I’ve been using a Copilot+ PC from Asus for the last few months, and one of my favorite things to do with it is try out it’s many integrated AI features while I have Task Manager opened to the NPU page, so I can see how much Microsoft’s AI features hit the fancy NPU built into my laptop’s processor.

I’ve never seen it move. Not once.

Up until Tuesday, every one of Microsoft’s AI features that I’ve tried have always hit their cloud services, not the NPU that was a key requirement for laptops to be labeled “Copilot+ PCs”. That all changed when Windows Update installed KB5062553 on Tuesday.

I hit the Start button on my keyboard, typed “fire” and pressed Enter to launch Firefox, the same way I do every time I reboot a computer. But Firefox didn’t open. Instead, it stalled and eventually opened an Edge tab to a Bing search for “fire”.

“That’s weird,” I thought, because one of the first things I do on every time I setup a new Windows machine is disable Bing search in the Start Menu. That’s when I noticed the new icon.

Image courtesy of Microsoft, because I permanently disabled the new search by accident.

Those twinkling stars and the slight animation when you click into it could only mean one thing. AI.

I opened Settings to see if there was an easy way to turn it off. I popped up to the search box and typed “Start” only to notice the same exact icon. A little swoop and some twinkling purple stars. It’s here too.

I figured this was part of a Windows Update, so I popped open the start menu again and typed “update”. The number 1 result was Discord (the executable for Discord is named DiscordUpdate.exe, so it makes sense, but it’s not the behavior I expected from the previous version) not Windows Update, which would have been the #1 in the old regime. I tried launching a few more apps. The AI search was slower and consistently less reliable. These are places that adding “context-aware” search is an active detriment to me. I know what I’m looking for, I know what it’s called, and I just want to have the least friction possible between the computer and me.

On a hunch, I popped open the Task Manager to the NPU pane and sure enough these searches were the first thing I’d really seen move the NPU’s usage needle. (The other usage I’d seen was using Windows Studio Effects, which blurs your webcam background and tracks your face, among other things.)

Microsoft calls this feature Semantic Search, and considers it a core part of Windows 11, so there’s no easy way to revert to the old functionality. It shows up in a few different places in Windows—the Start Menu, the Search box on the taskbar, Settings search, and Explorer’s file search.

After a few hours of testing, I’m frustrated that Microsoft isn’t letting us enable this new feature selectively. It’s incredibly useful for inside the contents of files, I was able to find photos based on their contents and the search inside documents worked better than it ever has before.

The Start menu search didn’t fare as well. After using it for a few more hours, the AI-assisted search is a huge net negative. It’s noticeably slower, it gives me Bing results when I search for applications that are on the PC, and it’s unnecessary. It’s an interface where my searches have zero ambiguity.

Settings search is also a regression. It’s another place where I actively prefer deterministic search. For me, it’s a shortcut to panels deep in the Settings app, rather than navigating an increasingly arbitrary tree. I neither expect nor want the search in Settings to try and help me solve problems with my PC.

I typically disable the Search box on the Taskbar, so I didn’t test it much. A quick look showed that it’s extremely similar to Explorer’s search.

The bummer about all of this is that Microsoft is taking an all-or-nothing approach to it. Because Microsoft views Semantic Search as a core part of Windows, they don’t provide an easy way to pick and choose where it’s enabled. While you can uninstall the core Copilot app and disable Copilot in Notepad (or use one of a bazillion great alternatives, like Notepad++), not providing users with an easy, reversible way to disable AI “enhancements” to a core OS feature seems like malpractice. They know it’s not good, there’s a disclaimer everywhere that it shows up and the OS asks you for feedback constantly when you use it.

When it comes down to it, I use the Start menu search dozens of times a day and search using Explorer at most once or twice a day. Having a Start menu search that opens an Edge window with Bing results for Slack when I just want to open Slack, an application on my computer, is not acceptable.

Disabling Semantic Search

The good news is that, for now at least, Semantic search is limited to computers that include a Copilot+ class NPU. That means it’s limited to a handful of laptops that came out in the last year. Your desktop PC won’t see this feature, even if it includes a dedicated GPU that can provide 10x the number of TOPS that the Copilot+ spec requires.

And if you do have one of those laptops, there’s an easy way to disable Semantic Search. Unfortunately, it seems to be irreversible. (I didn’t expect it to be a permanent change, so I tested ways to disable it before I grabbed all the screenshots I wanted, which is why I don’t have the usual complement of screenies here. Sorry about that.) Because the change was permanent, I didn’t get to test any of the other potential techniques I found using Group Policy Editor and Regedit.

So here’s how I permanently removed Semantic search from my Windows 11 Copilot+ laptop. Please don’t complain to me when you do this and you can’t turn it back on. This is courtesy of Reddit user PhantomOcean3, who pointed another user to vivetool. Vivetool give you an easy way to manipulate the A/B testing mechanisms that Microsoft has been using in Windows for the last few years. (Full disclosure: Semantic search will likely come back with a future update, but I haven’t been able to turn it back on in a current build and I’ve tried everything I know to do.)

Once you’ve downloaded vivetool and chucked it in a folder somewhere, all you need to do to permanently disable the AI-assisted search is to open an Administrator Powershell Terminal, navigate to the folder you unzipped vivetool into, and type ‘./vivetool /disable /id:47942714’. Reboot, and your local AI search will be replaced with the faster and more reliable deterministic search that’s been around since Windows 7.

What’s Next?

So anyway, I just went ahead and installed CachyOS on my laptop. I haven’t tried to run a Linux distro on a laptop in a really long time, but the process was painless and after a little bit of work I even managed to get fancy features, like my Windows Hello-style face authentication working in Linux. I’m a few days in, so there will be more soon, but for now, the way I use my laptop is much more conducive to a Linux lifestyle than my desktop, where I actively use Adobe applications and regularly play games that use anti-cheat that’s incompatible.

I’m still working out what the right cadence should be for these posts, so please bear with me while I figure out what’s sustainable. I’d rather miss a week than churn out something low effort nonsense just to keep to an arbitrary schedule. In the meantime, check out the video I just did over at PC World where I tested the performance impact that running a crapped up Windows install will cause. This video will shake the very foundations of your understanding of Windows.

Things I’ve enjoyed this week: the FunFactor Podcast digs into classic gaming magazines in a way that I love. Each episode, Aidan and Ty dive into a different magazine issue, but my favorite so far was Nintendo Power #88, aka the issue where they unveiled Super Mario 64. I wasn’t a subscriber by that point, but the game was such a big deal I went out and bought a copy at the newsstand.

There comes a time in every young man’s life when you sit down at the PC just after you should be in bed and decide to reinstall Windows to solve your PC problems. Mine came after a week locked in the bedroom with COVID, on the evening when I made my return to my desktop PC after a laptop-bound week in bed, gaming on GeForce Now and the Switch 2. My shit was busted.

A failed Windows update left my desktop suffering from random bluescreens, stutters in games that shouldn’t be stuttery, and just general across-the-board jank with Windows. It was not the triumphant homecoming to my desktop PC that I had anticipated.

After digging deep into the minidump files Windows generates when the PC bluescreens, I found that my bluescreens weren’t happening in any one particular file or driver. After consulting with an expert, I suspected that the failed Windows update was probably the culprit. To confirm, I installed a clean version of Windows 11 on a small portable SSD and ran it for an hour or two to see if I got any bluescreens. At the end of that session, when I hadn’t experienced any wonkiness, I had the confirmation I was looking for. It was time to reinstall Windows, but it was also midnight.

With a live recording of The Full Nerd looming the next morning at 11, and a busy week full of meetings and recordings after that, I realized that if I wanted to get the reinstall done without affecting the rest of my week, this was the time, even if it was midnight on a Monday.

Just under two hours later, my PC was back up and running, with all of my applications installed and everything I needed to work the next day in place. My games were reinstalled and all of my applications were configured just the way I left them on the old Windows install.

So how’d I do that? I’ll explain it by hitting three topics today—first I’ve got a quick primer on how to look at the dump files Windows generates when you bluescreen, then we’ll talk about how I set up my system so that Windows reinstalls are fast and relatively painless, and then I’ll close with some basic backup strategies for Windows.

A Three Paragraph Primer on the Importance of Inspecting Your Dumps

By default, Windows 11 saves a small file that contains key debugging info from your memory when your computer blue screens, called a minidump. To inspect your minidump, you’ll need to grab WinDbg from the Microsoft Store, and then double-click the dmp files you find in c:/Windows/Minidump. The dumps include a time/date stamp in the file name, so you can use that to know which dumps are relevant to your current problems.

KERNEL_SECURITY_CHECK_FAILURE is never what you want to see in a BSOD

When WinDbg opens, you’ll see an analyze -v link in the Command pane. If you click that, the app will scan the dump and you can pull out some key info. The two most important bits of information for our purposes are the type of bluescreen and the software which caused the bluescreen. The type of bluescreen is in the first line under the big “Bugcheck Analysis” block and the offending software can be found by searching for “Process_Name”. You can hit ctrl+f to search through the wall of text in the command pane to dig out these nuggets.

Once you get that info, you have what you need to effectively diagnose your problem. If you have multiple bluescreens and you’re seeing the same kind of error in the same software consistently, it’s probably a problem with that software. You can solve it by reinstalling that software or driver or whatever the offending application is. Unfortunately for me, I was seeing different bluescreens in different applications, which was indicative of a larger problem with Windows.

The Reinstall and My Two-Disk Strategy

So now I’m faced with a reinstall. I suspected that this was coming, so I made a fresh image of my c: drive earlier in the day (more on that in a moment), and was prepared. I rebooted my PC with my Windows 11 24H2-equipped Ventoy thumbdrive in the USB port and kicked off the install process.

As an aside, if you regularly install multiple operating systems, Ventoy is the bomb. You make a specially formatted USB thumbdrive and it automatically scans a partition on that drive for bootable ISO files, which it lets you choose between them in an easy-to-use menu when you boot off of the drive. It’s turned my sack of OS install USB thumbdrives into a single drive that I can use to install different versions of Windows and different flavors of Linux. Highly receommended.

After deleting my Windows partitions (and leaving my Linux partition intact), I let Windows run its reinstall process. I use a mostly stock Windows install, the only thing I do is bypass the online account requirement during install, because I don’t want my Windows profile folders stored in OneDrive. You can do that by pressing Shift + F10 when the installer prompts you to connect to the Internet, then typing ‘start ms-cxh:localonly’ at that prompt, and following the instructions to create a local account. (For this to work, it’s vital that your old account and new account be the same because Windows creates your user profile folders based on this path. Once Windows boots up, you can safely connect your local user account to a Microsoft account without invoking OneDrive, if you want.)

While Windows was installing, I fired up my laptop and downloaded drivers for my hardware, installers for the applications I use most frequently, and the other assorted crap I install on my PC. I could have used WinGet or Chocolatey or another Windows package manger to gin up a quick script that automates installing this stuff, but clicking the installers is relatively easy and fast enough for infrequent, one-off installs. If I was unfortunate enough to reinstall Windows all the time, it might be worth automating the process more.

Once the Windows install is done, my two-disk strategy comes into play. I keep all my work on a separate SSD from c: drive, so my post install process was simple. First, I installed drivers for my hardware (chipset, graphics card, and other assorted motherboard stuff like network, Wi-Fi, Bluetooth, Thunderbolt, etc), run Windows Update, and reinstall my applications, making sure to reboot as instructed throughout the process. I also went into my folder view options in Explorer and turned on “Show hidden files, folders, and drives.”

Then, I rebooted the PC and immediately closed all the applications I’d just installed, including the stuff hidden in my system tray. I mounted my backup image of my old c: drive using Disk Management > Action > Attach VHD and assigned it a drive letter. Once I mounted that image, I was able to copy my entire profile folder over from the old image into my new profile directory, replacing files with reckless abandon. That’s c:/Users/<yourusername> It’s important to note that this only really works if the path to your user folder is the same on both your old install and new install and if you really don’t have anything running when you do this.

The copy just took a few minutes, thanks to the speedy external SSD I saved the backup on. After it finished, I rebooted the PC again and was back in business. While I’ll undoubtedly find things I need to do over the next few weeks—like removing search results from the Start Menu—the core applications and utilities I use on the day-to-day were all back up and running, in just under two hours.

This works because most applications store their configuration info in one of the subfolders of the c:/Users/<username>/AppData folder. When you copy your old profile folder over, your apps will find your old settings where they expect to be, and since I save all of my work files on my d: drive, the old work is where apps expect it to be when I fire them up.

Backing Up and You

The backup situation on modern Windows is, frankly, grim. With robust backup solutions that will use existing hardware built into OSX and readily available on Linux, the fact that Microsoft’s backup solution is exclusive “Use our cloud-based OneDrive solution and also give us money please” leaves a lot of users in the lurch. While OneDrive is great for backing up lightly-used PCs that are mostly glorified web browsers, I don’t think it’s sufficient for people who use their PCs for more than that. Anything that uses a large amount of space—gaming, photo and video creation, audio editing, or programming—will fill your available OneDrive space too quickly. So what’s the right local backup solution for Windows users?

This is where the classic Windows 7 Backup and Restore tool comes into play. It regularly backs up my important data to a NAS in my garage, but you could just as easily use these same tools with an external hard drive or SSD. (I also use Windows File History on some machines, as it’s a fast/easy backup that’s reasonably well supported in Windows 11. We discussed it in detail in a Tech Pod episode last month, but I can go into it in a future newsletter if folks want. Reply to this email or post a comment!)

Windows 11 does include a decent local imaging tool, it’s just buried in the pre-Windows 8 Control Panel where no one will find it.

Microsoft hides the Backup and Restore tool deep in the legacy Control Panel. To find it, type “Control Panel” in the Start Menu and go to System and Security > Backup and Restore (Windows 7). Select “Set up backup”, and follow the prompts until it asks you ‘What do you want to back up?’ You should choose ‘Let me choose’ and then make sure that under Computer the drives you want to backup are selected as is the check box labeled “Include a system image of drives…” at the bottom of the pane. I typically uncheck the two options under Data Files, since the drive backups will cover them as well. Hit next, and let the backup run. Depending on the size of your drives and the speed of the drive you’re backing up to, it might take a while.

To restore using this backup, you need to reinstall Windows 11, then open the Backup and Restore tool and walk through the Restore process. I’ve had mixed results trying to do a complete system restore using this process, but it’s fabulous as a way to cherry pick just the files you need and/or keep an image of an old install around just in case you discover you missed something important six months down the line. You can always mount that backup image, dig into your old file system, and restore whatever you missed.

What’s Next?

So I’m generally back in business, and happily running a crisp, new Windows install. Fixing this in a relatively short period of time feels really good too. Understanding how your computers work and how to fix them is an important way to fight enshittification. If the solution to every problem is just “yo, just buy a new one” that’s not great for anyone but the people selling the new one. It just strikes me as funny that I’m here again, writing about how to reinstall Windows, because that was one of the first cover features I wrote at the start of my tenure at Maximum PC.

If you’re looking for more great newsletters, I’ve really been enjoying my friend Patrick Klepek’s Crossplay newsletter. Partly because the topic is immediately relevant—navigating the intersection of parenting and games is challenging in a world where there’s enormous peer pressure pushing your kids toward predatory services like Roblox—but mostly I really missed reading Patrick’s writing about games.

Between being sick and my recent Windows misadventures have stalled out my time on Linux a bit. The Windows reinstall nuked my EFI partition and took out my boot manager. I’m planning to reinstall rEFInd on the EFI partition, so I’m sure I’ll talk about that next time.

I managed to pick up a cold this week, so this week’s newsletter will be a bit thinner than usual this week. Any typos or errors are the result of head congestion.

Desktop Linux Notes

I did some distro research, and I decided to try CachyOS first. It’s based on Arch, the same as the SteamOS, but they recompile software to require the latest CPUs. That awoke something inside my performance obsessed heart.

The install and setup process was really straightforward, with NVIDIA drivers and some of my most commonly used apps working seamlessly on my first boot. I followed the Cachy install guide, which worked for me after a few false starts. I had originally resized the partition on my SSD using the Disk Manager in Windows, but the Cachy installer didn’t like that. Once I restored its original configuration and let the Linux installer handle resizing it worked fine.

A desktop screenshot of a fresh CachyOS install.

Before I committed drive space to installing Cachy, I tried it out for the better part of a day using the live install USB image. Most Linux distros provide something similar now, which lets you take the OS for a test drive before you commit SSD space to it. Instead of building a dedicated bootable installer, the Cachy installer is a bootable version of the OS with an installer app that you can run when you’re ready to commit. The installer will help you format or repartition your drive, walks you through your configuration options, and copies everything over to the appropriate partitions.

A note on dual-booting: In order to boot multiple OSes on the same device, you need software that sits between your OS and your UEFI that will let you I decided to use rEFInd, which is a boot manager that sits between your UEFI (what you probably think of as your BIOS) and your operating systems. I put it on its own partition and set my motherboard’s UEFI to boot to the rEFInd partition. When I rebooted my PC, rEFInd detected my Linux and Windows installs and presented me with a slick graphical UI to choose the OS I wanted to boot into. Each subsequent boot goes into the OS that I used last, although I’m sure that’s a configurable option somewhere in the config files.

Once I had the OS installed, I installed flatpak, using the instructions at Flathub. Different distros package applications in different ways, and your favorite services that distribute Linux versions of their applications—like Dropbox, Notion, and Slack—may not provide their software in your distros preferred format or repositories. This is where flatpak and Flathub shine. They provide a cross distro way to distribute software that handles installation and updating across multiple distros.

Once flatpak was installed, I was able to download and install almost all of the software I use on the regular, including Audacity, OBS, Notion, Slack, Discord, Dropbox, and a bunch more stuff that I’m forgetting.. Up next is getting my video and photo editing software working, and setting up a backup process to save everything to my NAS.

More updates soon.

Switch 2 Thoughts

It’s lovely to pick up a piece of hardware that’s barely been touched by the enshittification hammer. Products cost money not attention, I’ve only spotted one dark pattern (it’s difficult to figure out how to download the free-to-subscribers updates to the Zelda games), and they seemed to have a whole lot of hardware available for sale. So far, it’s the smoothest console launch in years.

We discussed first impressions at length on this week’s Techpod (posting tomorrow), but the TLDR is that they made another Switch and it’s good. I’ve been playing a few older games at a smooth 60fps—Breath of the Wild, Tears of the Kingdom, and Super Mario 3D World mostly—as well as Mario Kart World, but my favorite part of the launch lineup is the Switch 2 Welcome Tour.

It’s hits me exactly in my nerdiest parts, with surprisingly detailed motion graphics and demos showing how the new (and old) stuff in the console actually works. It’s unfortunate that it costs $10, because it’s packed with the kind of tech demos and cool facts that makes me feel good about spending money on the hardware and gives players a ton of material to evangelize the console.

Hopefully I’ll be back for a regular newsletter next week, but in the meantime please keep sending in your favorite Stream Deck tips and tricks. I’m collecting them for a future newsletter, and you all have sent me some really clever stuff so far. It’s changing the way I think about my Stream Decks, and I already think about them quite a bit.

It would be easy to discount Elgato’s Stream Deck because it was originally built for Twitch streamers. But, you’d be doing yourself a disservice. I find that having a small set of software-configurable buttons attached to my PC is incredibly powerful. And the secret, as is usually the case, is good software.

What launched in 2017 as a tool for streamers to create macros that would let them create single button presses that triggered common tasks on their streams—changing scenes, adjusting lighting, muting audio inputs, and running chat commands—has become much much more. My Stream Decks have become powerhouses of productivity because Elgato included multiple ways for the Deck to interact with other applications. Full disclosure: This is not a sponsored post, I bought one when they first launched, because it seemed like a fun novelty, and over the years have upgraded and added a few more. I currently have two Decks in-use on my desk today, a Stream Deck + and a Stream Deck XL.

I love my Stream Deck +, which has knobs I use to adjust audio on my streams and recordings.

The Deck has three basic superpowers. It can interact with pretty much any application on your computer, it can automatically change modes based on what your using your computer for at any given moment, and it lets you stack multiple actions onto a single button press or toggle. Combined, these three powers make these devices incredibly flexible and capable of automating complicated workflows. Let me explain.

The Deck’s first superpower is that any button can trigger pretty much any keyboard shortcut, which lets you interact with virtually every piece of software running on your desktop. You can also set buttons to open or close specific programs, run a command line, perform common system tasks (like muting audio), or adjust settings on the Stream Deck itself (brightness, switching between pages of buttons, etc). The Stream Deck also includes a full plugin infrastructure, which people have gone wild with, adding support for thousands of their favorite applications to the devices. And finally, if that isn’t enough for you, the SDK lets you write your own code that interacts directly with the device.

The Deck software’s second superpower is that it’s context-aware. It knows what software you’re running at any given time and can adjust the buttons that it displays accordingly. As an example, when I open Premiere, it becomes a bank of hotkeys for timeline manipulation, exporting videos, capturing screenshots, etc. When I’m in Audacity or Audition, I get hotkeys for audio editing and shortcuts to the folders where audio lives on my computer. And when I open Zoom, it swaps to buttons that let me control my OBS Virtual Camera and turn on my lights for a conference call. The rest of the time, my Stream Deck is a hardware interface for my audio mixer, a home automation control pad, and a tool that lets me control OBS. Your use will almost certainly vary.

But you can just use keyboard shortcuts to do all of this, right? That’s true for the most basic functionality, but the Stream Deck’s third superpower is that it lets you bundle up multiple actions into a single hotkey, including supplying pauses between steps. This lets me turn checklists and other multi-step tasks into single button presses.

For example, before I bought a Stream Deck, I had a checklist of things to do before I started a stream on a post-it stuck on my monitor. It included software to launch, settings to toggle and confirm, lights to adjust, and more. I went down that checklist every single time I started a stream. Now I hit a single button to launch all the software I need, arrange the newly-opened windows in their proper positions, configure my lights and audio for broadcast, and more. Then I hit a second button to launch the stream, which starts a pre-roll reel for whatever game I’m playing, mutes my audio inputs, joins the appropriate Discord channels, and pushes the stream live.

The thing that happened, after I started using a Deck for a while, is that every repetitive task started to feel like a Deck problem. When I was testing microstutter for PC World a few weeks ago, it was a moment’s work to setup a hotkey that triggered my profiling tool and started a video capture within a few milliseconds of each other. That made capturing a 60 second loop of gameplay and performance metrics a single button press that I could easily trigger while I was playing games. I have hotkeys that automate my common tasks in Premiere and Affinity Photo, clip web links to my Notion story ideas page, toggle my online presence in all my chat apps between Online and Invisible, and a bunch more.

(The only downside of this is that I have to resist the urge to over-automate tasks, lest I become a living example of this XKCD.)

Yeah, this has never happened to me.

The Hardware

The biggest knock against the Stream Deck family of devices is that they’re overpriced for what they are, which I think was more fair in 2017 than it is today. With 1000s of software integrations for the Deck hardware, I think the pricing the the Deck hardware is justified (but they also run frequent sales on their store and Amazon).

What started as a 3x5 row of buttons has become kind of a complicated product family, especially as you dig into the new stuff that they rolled out at Computex last week. Here’s the quick rundown:

• Boxes with buttons and knobs - This is the core hardware but it comes in a bunch of different shapes and sizes now

  • Stream Deck - This is the overarching category for physical Stream Deck hardware with buttons

    • Stream Decks are basic grids of buttons. They come in three different sizes: Mini with 6 buttons, XL with 32 keys and standard with 15. They’re rolling out versions with more tactile scissor keys now as well.

    • Stream Deck Neo - This is an 8 button keypad with a small screen for date/time/weather and a pair of dedicated touch buttons for switching between different pages of buttons.

    • Stream Deck + - The Stream Deck + adds 4 analog + click knobs and a touchscreen slide bar to an 8 button Stream Deck. It also supports a handful of clamp-on accessories, including an XLR-audio interface and a USB hub.

• Software options - There are also a handful of software implementations

  • The Steam Deck Mobile app for phones and tablets gives you a way to test drive the whole ecosystem for free. The phone apps give everyone access to 6 buttons. Beyond that, they charge $25/year or a one-time charge of $50 for unlimited access to the app, which feels kind of expensive for an app.

  • You can also add software buttons to your Windows or Mac desktops using the Stream Deck version 7 software (in beta right now). In order to unlock the Virtual buttons on your desktop, you have to own a hardware Stream Deck or certain Corsair peripherals.

• The weird hardware - This is where the Deck expansion is happening.

  • Studio - A rack-mounted Stream Deck that works with Bitfocus Companion. This one is really designed for professional audio and video applications, so it’s safe to ignore.

  • Pedal - a really sturdy, 3-button pedal that’s used for Mute buttons, triggering video capture, or other hands-free activities. If you need a pedal, you probably already know it.

  • Stream Deck Module - If you want to build your own custom hardware that includes Stream Deck buttons, this is for you.

The cheapest and fastest way to get started is to download the Stream Deck Mobile app for your phone. It walks you through the initial set up and process of connecting your phone to your PC or Mac, and the free 6 buttons, combined with the auto-switching profiles and paging features may be enough for you.

I think that both the Stream Deck Neo and Mini offer good jumping off points if you’re interested in trying the hardware. The paging buttons built into the NEO make it really flexible for people just getting started, but I don’t think you can really go wrong with any of this hardware—unless you just buy it and leave it sitting in a box on your shelf for two years.

If you decide to purchase some hardware, please consider using the links in the post. They’re affiliate links, which generates a few bucks per purchase and helps support the newsletter (and me and my family, which I appreciate).

If you do end up integrating the Stream Deck with your workflows, please smash the reply button and share your favorite tips and tricks! I’ll share the best ones with folks in a future edition of the newsletter.

First off, I need to apologize for the unplanned hiatus. I had a few prospective topics in a row fail to work out as newsletters, which was compounded by an incredibly busy few weeks in my day job, and a full-blown obsession with getting Linux up and running as a daily driver on my main PC. The good news is that I’m using Linux just about half the time now, and there are a lot of potential topics there. There’s so much that it’s a bit overwhelming (more on that at the end of the letter).

At the same time, and partly because it’s much more useful as a day-to-day Linux user than a Windows user, I’ve been teaching myself Python. And I’m doing it with a game. The Farmer Was Replaced is a game that helps you learn how to use Python (technically, it uses a language similar to Python, but the syntax is extremely similar) to solve a series of increasingly complicated tasks. After spending a bunch of time with it over the last few weeks, it’s also the first programming game I’ve played that actually helped me learn how to go from an blank text window to a functional program.

I’ve been trying to learn how to write useful programs since I was in college. I took CS100, where I learned to write C code on Sun workstations back in the 90s. I learned some fundamentals about writing software there that have stuck with me for 30 years—how to use loops, recursions, and basic data structures to make useless software that only ran in CLI mode on the CS department’s Sparcstations. I bailed out of the CS department after I realized that I was less interested in learning about the theoretical underpinnings of software than I was in writing functional programs for computers people actually used.

I once learned how to make an audio CD library program using CLI mode on a SparcStation just like this one. It wasn’t particularly useful software.

Later on, I wanted to build webpages, so I learned markup languages, CSS, and a little bit of Javascript in the late 90s and 2000s. When the first Arduinos came out, I found that my CS100-level knowledge of C was useful again and I’ve used that to make a handful of simple, but useful Arduino devices over the years. The ability to start with a blank page and write a program that did something I needed done was always a wall for me, but I could look at other people’s scripts and figure out how they worked and then adjust them to do what I wanted.

When companies like Zachtronics started releasing games that required programming skills to play, I tried a bunch of them. But I inevitably hit that same wall, where the problems the game presented exceeded my ability to come up with an algorithm and then turn that algorithm into code that can solve the problem . I think part of that is that Zachtronics games seemed to be tuned for people who were already programmers—people who understood how to assemble the basic pieces of a program from scratch and how to create algorithms that could solve complicated mathematical problems.

The Farmer Was Replaced seems to have been made for people who want to learn to program. It’s helped me push through several of the walls I’ve hit in previous attempts, and it’s done it by building a very satisfying loop. It presents me with a problem, I figure out how to solve it, I write the code to execute that solution, and then I get to watch the code execute (and/or spend some time debugging, if I’m being honest).

It starts with a simple metaphor, some land that you—a drone farmer—need to tend. You learn how to monitor growth, cut the grass, and create loops that run based on specific conditions. Then your single tile is turned into a single row, and eventually more rows are added until you have an arbitrary-sized grid to manage and are harvesting tens of thousands of crops per minute.

The Farmer Was Replaced starts simple, with one tile to maintain. You expand your farm as you expand your abilities.

The game has a skill tree, that lets you unlock ability to grow different crops. Different crops have different requirements for optimal growth, some require tilled soil, some require watering or fertilizer, and some require that you plant them between other crops. Each of the new crops provides an additional challenge, and each one teaches you new programming and algorithmic concepts—everything from recursion and object-oriented programming to sorting and maze-solving algorithms—as you progress up the game’s tech tree.

And the game does it with a gentle difficulty curve. The problems are presented in a straightforward and incremental way and the game includes a well-developed hint system that holds your hand when you need that but also knows when to get out of its own way. (If you’re new to programming, there’s no shame in using the hint system, it includes absolutely crucial information that you’ve probably never been exposed to.) When you reach more complicated computer science topics, like sorting algorithms or maze-solving techniques, the hint system gives you the keywords that you need in order to find useful results on the web, rather than reinvent the bubble-sort wheel.

Eventually, you unlock abilities that let you perform every action that you’ve had to manually trigger with code, which let’s you write code to perform all the tasks in the game, including meta-progression stuff, like unlocking skills. The endgame for The Farmer Was Replaced is to replace the player—that is to write a program that plays the entire game. Once you’ve done that, you can optimize your speed until the game is playing itself as quickly as possible. And yes, there are leaderboards. I’m not quite ready for leaderboards yet, but I can see the path to get there!

A screenshot of my pumpkin/wood farm midway through my playthrough. It pretty consistently farms most resources now, but I haven’t gotten logic for handling upgrades in yet.

Why am I talking about this today? I think that it’s important to know how to write programs for our computers. Even if it’s just a simple script to manipulate file names on your computer or organize your photos by date taken, it’s worth taking the time to learn. The Farmer Was Replaced is the first tool I’ve used that actually helped me make that leap from a blank page of code to a working program.

What’s Next

Part of the output from being “busy at my day job” is available online to watch now. There are a handful of videos on the way, but the first of them, a look at why id software’s modern Doom games feel so fast despite running at relatively modest framerates, is live now over at PC World’s YouTube channel. Coming soon are deep dives into performance degradation on Windows over longer periods of time and the impact of system cruft on different system configurations.

I’m still a little swamped at the day job, but unless something drastic happens, the next edition of the newsletter will cover some good backup strategies that don’t involve paying Microsoft to do a half-assed job of it with OneDrive. I’ve done a bunch of research on this topic and tried out a handful of the most recommended tools and I have thoughts. I’ll be including guides for people who have some sort of network-attached storage, use simple external drives, or are just looking for a simple cloud solution.

Folks who also listen to either the Tech Pod or The Full Nerd know I’ve been half joking that Microsoft abandoning Windows 10 this fall makes 2025 a strong Year of Desktop Linux candidate. I could really use your help figuring out exactly how I should cover desktop Linux here.

I’ve got some basic guidance in the works that covers how I evaluated potential distros, what I looked for, how I chose things like desktop environments and boot management tools, etc. I’ve also got some “trip report” type coverage, which gets into what’s working well for me and what I’m having trouble with, as well as what I’m learning along the way. I’m also planning on covering some more specific topics—like getting games and other Windows software running, how you can keep your PC safe when you’re using software that doesn’t include guardrails, and collections of tips and tricks. I’d love to know what other questions you have and what topics you’d like to have included, so please mash that reply button and let me know!

A few weeks ago, a reader wrote in and asked me if there was an easy way to tell the difference between good home automation gear and bad home automation gear. It’s a good question. I love my Home Assistant setup. Everyone in the family uses it, it’s made life more convenient for us all, and we use it to help reinforce positive habits. That said, the Internet of Shit is undoubtedly real and worth talking about. I just wasn’t sure I would be able to break down the different ways that that IoT gear can go bad in a reasonable number of guidelines.

But I figured it was worth trying and once I got into it, I realized that there are some commonalities to the most common failures for connected devices¹ .

And it’s worth saying that this is likely the first version of this list. As people suggest new categories and new and novel fail states, I’ll add them to the web version of the article. Basically, if you let this one age in your inbox for a few days before opening it, you might want to check the web version of this post too.

Without further ado, here are five simple rules to help you distinguish good connected devices from the Internet of Shit

1. Devices Should Not Require Online Services to Work

This is the big one, and you all probably already know it’s a problem. If your home automation gear and other connected devices require online services to work, they’re living on borrowed time. I talked about this with Home-Assistant founder, Paulus Schoutsen a few years ago, and this was his main complaint with the category. His recommendation was simple, avoid any products that require an online service to work. This applies to everything, from doorbells and speakers to smart bulbs and switches. As soon as a company turns those services off, the products are almost always done for. (As a bonus for Home Assistant users, HA displays a globe on Services that require Internet access to function in the HA user interface and on their website.)

Which of these services requires an Internet connection? Hint: It’s Wyze.

While there are a handful of beloved products that dedicated nerds have managed to revive after their manufacturers killed them, but they are rare compared to the piles of remotely disabled gear that people have been forced to truck to e-waste. At the same time, my 13-year-old Hue hub is still going strong, because my Home Assistant install knows how to control it, years after Philips turned off their servers.

It’s worth mentioning that this rule explicitly means that high profile devices from Google and Amazon, like pretty much everything sold under the Nest or Ring labels, fits this category and should be avoided. (Full disclosure: I’m a bit of a hypocrite here, because I actually have a Nest doorbell. Rather than replace a working device with something new, I’ll keep using this one until it stops working. I’m confident that Google will support it for the foreseeable future and I’m ok replacing it with something more open if and when they end up killing that support.)

As an added bonus, using IoT devices that work locally are almost always faster to respond than devices that require some sort of cloud connection. For example, Siri is setup to connect directly to my Home Assistant without a cloud component, and its response time is basically instantaneous. When I try doing the same action through a Google Home device, which requires cloud servers to work, it takes a few seconds to trigger. Doing it right is actually better!

2. Devices Should Not Require Apps for Configuration

This is another important criteria for hardware that should last for a decade or more. Smart light bulbs are a great example here. While light bulbs have used the same physical interface for a hundred years, the app infrastructure to connect and setup your smart bulbs is ever changing. So, if your bulb vendor goes away or your phone provider changes their app ecosystem sufficiently, you can lose access to the apps that you use to configure your smart bulbs. That will leave you with a bunch of smart bulbs that work just like much cheaper not-so-smart bulbs, since you have any way to connect them to your network.

Devices that are configurable using standard protocols, like Zigbee or Matter, are fine because those standards should allow the devices to operate independently of the manufacturer’s app. Typically you configure them using your smart home hub, like Home Assistant. Likewise devices that you can log into a web interface to set up are OK. Browsers may not be trendy, but they are eternal.

3. It’s Not OK for Devices to Collect and Report Unnecessary Data

For some types of smart devices, there’s a certain amount of connecting to online servers and transmitting data that’s necessary for the device to function. Yes, I know I just said not to do this two items up, but if you have a doorbell camera that does face recognition or a smart speaker that accepts voice commands but doesn’t have the horsepower to handle speech-to-text onboard those will need cloud assists to work.

But none of your smart home devices should scan your network to catalog other devices, track your physical address, clock everything you watch on your TV, or track viewer’s presence in or around the TV set. They certainly don’t need to hook into social media, advertising, or data broker services to build a more fully-fledged profile of you as a consumer (derogatory). Think this sounds bad? It is, and it’s exactly what Roku streaming boxes and smart TVs do. Smart speakers are just as bad.

What a friendly-looking TV that’s also sending a ton of data about you back to its mothership. Source: Roku

And that’s without getting into the data tied to ultrasonic and short-range radar presence detection or devices that use inaudible ultrasonic waves to communicate with each other, even across air gaps. There are even some advanced/hypothetical hacks that can jump between devices using ultrasonic comms.

There isn’t an easy way to scan your home network for offenders, unfortunately. The tools you use to find out when a badly behaved device are snooping on your network are designed for people with a deep understanding of how networks work. That means the easiest way to check in on your stuff is to look them up on a privacy watchdog’s site. I like the Mozilla Foundation’s privacy reports for devices, they tend to do a good job breaking down the actual problems with specific products without sensationalizing non-issues and with a no-nonsense analysis of what the vendor’s privacy policy allows them to do with your data. I also really like that they pair the data the vendors are allowed to collect with the worst that can happen if you or that vendor get hacked. My only complaint with the Mozilla offering is that they update infrequently, and tend to hit only the most popular products.

Apple’s Homepod products generally get decent marks for privacy because they don’t call back to the Internet for every single user interaction. Source: Apple

There are a handful of vendors that aren’t harvesting every spec of data that they’re able to at every opportunity. Mozilla’s generally recommends Apple hardware as safe, with positive looks at the Apple TV and HomePod Mini.

4. Is It Easily Repairable?

Part of making things that serve users means making them repairable. That means replacing the hardware that is likely to fail over a reasonable lifespan, which usually means that you need to be able to replace the battery for any battery-powered devices.

I love the iFixit Repairability Index for computers, phones, tablets, and other devices. While they don’t specifically list a top-level category for home automation stuff, they do include repair guides for a large number of smart speakers, screens, and other IoT stuff. You just have to search for them.

iFixit is a great resource for finding products that you’ll be able to do both simple and advanced repairs on.

5. Does it create a potential privacy nightmare for you?

There’s a whole category of devices that can be fine if used properly and a nightmare if used carelessly.

The classic example of this is the smart lock. Sure, anyone who locked themselves out of their house at 2AM and had to wait an hour for a locksmith to show up, open the door, and charge you a few hundred bucks can see the obvious benefit of a lock that you can open without a key. But if you pair a smart lock with a voice assistant that’s placed too close to the front door or window, you can end up in a situation where a random stranger can unlock your house by yelling “Alexa, unlock the front door” at a bedroom window.

Using Internet-connected cameras inside the house is also a big no-no for me. I have a few cameras setup around the perimeter of my house, to help deter porch pirates and catch video of the critters that love to dig up my backyard, but none of those cameras face private areas, windows, or places people wouldn’t have a reasonable expectation of privacy. (If a malicious actor got access to videos of the skunks and gophers in my backyard, it wouldn’t cause me any problems.)

There’s two reasons for this. The first is that video that isn’t available outside my LAN is much less subject to leaks. While I know Google is going to use whatever data I grant them access to to sell me stuff, they have a pretty good track record with leaks of users’ personal data, unlike Amazon, which has has a series of problems, including one where contractors passed customer’s video around on an open network share inside their office.

When you’re adding new tech to your home, it’s worth taking a few minutes and thinking about the potential threats that whatever you’re adding will expose you to. How can you ameliorate that threat? In my Smart Lock example, you could just disallow access to locks from your voice assistant (this is the default for most smart locks I’ve tested now). If you add presence sensors that track your movement inside your home, how could a malicious user who gains access to your HA install use that info? Too much paranoia is bad, but too little is worse.

What’s Next?

For next week, I’ve been thinking about the default settings quite a bit. I’ve had to use a few fresh Windows installs lately, and the difference between my cleaned up version of Windows and the new out of the box Windows is grim. I think about the data that shows that something like 90% of users never change default settings, and I wonder if there’s anything we can do to help those folks out. If you have ideas, please smash reply and let me know!

This week’s recommendation is Doc Burford’s kick ass game design blog. Doc’s one of the most thoughtful people writing about games today, and every time I read one of his posts, I come away with a brain full of fresh ideas.

Sorry for the late newsletter this week, I’ve been a bit under the weather for the stupidest possible reason. I won’t get into it here, but I discussed it under separate cover. All I can say, is in this season of extreme tree pollen, please be careful when you’re sneezing.

This is what Quake 2 actually looks like. More than 25 years after release, it will run on every computer you own.

I was planning on writing about the importance of default settings this week, and I’ll come back to it in the not-too-distant-future, but I got sidetracked by this Copilot AI-generated Quake 2 demo, and I can’t stop thinking about what an enormous disconnect this whole thing represents.

And once again, Microsoft is desperately looking for a use case that people actually engage with to justify their multi-billion dollar investment in AI. And once again, they’re not finding it.

What’s the demo?

First, you should probably go try the demo, then I’ll tell you what’s happening here. (No, I don’t have any info about the cost of running this thing, but it’s safe to assume that it’s the rough equivalent of chucking a few cute woodland animals into a woodchipper every couple of minutes .☹️)

What’s going on here? First off, the thing you just tried out isn’t really a game. It looks like a game and sounds like, well, it didn’t make sounds for me. At best, it’s an interactive video, almost devoid of game. Sure, it looks like a game, maybe even one that you’re familiar with. But it lacks the complicated interconnected systems that make play possible and without those systems, players have very little in the way of agency or direction.

For example, look at the world geometry. The space is aggressively non-Euclidean. What does that mean? Well, the structure of the world you’re experiencing has trouble maintaining coherency, even when you’re just moving forwards and backwards down a hallway. Walk forwards down a hallway, and there’s a window on your right. Back up and the window is gone. Walk forward, and it’s there again. Walk up to a door, open it, and see one area. Let the door close and reopen and a completely different area is on the other side. Worst, when you walk into an area that’s dark enough to black out the screen, like pretty much any corner in Quake 2, you’re not going to exit that darkness in the same spot, whether you rotate 180 degrees or just back away. Every texture or dark area is a potential portal to any visually similar area of the same level.

The non-Euclidean geometry isn’t a problem on its own. There are loads of great games that utilize non-Euclidean geometry intentionally. While those games take place in worlds that don’t follow the rules of 3D space that our brains are accustomed to, the rest of these games have well-defined structure and direction, which helps the player avoid getting lost and frustrated. (Check out Antichamber, The Stanley Parable, and Superliminal if you want to see games that use non-Euclidean spaces for novel gameplay.)

And that’s the crux of the problem. I generally find it impossible to define what is (or isn’t a game), but one of the few consistent features of every game I’ve ever played, from Go Fish to Caves of Qud to Super Mario Bros, is that there are systemic rules that players can understand and predict outcomes from. And, as Austin Walker said on Bluesky yesterday, exploring the edge cases in those interactions and outcomes is a large part of what makes games fun. Games combine simple rules to create emergent or unexpected behavior constantly. It’s fun!

As an example, in Quake, there are several straightforward rules that govern damage done by weapons.

• Taking damage imparts momentum on enemies, which is used to show a stagger effect when you shoot an enemy.

• The momentum imparted on an enemy scales with the damage applied. More damage equals more momentum.

• Damage and momentum are additive and can come from multiple sources simultaneously.

• The same damage rules that apply to enemies also apply to players because Quake 2 is also a multiplayer game.

• Some weapons, like the rocket launcher, do splash damage in an area of effect around their impact spot.

Players created an intersection between all these rules by simultaneously jumping and firing a high-powered weapon (like the rocket launcher) at their feet. You impart the momentum from the rocket explosion and the jump and you go much higher in the air than if you just jump alone, if you survive the explosion, at least. This is the origin of the rocket jump. (As a bonus, if you time it perfectly and stand on an exploding grenade while you rocket jump, you’ll go even higher. If you survive.)

Needless to say, there’s no rocket jumping in Microsoft’s AI version of Quake 2. Not only is there no rocket launcher, but there’s no gameplay framework to allow that emergent behavior to happen.

Their AI model is just generating images that it knows are related to each other very quickly and sequentially based on input from the player. It was trained on video of players playing the same Quake 2 level over and over again. There’s no underlying logic, it just knows that when the arrangement of pixels that you recognize as a health pickup disappears under the bottom edge of the screen, a certain UI element changes. When an arrangement of pixels that you recognize as an enemy’s bullet traverses the screen in a specific way, the screen flashes and the player’s health number on the UI goes down.

In the MS demo, there are occasionally situations that predictably mimic real-life game rules. For example, almost every time I walked down a long hallway with a left turn at the end, the model drew a monster at the at exactly the same spot, jumping out at me in exactly the same way, presumably because that happened every time players in the training dataset traversed a similar hallway. It happens even if I walk back and forth down the same hallway. While this is predictable behavior, it also breaks the rules that players expect from this kind of game around enemy spawning and behavior.

What does the training dataset look like for a generative model that can do this? Microsoft didn’t release specific training data on this model, but in an earlier paper, they stated that it took 7 person-years of gameplay to generate a model using an unreleased game from Ninja Theory. To give some reference, Quake 2 was built in about a year by 10 or so people.

After playing this a few times, I really didn’t understand why Microsoft would release such an obviously flawed demo. The game constantly breaks, there’s no player progression, and the same geometry pops up again and again. And then I went out and read what people were writing about it. The tech press was neutral to cautiously positive, and the games press had the mostly the same negative response that they did to AI-generated Minecraft demo from last year.

How does this happen?

The bit where I talk about my old insecurities

To figure this out, I had to interrogate my own experience as a tech journalist. I spent the first decade of my career reviewing and testing gaming hardware. I never really talked about it, but I always felt like a second-class citizen because I focused on the hardware and software that normal people use on home PCs, instead of the kind of computers that Fortune 500 companies buy. Almost every PC hardware and software event I’d attend was focused on businesses first and games sixth, if at all.

I learned about in the beginning computers because the PC was the place to play the most incredible games people were making in the mid 90s. Wolfenstein and X-Wing were only on PC, and strategy games like X-Com didn’t exist anywhere else. I learned how to make computers work because it was the only way I could get games to run on my 25MHz computer with 4MB or RAM and a 25MB hard drive. When I started working at Ars Technica, during the early days of 3D acceleration and the explosion of PC gaming that came with DirectX and Windows 95/98, it just made sense to me to dig into gaming hardware.

The trick is that games have always been at the bleeding edge of computing. And if you really wanted to learn what you could do with computers—if you wanted a glimpse of the future of computing, games held it.

When I started covering Microsoft and Windows for MaximumPC, it became clear that most of the folks who worked in that space, in the press and at Microsoft itself, looked at gaming as an interesting novelty that didn’t ever really meet the bar for serious consideration. After all, Fortune 500 companies weren’t buying 3D accelerators for their fleet PCs and before Steam grew into a sales juggernaut, PC game sales were tiny compared to consoles. Hell, even in meetings with MS executives talking about the pre-release prospects for Windows XP, they downplayed the importance of the home gaming market as a niche of a niche.

Fast forward 20 years. The PC gaming market is enormous, the AI models that are driving billions of dollars of investment in datacenters are running on tech originally designed for playing games, and the gaming consoles are essentially PCs with lightly-customized processors and vendor-specific operating systems. Gaming has eaten the tech world.

And the tech press and the folks at Microsoft and OpenAI making this AI software are still woefully under-equipped to understand games. I’m sure they know a ton about the business of games, but have no idea why people actually play games. They don’t get why someone would choose to play Control instead of watching a baseball game or a reality TV show. Or why my daughter is just as captivated with Super Mario Bros as I was when I was her age. They don’t get why games are fun.

At a quick glance, this Quake 2 AI demo looks like a game. Someone who doesn’t play games could look at it and think “that’s a game”. When I look at it, I see something that looks game-ish but doesn’t behave like a game. No one is going to spend more than a few minutes checking this out as a novelty.

And yet, Microsoft is spending billions of dollars on this nonsense. The people driving these product decisions seem to have all of the unearned confidence of a middle-aged MBA who watched a few YouTube videos about making video games and never heard of the Dunning-Kruger Effect.

(FWIW, there seem to be quite a few people at Microsoft and other companies who precisely understand this problem. They send me messages on Signal every time I write a column about this. Feel free to share your side of the story if you work at companies using AI for stupid stuff and want to let me know how it’s going. I’m happy to keep your identity secret.)

The TLDR is that this AI-generated Quake 2 thing is a gimmick. It’s interactive video, and yeah, I guess that’s impressive, but it doesn’t capture the magic of interlocking systems that still makes playing Quake 2 fun in 2025, despite the fact that it’s almost 30 years old and dates to the earliest days of 3D graphics.

What’s Next?

This week’s actionable takeaways are a little more ephemeral than normal. Interrogate the folks delivering your news and talking about AI in the press. If they’re promising stuff that sounds too good to be true, consider that it probably is too good to be true.

In the case of this Quake 2 AI demo, I don’t understand why anyone at Microsoft thought it was a good idea to release it as a playable demo. While it’s possible to get a carefully controlled 90 second video out of it that almost looks like a real video game, the moment you actually control it yourself, its flaws are obvious.

One of the stated use cases for this model, in the Microsoft Research paper that was published in Nature, is to allow developers to build prototypes faster and more efficiently than traditional game dev makes possible. This sounds really good, right?

If you don’t know, most publishers want to see some sort of working prototype as well as a fully-fleshed out example of the eventual art direction before they’ll write a check for any games today. Combined, these typically represent many person-months of work for some subset of your team and as a result they can cost tens or hundreds of thousands of dollars to create.

As someone who has pitched a bunch of games to publishers, content partners, and venture capitalists over the last decade, I’m enough of an expert to say that this dog won’t hunt. If I treated this demo as a serious project and pitched it to anyone I’ve sold projects to in the past, they’d stop taking my calls after seeing this. It’s laughable.

This week’s shoutout goes to this fantastic post, which lists free, open, or at least pay-once options for key Adobe Creative Cloud software. So if you want to get clear of annual subscriptions with questionable AI policies, you probably have more options than you knew about. I’ve been using Affinity Photo and Designer for years now, and like them quite a bit, and I have a handful of new audio editors to test out after checking out this list.

It’s springtime in my hemisphere, and as tradition demands, I’ve spent the last few weekends clearing everything out of one room in my house, then cleaning everything from ceiling to floor (if you think about it, starting at the ceiling and working your way down makes a lot more sense and lets gravity help with the detritus removal process, but I digress). It’s a lot of work, but anytime I feel my back twinge during the week, I get a blast of the intense sense of satisfaction that was my reward for all that hard work.

Yes, that keyboard is extremely dirty, but it isn’t making your PC any slower.

I don’t think we really have an equivalent of spring cleaning for technology. In the old days, an operating system reinstall was the second or third thing Apple and Google recommended if your phone was behaving oddly, and the annual reinstall of Windows or OSX was just part of the cost of doing business. It was so popular a topic that we did annual cover stories at Maximum PC telling people what tools they could use to make that reinstall as painless as possible. People loved that new Windows feel, before you installed all the software that jacked up your computer.

But over the last decade, reinstalling the operating has become the nuclear option, for when you have Really Serious Problems with your devices. The data on our computers and phones is so important to us that it doesn’t make sense to take the risk of a clean OS install unless we absolutely have to.

Overall, this is a good change. It means that our computers are reliable enough that they usually work from the time we buy them until the time we replace them. And our computers are lasting longer than ever before—my typical laptop lasts at least 4 or 5 years now, and I’m fairly typical, with a fair amount of variation based on whether you buy high- or low-end hardware. The only downside of computers lasting longer is that if you don’t make a concerted effort to clear out stuff you don’t use, your computers will end up full of cruft.

Cruft Sounds Bad. It’s Bad Right?

What is computer cruft exactly? It’s all the little stuff that lingers on your PC even after you’ve removed the software that installed it, from leftover files and applications that you haven’t uninstalled to the unnecessary options in your right-click context menus. It’s the programs that you leave running in your system tray, the weird drivers that some software installs, and a lot more. (There are a boatload of somewhat scammy applications that promise to clean this crap out of your PC, but in my experience those apps usually just make your PC problems worse.)

If you like this kind of a block map, you should check out Eclectic Light’s article about OS X becoming more immutable every year.

Does that cruft make your computer slower? The answer is “It depends”, mainly on what operating system you use. Since Apple started transitioning from Intel to Apple CPUs, they’ve been slowly transforming MacOS to essentially be an immutable operating system.

Immutable, in this instance, means that big chunks of the system files that the computer uses to boot and run software are locked down so that they can only be changed as part of operating system updates. It’s common in phones and has seen growing popularity in the Linux world for certain types of installs, including the Steam Deck.

The immutable OS renders modern Macs running Apple silicon relatively immune to cruft. You can install software that runs in the background and crushes your performance, but uninstalling that software will fix the problem. The worst-case fix for a crufted up Mac these days is to create a new user account and start using it instead of the one you ruined.

Windows is a little more complicated. Because you can install software that interacts with every part of the OS all the way down to the kernel, you can end up in situations where software you’ve installed makes Windows janky and slow. I’m in the middle of some long-term testing that’s quantifying the impact of a crapped-up Windows install on performance. More on that soon, but the early results indicate that there’s isn’t really a perf hit for running an old install of Windows, beyond the stuff running in the background and taking up resources. These tests aren’t making me rethink my basic strategy of only reinstalling Windows when I absolutely have to.

What Do You Actually Need to Maintain?

This is weird advice to give for a newsletter, where headlines like “YOUR COMPUTER IS MAKING YOU CONSTIPATED” or something more sensationalist would probably drive mad engagement, but generally speaking, computers these days don’t need a ton of actual maintenance. They’re reliable and will keep working for years on end, even when faced with almost malicious levels of neglect. I try to focus only on the work that will have an actual impact on my user experience, so with that in mind, here’s the regular maintenance work I do on my PCs and my family’s Macs.

• Pay attention to the software that’s allowed to run all the time. It can have a substantial impact on your computer’s performance and battery life.

• On Windows, these apps are usually hidden in your system tray (you’ll need to click the ^ next to the networking icon to see everything that’s running). Those applications make your computer take longer to boot and use CPU, memory, and also battery. For me, tools that I use all the time like Dropbox, Slack, Discord, 1Password and Fan Control make the cut. I disable all of the other crap in there using the Startup tab in Task Manager. (If you accidentally disable something that you need to make your computer work, you can always reenable it in the same spot).

On modern versions of Windows, the apps running in your System Tray (and using system memory and CPU cycles) stay hidden unless you explicitly choose to show them.

• On the Mac, the always running programs live on the right side of your Menu Bar. You can see what runs at startup and what has permission to stay running in the background by going to Settings > General > Login Items.

• I make sure that I keep my operating system, applications, and drivers updated. Phones, tablets, and Macs have spoiled me over the last decade, because that work usually happens invisibly in the background on those platforms. But Windows users still need to hit Windows Update as well as the tools provided by your hardware vendors to keep everything updated.

• I monitor the amount of storage that’s free on my drives—for reasons I don’t fully understand, running out of disk space makes Windows slow to a crawl. There are a bazillion tools that help you visualize where all the space on your hard drive went, but my current favorite is WizTree. Unlike WinDirStat, which was built for now-ancient OSes and file systems, and has to actually traverse each folder on your drive to suss out how much space each file takes, WizTree reads the filesystem itself and is really fast as a result. It’s free for personal use, and I highly recommend it. PS. The thing that’s eating all of your drive space is probably games.

  

  WizTree gives you a good way to find out which applications are hogging space on your drive, even if the individual files are buried deep in subfolders.

• It’s also a good time to look at your phones, tablets, and other devices and remove the software that you aren’t using anymore (or at least take a look at the Privacy settings panel and audit the apps that have access to location tracking, health data, microphone access, local network access, and anything else that you’d like to limit. Either uninstall the offending apps or disable their access.

• That’s really it. I generally only recommend you do things like update your BIOS or firmware if you have a specific need. While BIOS updates are much lower risk now than they used to be, recovering from a failed update is still a pain in the ass and even if things go well, there’s a strong likelihood that you’ll have to reset your settings, potentially hammering perf. The exception is if you’re running a 13th- or 14th-gen Intel CPU and haven’t updated your BIOS since last fall, you should 100% update your BIOS with the current one, which will apply the fix for the over-voltage problems that killed a bunch of Raptor Lake processors over the last few years.

That’s pretty much, aside from physically cleaning my devices.

An Aside On the Importance of Actually Cleaning Your Devices

Full disclosure: This section has a bunch of affiliate links to Amazon, so if you buy something I recommend here I’ll get a portion of each sale. Also, I get it if you don’t like Amazon. You won’t hurt my feelings if you buy stuff elsewhere. It’s probably a good idea to get in the habit of physically cleaning your devices more often than once a year. Most of the recommendations I found about phone cleanliness date back to the pandemic, when people seemed slightly more concerned with hygiene. The current guidelines are that you should wipe your phone down daily, and remove the case for a thorough cleaning on the regular.

I like these Zeiss lens wipes for phones and computer screens, but they include a ton of packaging (and disposable wipes are never great for the environment), so I try to only use them on the go. I keep good microfibers and bottled cleaner around when I’m at home, but you want to make sure that you’re always applying the cleaner to your microfiber, rather than directly to the device so that you avoid drips that can kill sensitive electronics. If you get larger microfibers, they’ll do double duty—a slightly damp cloth is a great way to clean dust off the outside of your computer case or laptop.

I wipe my keyboard and mouse down less frequently than I do my phone, but I also tend to have clean hands at my PC, since I rarely eat at the computer. I have a soft dust brush on my desk, which is great for brushing crumbs and bits of dead skin and hair off of my keyboard, mouse mat, and desktop. Hidden in its handle, it has a small metal pick and a few tiny brushes for cleaning small, delicate devices with lots of crevices, like Airpods. And it does triple duty as a fidget when I’m on calls.

I live in a particularly dusty house, because we keep the windows open roughly 300 days a year. That means the dust filters on my devices are essential and need regular cleaning. While you use canned air for minor dust problems or pop them off and clean them in the sink for bigger problems, I invested in an electric duster a few years ago, and use it all the time. It’s great for blasting crud out of dust filters on my desktop PC, blowing dust out of laptops and consoles, and cleaning stuff like reusable vacuum cleaner filters. I even use it to deep clean our dehumidifier and air filters (the inbound areas of household air filters get surprisingly nasty).

When I bought a fancy mechanical keyboard with white keycaps, I didn’t consider how nasty the natural oil on my skin would make the caps. To remove the finger goo without a bunch of scrubbing, I got an ultrasonic bath cleaner. A few drops of dish soap in the bath with nastiest of keycaps blasts all the crud right off. Be careful if you have fancy multishot caps though, as the ultrasonic bath can cause them to delaminate if there are voids in the material. I also use the ultrasonic bath to clean my eyeglasses, but again you should be careful if your glasses have scratches. Using the ultrasonic bath on scratched lenses can cause bubbles to form in your lens coatings.

What’s Next?

I’ve got more real-life spring cleaning to do in the house, and it’s about time to plant the summer garden (tomatoes, lettuce, spring onions, cucumbers, and maybe a watermelon vine or two are on the plan for this year). Eventually I’ll get to cleaning my office and probably also the garage, but probably not while I’m in the post-GDC busy season.

I’ve been playing 33 Immortals this week, which is an incredible large-scale co-op roguelike. It’s a bit intimidating at first, but it’s giving me real early-2000s MMO raid feels, without the awful early-2000s MMO grind.

Disclaimer: This newsletter contains links that generate affiliate revenue for me if you buy stuff using them.

Your Alexa is in the news lately, and it’s not because the device has suddenly become good for something more than setting timers and turning smart bulbs on and off. Your Amazon Echo devices that are powered by Alexa are getting a souped up AI mode called Alexa+ later this year, but they’re disabling the feature that allowed users to specify that all queries should run locally at the same time. After March 28th, all queries on these devices will now be sent to Amazon’s cloud for processing (and storage and future retrieval by Amazon employees or nefarious hackers).

Maybe it’s time to take your Amazon Echo out with the trash?

The local processing feature was only available on relatively new hardware, the Echo (4th gen), Echo Show 10, and Echo Show 15, so if you were using any other Alexa devices in your home, they were already sending all of your queries to Amazon’s servers.

Yes, it’s bad that Amazon is pulling an advertised and supported feature from existing hardware and they should have to refund anyone who bought one of these devices. But for the vast majority of people using Alexa devices, this is no change. It’s exactly how Alexa has worked from its launch in 2014 until these newer devices rolled out.

The more important question is: “Should you have ever put one of these devices in your house in the first place?”

When Amazon rolled out the Alexa, way back in the heady days of 2014, I was captivated by the little black cylindrical computer that you could talk to and kind of get to do useful stuff, when everything worked right. Amazon played up the Star Trek library computer vibe of the thing, and it was a fun, if mostly useless, way to interact with a computer.

I enjoyed playing 20 questions with the Akinator and I used it to listen to music in the rooms where I didn’t have a Sonos setup. Eventually I connected Alexa to my smart home hub, and it quickly supplanted the tetchy web of sensors I was using to turn lights on and off automatically in my home. Has it ever been easier to say “Alexa turn on the bedroom lights” than it was to flip a switch? Maybe when I had both hands full of a squirmy baby, but not since then. But that’s where it stopped. In 2025, I use voice agents mostly the same as I did in 2015, for music playback, setting timers when I’m cooking or doing chores, and controlling smart home stuff.

Why do we keep these things around?

At a certain level, we’ve become inured to the high level of surveillance that we experience in the service of selling us new shoes and underwear. Do an idle search or spend too many seconds looking at a paid influencer spot on Instagram, and suddenly every ad you see will be for the advertised product. Whatever you do, don’t click on this affiliate link for a 40 pound bag of dehydrated cereal marshmallows, or you’ll start seeing marshmallow ads everywhere.

In the beginning, I felt OK with these things because the promise has always been “It detects the wake word on the device and then sends everything after that to the cloud”. And while there were going to be some errors with that system, sometimes you take the good with the bad, right?

My thoughts on this in the beginning were that we should assume best intent—that is, believe that these devices are actually working the way they claim to be. But then Amazon had a series of problems with data privacy—they stored recordings of kids indefinitely, they let employees and contractors access data from Ring cameras inside people’s houses, they handed user data over to the police without requiring warrants for a decade, they pay employees to listen to recordings to make the service work better, rampant allegations of improper data security, and the list goes on.

I got rid of my Alexa devices a few years ago¹ , but I kept speakers from Google in the house, and still have Siri turned on my Apple devices. Given the rise of authoritarianism in the United States, I’m not sure having these speakers around is such a good idea. When people are being turned away at the border, seemingly because they criticized the current administration, having an always-on, digital snitch connected to servers that are accessible to the US government seems iffy at best.

What are the other alternatives?

Now that it seems like getting rid of all of the corporate-controlled voice assistants is probably the right choice, I’d hate to lose the convenience of voice control for things like kitchen timers, music playback, and smart home control. There are a lot of options from a variety of companies, but as is often the case, the Open Source folks are here to save the day.

Home Assistant’s Voice Preview is a microphone and speaker that ties directly into the Home Assistant ecosystem. You can connect it either to a variety of cloud providers or run local infrastructure inside your home to parse voice queries.

Home Assistant has a voice module that you can use to replace some of the functionality of an Alexa, Google Home, or Siri with, and run it locally (or in Home Assistant’s Nabu Casa cloud instances). From the reviews I’ve read, it’s not a one-to-one upgrade, the audio quality on the HA Voice speaker is such that you won’t use it for music, and it lacks access to the deep information graph that Google, Amazon, and Apple’s voice assistants can tap into by default.

In true Home Assistant fashion, they’ll let you replace any of the three core components—speech-to-text, the conversation agent that parses your voice commands and infers your intent from them, and text-to-speech with pretty much anything you’d like. If you want to put a giant NVIDIA GPU in a box in your garage and run a local model as the conversation agent, they’ll let you do that.

If you want to hook it up to a cloud-based model from ChatGPT or Google Gemini you can do that too. And of course, it’s much easier and cheaper to get the cloud-based solutions running than to do the same thing locally. (Please don’t do the work to ditch Google or Amazon’s voice assistant and then run ChatGPT in its place, that completely defeats the purpose.)

The unfortunate reality is that running these kinds of complicated, interconnected services locally is difficult. There are a lot of moving pieces, it’s likely to be fiddly, and some stuff just won’t work the same way it does with the commercial solutions. There are other contenders in this space, using varying levels of Open Sourced-ness, but the only ones that are turnkey have the same (or worse) potential privacy problems as the solutions from Apple, Google, and Amazon.

Or you could just keep using Siri and hope that Apple doesn’t roll over for the US government the same way they just did for the UK government.

What’s Next?

I’ve been at the Game Developer’s Conference this week and have had a bunch of lovely chats with people from around the world who work in all aspects of game dev. But after a week of non-stop days and nights, my social battery is empty and I’m looking forward to coming home this weekend and spending some quality time with some of the games I saw. I talked about some of the different things I saw on this week’s episodes of The Full Nerd and the Tech Pod (The Full Nerd is up now and this week’s Tech Pod will drop early Sunday morning).

I’ve also been enjoying Om Malik and Fred Vogelstein’s newsletter, Crazy Stupid Tech. Om and Fred have both covered emerging tech for investors for a long time and they have a good track record of focusing on the products and teams that end up making an actual positive impact for people, not just investors and shareholders. Om’s latest post about cheap Chinese IoT devices that phone home has left me rethinking the devices I want to give the ability sniff around in my private subnets.

I’m pretty sure that I had the most strongly positive response to Fastmail that I’d ever had to any piece of software. Sure, Fastmail is just email, but it’s a different kind of email service. It feels like it’s designed to work explicitly for me. There’s no enshittification, no upgrades or paywalled features. I don’t think I’ve ever gone from a free trial to a paying user of a piece of software so quickly in my life. Looking back at my email history, I signed up for their free 30-day trial at 6:24PM on a Saturday night, set up a couple of email addresses, and then changed to a paid subscription less than an hour later, at 7:18PM.

I didn’t really understand why I signed up so quickly until I went to a flea market and bought an old CRT TV. How’s that for a segue?

Anyway, last weekend, I went to the Silicon Valley Electronics Flea Market with some friends. It’s a once-a-month, warm-seasons-only swap meet that one of my Bluesky followers described as the place to unload all the “crap you stole from work 30 years ago”. There’s every kind of ancient computer on-hand, ephemera from the chip-making business, and all sorts of still useful retro tech, like CRT monitors and TVs. It’s the kind of place that you’ll see someone selling a tub full of stepper motors ($2 apiece) right next to someone who has 40 beautifully restored Xbox 360s ($30-50 depending on condition).

It’s one of those Bay Area fixtures—like the Musee Mechanique or Urban Ore—a destination unto itself that’s been around forever and is worth visiting just to experience. I’ve been going for the last few years, and I’ve enjoyed my trips there despite having to get up at the crack of dawn on Sunday morning and never leaving with more than a roll of Kapton tape or an apple fritter and a coffee or two.

Until this week, at least. At the last booth we hit, sitting on the asphalt surrounded by a slew of lesser CRTs, I spotted the exact model of TV that my first computers and my NES was hooked up to as a kid. In many ways, this TV is a core part of my origin story. I wrote my first BASIC program on a TI-99/4A that was hooked up to it. I discovered a deep love of video games that led to learning how computers work. And yes, it still works, complete with clicky knobs and bunny ears, and the classy walnut texture on the front that screamed “this is a great TV” in the 80s. Before I knew what had happened, I handed the seller $40 and I owned the same TV that I spent countless hours in front of as a kid.

An 11-inch color GE television, which is the exact same model that I played NES and SNES games on as a kid.

So, that’s how I ended up buying an 11-inch color TV in 2025. I’m pretty resistant to nostalgia, so up until that moment, I was pretty sure that my time with CRTs passed when I got rid of my last behemoth Sony Trinitron monitor 20 years ago. It turns out I just needed the right one to pull me back in and get me excited again.

And Here We Are, Back at Fastmail

Fastmail made me excited about email again. That’s why I signed up so quickly and that’s why I’m writing about it today. It did it with a series of workmanlike features and a steadfast approach to usability at all costs.

You see, I’d been thinking about email the same way I thought about CRTs—as a dead technology that I’d written off. I had given up, partly because of spam, but also because the tools just weren’t improving to match the way I use email today. I don’t want or need AI writing assistance in my email client. I don’t need complicated workflow automation for customer service inquiries.

I just need tools that makes it easy for me to manage a large number of daily emails, sent to a handful of different addresses, each with its own unique set of rules, signatures, and other contexts.

This is where Fastmail shines. Unlike traditional email services—Gmail, Proton, Hey, Outlook.com, and the rest—which are designed to be used with a single email address and have varying levels of accommodations for people who use multiple emails, Fastmail is designed for people who use multiple emails on multiple domains and want to manage everything from a single unified interface.

Sure, when you sign up for Fastmail, you can register bob@fastmail.com. But it shines when hook your Fastmail account up to every domain and email address you own. And unlike every other email provider )and my domain registrar), they don’t charge me a few bucks a month for each email address I use. More on the multiple domain stuff later, but I was paying more than the $60 a year that Fastmail charges for forwarding addresses on every domain I own.

Why Fastmail Is Good

• It handles the complicated business of setting up email on your own domain. No one should run their own email server in the modern era. It’s difficult, the penalty for messing something up is that the domain becomes unusable for email for all eternity, and there’s a bunch of sticky legal stuff that comes with accepting files from random people on the Internet.

• The folder structure makes a ton of sense. You can set up basic rules on a per-folder basis, allowing for automatic purging of old emails, setting a default account that new messages and replies to emails in that folder use, and even share access on a per-folder folders with other people (with a toggle allowing you to choose whether you want to share read status on individual messages).

• Email filtering on inbound messages happens based on the account it’s received from, not the information on headers. So you can say put all emails from Account X in this folder, and then run your filtering rules on them there. This prevents emails that use bcc: or fake headers from bypassing the filtering rules you setup on your inbox.

• The clients that I’ve tested (web and iOS) are fast, lightweight, don’t have a ton of tracking cookies and other nonsense, and generally get out of the way of the work I need to do. (You can also use traditional IMAP clients if you want, but you lose access to some of the cooler features when you do that.)

• There are several ways to generate one-off masked email addresses you can use in place of your actual email in case one of your accounts gets compromised or someplace you signed up for starts bombarding you with spam. You can either create them using some basic rules (which are established on a per email address or per-domain basis) or using their 1Password extension to create a new email address when you create a new login in the 1Password apps and extensions.

• It includes a fine calendar app, which will unify all of your existing calendars into one view, if that’s the kind of thing you like.

• You all know how I feel about notifications, and I’m impressed with how they operate these. Each device has its own notification settings—and you can set either general rules for notifications (all emails from contacts trigger a notification or only emails from VIPs notify you or nothing ever triggers a notification) or you can set up rules that trigger notifications based on criteria you specify.

• If you want to use your Inbox as a To-Do list, you can snooze messages. (I still think that it’s a bad idea to do this though.)

• Their anti-spam tools provide good base level protection out of the box and a great deal of granular control for folks who have edge cases (if you want to forward all of their email from a spammy Gmail account or if you want to auto-delete especially spammy messages on arrival).

• The default migration path, which has you log in to your old account from Fastmail, will pull your contacts, calendar, and email from your other accounts and set them all to stay in sync with the old accounts, if you choose to do that.

• One click unsubscribe from mailing lists you’re having second thoughts about getting in your inbox (not this newsletter please!)

• The setup, especially if you’re using your own domains and already have websites hosted on them, is tricky and involved adjusting some DNS settings. If you don’t know, adjusting DNS settings is fraught with peril, so if you’re nervous about this, their online help responded really quickly when I asked some questions. I’m competent but extremely nervous about DNS changes, and I found their documentation to be good enough that I was able to set up all of domains in a few hours one night. (As we all know, making scary DNS changes is always best to do late at night.)

I could go on, but the TLDR is that this service is worth much more than $60 a year (but please don’t tell them that).

If you want to try it out, I have a referral code you can use to get 10% off your first year with Fastmail (Full disclosure: I get $10 when you activate your account, but I also wouldn’t recommend something like this if it didn’t kick ass).

What’s Next?

As I’m waiting for the hardware I need to connect my MiSTeR to my new TV, I’ve spent some time over at my pal Wes Fenlon’s excellent twice-monthly emulation news newsletter Read-Only Memo. Wes is a Tech Pod regular and has been doing amazing writing on the topic for more than a decade now, so I’m always excited to see a new issue drop into my inbox. PS if you don’t know about ShaderGlass but want an easier way to drop fancy filters on your old games, you should check out the most recent edition.

The first week of my paid subscription over here has gone great, despite forgetting to add the “Hey, this newsletter is reader-supported!” reminder in the post. Thanks to everyone who signed up, gave me feedback, or just sent a nice message. I’d always rather work for readers than advertisers, so if you find my work here useful, please consider upgrading.

This week’s newsletter has me thinking about all of the core infrastructure that we’ve ceded to spammers. Phones, text messages, faxes, paper mail, and email are all overrun by spam now. I feel like better tools have helped me reclaim my email, but I don’t think I have the same options on more closed networks, like phones and text messages

The place I’m most worried about spam is the web. I talked about it last week, but AI is empowering new frontiers in spam on the web, and I’m not OK ceding it to the spammers. This week I’m going to spend some time looking at the tools and techniques that people are using to excise different types of spam from their lives. If you have a favorite one, please mash that reply button and let me know

Thanks for reading. I hope you have a lovely week, and as always, if you enjoy the newsletter, please consider subscribing!

One of the thoughts I see regularly is that AI is going to save us from the deleterious effects of the attention economy. AI is somehow going to magically help us deal with the increased cognitive load that technology requires, it’s going to solve the problems with the attention economy, and it’s going to be the end of enshittification. That’s a lovely thought, but I don’t believe it for a minute.

If you’ve followed me on podcasts or Bluesky, you know that I’m extremely skeptical about the likelihood of replacing human artists and writers with generative AI. Sure, it’s bad that it makes people who use it stupider and yeah AI reverts to the mean by design, and no, I probably wouldn’t build any commercial product on something that’s unprotected by copyright unless you do substantial modifications. But I do think AI is quite good at a few things—translating work from one format to another, generating certain types of code, backstopping humans in difficult pattern recognition roles, and raising the for accessibility in video and audio.

But the thing I think generative AI is the absolute best at churning out content that sounds plausible. It can do that at a rate that’s limited only by your budget. That means the easiest way to make money with AI right now is spam.

Air Purifiers Were a Bellwether

This famously came to a head a few years ago with air filters. There used to be a handful of sites that did in-depth testing of them, but because each filter is a relatively high-dollar item, reviewing them was a profitable enterprise. People who are looking at air filter reviews are typically in the market, so a sizeable number of reviews led to clicks into online stores, which in turn generates sweet affiliate paychecks. I actually love using sites that are supported by affiliate revenue, because I know if they give me bad purchasing recommendations, I’m less likely to trust their reviews (and click their affiliate links) again in the future. It’s a powerful incentive to write good reviews.

The sites that specialized in reviewing household air filters had to spend a ton of money and time buying a bunch of them, establishing a solid testing methodology, spending untold hours testing them, then collecting the results into a meaningful, easy-to-read reviews. Doing this kind of meat-and-potatoes review is incredibly labor intensive, but it was worth the effort. Separating the great from the merely good products would make or break a site.

Pre-AI, these high-investment, high-return sites weren’t particularly good targets for spammers. Sure, people copied reviews from the originator’s sites, but Google is constantly scanning the web, and it was pretty good about prioritizing results from the original creators instead of some yahoos who just lifted another site’s reviews. Sometimes being the first poster has an advantage.

But generative AI upended that model. These days, if you want a site with a bunch of air filter reviews on it, all you have to do is ask ChatGPT to write a reviews of each of the most popular air filters that are on the market. Then ask it to make a best of list, and link them all together. Are the reviews accurate? Who knows? Are they fair and representative of the products? I don’t know. Do they provide actionable buying advice? Sure. Do the people who use this tech to generate a bunch of spam care if people spend their hard-earned money based on terrible machine-generated advice? Not a whit, at least as long as they keep whatever they buy past the return window.

It’s really easy to review products using generative AI. You don’t even have to spend any time with them!

When you use something like ChatGPT to generate a bunch of bogus reviews, it indiscriminately aggregates reviews from all over the web—editorial reviews, user reviews, comments on social media, YouTube comments, and random blogs. It’s untraceable and makes generative AI the steam engine that empowers a new Industrial Revolution of stealing other people’s content.

Once you’ve built out a toolbox of relatively simple scripts and prompts to generate a site, you can get ChatGPT to generate a list of high value search terms. Then you use that list to plant AI-generated seeds for each of those terms in the form of real looking sites, complete with outbound links to other sites, images, and all the things you’d expect to see. You can create a positive feedback loop of content generation that only stops when you run out of money. But when you pair the AI-generated sites with low-effort affiliate and display ads, you won’t ever run out of money! Instead, you’ll create a flywheel of spam—an infinite ouroboros of spammy sites that each generate a little bit of passive income. This is an actual business model, presumably from the same people who brought us the plague of ad-driven content that looks kind of like it belongs on this website, like Demand Media and Taboola.

ChatGPT output showing how excited the generative AI is to help you build a spam empire

And all you had to do to make it happen was make the web just a little bit more useless.

On a positive note, there is actually a happy ending to the air purifier story. The search results today for that particular term are quite useful. It’s because a small site made a big stink about Google ranking AI-generated slop that was being hosted on sites with general search ranking ahead of smaller specialist sites a few years ago. It went viral, and apparently Google made some tweaks, at least to the way the search algorithm handles air purifiers.

But the overall impact of garbage AI-generated content is everywhere and it’s going to continue to outpace the tools designed to combat it. Since I started working on the newsletter, I’ve been looking for decent stock image services, and the low end of that market is completely overrun. The top results in every category on all of the big stock image sites are low-effort AI slop, to the point that you can’t find interesting human-generated Creative Commons or public domain work outside of geriatric sites like Flickr. While the high-dollar services, like Getty, are uninfested so far, anyplace you can buy rights to a specific image for a few bucks has been overrun.

How Do We Combat Webspam?

I like to end these with something actionable that you can do in a few minutes to make your computing experience better. But unfortunately, I can’t do much to help you with webspam. I’ve started testing some alternative web searches, but I’m not thrilled with any of the results so far. (I’ve spent time with both Duck Duck Go and Kagi so far, so if you have other recommendations you’d like me to test out, hit reply).

My basic rule for generative AI is the same as it is for vampires—I don’t invite either of them into my house. That means while I’m continuing to use Google, I disable as much of its AI-generated business as I can using the UDM=14 trick. (Ernie Smith, over at the always awesome Tedium blog, clocked this filter last year, and found that it not only disabled AI, but also all the other crap they’ve been adding to search results over the last 15 years.) What’s the trick? If you append your Google searches with ?udm=14 then you get 2004-style Google results—no knowledge graph, no additional context from the search engine, no embedded news results, no image results, nothing. Just straight web results that take you to the website you were probably looking for in the first place. It’s really nice.

Appending some cryptic combination of letters and numbers manually every time you search is a pain, so there are a variety of tools that will help you. The fastest and easiest thing to do, at least for Chrome and Firefox users is to go to tenbluelinks.org and follow the instructions on the page to create a custom search for your browser and then set it as the default for location bar searches in your browser’s settings pane. It should take just a few minutes and it’s been a significant upgrade to the way I search. (If you’re worried about passing all your searches through a third party, don’t be. I tested it on Chrome and Firefox on PC, and the people running the site have just used the OpenSearch spec to create a custom search that goes directly to Google. They’re not going to know anything about whatever dark business you’re searching for.)

If you’re on Safari, it’s a little more complicated. You can head to this page and use this userscript and the Userscripts extension for Safari to replace all Google search links with the UDM=14 versions.

So while you and I can’t fix webspam, we can prevent you from getting hit with crappy AI-generated summaries of AI-generated pages every time you search for something.

What’s Next?

A few weeks ago, I wrote about turning off almost all of my notifications, and the results have been pretty impressive so far. A good day of work conversations in Discord or Slack can still blow out my daily average, but as a whole my time spent on the phone is more directed by me and less directed by apps interrupting me. The first few days after I trimmed the notifications way down were weird. I felt lonely, in the same way I do when my group chats and small Discord groups are quieter than normal. But after a few days with that sweet, sweet brand interaction and my endless thirst for dopamine was successfully being slaked with a handful of good mobile games and an ebook or two instead of ads for cheap chain pizza. (Shout outs to my friend Sandy Weisz, who just launched Raddle this week. It’s a kick ass daily word ladder puzzle game that you should check out.)

Before and after images from the ScreenTime app in iOS, showing that my average daily notifications have dropped ~20% in two weeks.

This week I also kicked off premium subscriptions for the newsletter. As is always the case, the main newsletter is and will remain free for everyone. But I’d love to spend more time on it, expand it to other platforms (like podcasts and maybe even YouTube), and have time to track down things like interviews, and all that takes cash. If you want to chuck a few bucks my way, you’d get my eternal gratitude, an occasional special issue, and some cool bonus content. On a related note, I got my first ad offer on the newsletter, and you’ll all be pleased to know that I turned down the opportunity to annoy a few thousand people in exchange for $12.

If you’d like to talk to me about sponsoring the newsletter, please reach out. Also, it’s going to cost more than $12.

I have a couple of different potential topics in the hopper for next week, but I haven’t honed in on one yet. As always, feel free to hit the reply button and let me know what you’d like to see. I got a bunch of great suggestions last week, and really appreciate the general feedback on what I’m doing over here. Until then, please consider subscribing, sharing the newsletter with a friend, or saying something nice about it on social!

If you follow tech news, you’ve likely heard about the impending Windows-pocalypse. On October 14, 2025, Microsoft is officially ending support for Windows 10, which means there won’t be any more security or feature updates for what is, at the time of this post, the most popular desktop operating system in the world.

Why do you need to care? Having a billion essentially unsecured computers connected to the Internet is a recipe for 15 different kinds of disaster. I’ll tell you a bit about what to expect and how to avoid disaster in a bit. First though, a quick history lesson.

A Quick History Lesson

As someone who has covered Windows for a long time now, this feels like a problem that we shouldn’t have to face. After all, when Microsoft launched Windows 10 in 2015, it was soft-pitched as “the last Windows you'll ever buy”. At the time, Microsoft was high on the idea of Windows as a Service. Windows 10 was envisioned as a direct pipeline from Redmond to all of our wallets in the form of new subscription services like OneDrive, Microsoft 365 (aka Office), and Xbox Game Pass.

After only six short years, for reasons that no one really understands, Microsoft ditched the “last Windows” plan and introduced Windows 11. Windows 11 adds support for new hardware and more advanced software, but unfortunately, it didn’t ditch the software-as-a-service business model—it just entrenched it further.

HIstorically, upgrading from one version of Windows to another is pretty straightforward. You might need to add some RAM or update drivers and software so that everything’s compatible, but most computers were good for three or four versions of Windows. In fact, from Windows 7 to 8 to 10, the Microsoft-recommended spec for hardware stayed the same—a 1GHz CPU with 4GB of RAM. (Yes, this would be a laughably underpowered computer, but you have to remember that Windows runs on an enormous number of computers around the world, and many of them are old.)

Even if you wouldn’t have a great experience running Windows 10 on a minspec Windows 7 computer, it would at least let you upgrade the machine so you stay current on security updates. That’s changed with the upgrade to Windows 11. This is really the first time Microsoft has done a clean break—leaving behind an entire era of old hardware—since the earliest days of Windows.

Why did they do this? Well, one of the core additions to Windows 11 is designed to provide added security for users and business against the increasing number of threats they face today. To help with that, they added a requirement for a special hardware module that didn’t exist in the Windows 7 era.

In order to install and run, Windows 11 requires a TPM 2.0 on the device. TPM stands for Trusted Platform Module, and put simply, the TPM provides hardware cryptographic services to software on your computer. The TPM handles everything from random number generation to cryptographic key storage, and it allows services like Secure Boot (which prevents you from booting hacked versions of your OS) and Bitlocker (which provides whole disk encryption tied to your Windows login). TPMs aren’t new technology—the first versions rolled out in the mid 2000s, but the 2.0 revision that Windows 11 requires didn’t start shipping in hardware until 2015.

An add-on TPM 2.0 module for certain Asus motherboards. On most modern computers, the TPM is embedded in the CPU or chipset.

Because of its tight integration with the low-level processes of the PC, TPMs are typically either soldered directly on the motherboard or integrated with the CPU or chipset itself. Because they need low-level access to the system, you can’t just plug a TPM into a USB port and get access to the services it provides. For the vast majority of computers out there, either your computer shipped with the right TPM and you’re good to go, or it doesn’t, and you won’t be able to run Windows 11 on that machine. As always in the PC space, there are exceptions—some computers in the mid-2010s shipped with their TPMs disabled so you can flip a setting in the BIOS to enable them. Others included an internal connector for a TPM module you can add later.

What Happens In October?

First, your Windows 10 machines aren’t going to turn off on October 26th. The old OS will continue to work indefinitely, you just won’t get security updates that fix vulnerabilities in the OS or virus definitions for Windows Defender (Microsoft’s anti-virus and anti-malware software).

The problem is the sheer number of Windows 10 machines that are still in operation provides an enormous opportunity for bad actors.

According to Statcounter—which analyzes massive amounts of web traffic to determine for OSes, device types, browsers, and more—Windows is about 72% of desktop market share (the rest is divided between MacOS, ChromeOS, and other flavors of Linux. About 60% of those Windows PCs are still running Windows 10, as of Statcounter’s January 2025 reporting. Assuming we don’t see massive adoption of Windows 11 between now and October, that means on October 26, there will be more than a billion computers that no longer get security updates.

A billion unsecured computers represents a huge opportunity for bad actors.

The last time we had a billion essentially unsecured computers connected to the Internet, things were a little different. Windows XP famously had enormous security problems—it shipped at a time when always-on, broadband connections were becoming more common, but before most people had routers that let them share their broadband with multiple devices. Your router doesn’t just let you use Wi-Fi in the bathroom, it also provides basic a basic firewall, which gives your PC the minimum protection from millions of bad actors who are constantly scanning every single IP address on the internet for known vulnerabilities. Before routers were commonplace, most people were connecting directly to the Internet. shudder

A shot from Eric Parker’s YouTube video showing how infected a Windows XP install gets after 20 minutes on a public IP.

How bad was it back then? Well, if you plugged a fully patched Windows XP SP1 machine into a public IP address in 2004, it would be completely hacked and part of a botnet within about 20 minutes. Twenty years later, it still takes about the same time.

The rise of modern routers and a much more robust Windows 10 firewall will protect these newly insecure PCs from the kind of low-effort automated attacks that hit those older Windows XP installs, but new exploits are discovered every day and attackers will undoubtedly find new ways to infect and attack Windows 10 machines post EOL. That will accelerate as other software vendors—I’m looking at web browsers here—stop pushing updates to machines running Windows 10. I imagine it won’t be terribly long before we see Windows 10 machines infected when they load the wrong web page or maybe even the wrong ad.

What Are the Alternatives?

While there are a handful of ways to bypass the TPM requirement during the Windows install, but machines that bypass the TPM requirement are officially unsupported and won’t receive updates. This limbo state is arguably a bit better than running Windows 10, but it doesn’t really address the problem.

If you have a machine that is new enough to run Windows 11 (Intel Core 8 or newer CPUs or any AMD Ryzen CPU), but it’s showing up as unsupported when you try to run the upgrade, you may need to enable the TPM in the BIOS. There isn’t a good universal guide for this, as every computer’s UEFI (that menu you can open before your computer starts Windows to change low-level settings) is different. But the basic gist is that you press del or F1 as your PC is booting to enter the UEFI, then you can search for TPM or Trusted Platform Module and find the settings you need to enable there.

A TPM motherboard header, courtesy of PC World.

There are a handful of computers that shipped in the mid-2010s with internal connectors on the motherboard for a TPM 2.0 module. You’ll need to consult your motherboard manual to find out if your machine includes one of those headers. (If you don’t know what motherboard is in your computer, you can use HWinfo to get the vendor and model.) If you have a header for the TPM 2.0 module, you need to get a compatible module for your particular motherboard. They’re typically available at Amazon and Newegg.

If none of this applies to you, it’s time for more drastic options. You’ll either need to buy (or build) a new PC or switch to another operating system. The good news about updating your PC is that anything you buy in 2025 is going to feel tons faster than your old PC, even if you’re just grabbing a $400 laptop.

If outlaying cash for a new computer isn’t in the cards, there are several good alternative operating system options. I’m going to spend more time on each of these in future posts, but I’ll give you a couple of good jumping off points so you can do your own research.

MAKE SURE YOU BACKUP EVERYTHING YOU WANT TO KEEP FROM YOUR COMPUTER BEFORE YOU INSTALL A NEW OPERATING SYSTEM. INSTALLING A NEW OPERATING SYSTEM BY NECESSITY WILL ERASE EVERYTHING ON YOUR DRIVES.

• If you do all of your work in a browser window, using online tools like Gmail, Canva, etc instead of dedicated desktop apps, consider ChromeOS Flex. Flex is a version of the OS Google ships on Chromebooks designed to be simple to install and run on pretty much any computer. I’ve used it to revitalize early 2010s-era unibody MacBooks and the results feel better than any Chromebook I’ve ever used.

• Linux Mint is a version of Linux designed to be accessible and straightforward for Linux beginners. It ships with a robust suite of applications covering everything from word processors and spreadsheets to 3D modeling and games. The installer is actually a fully functional (but slower because it’s running off of a USB drive) version of the Mint desktop, so you can try it out before you wipe your drive and commit to this new life.

• Any other version of Linux will work too—Ubuntu is a good general purpose Linux distro, Bazzite pulls the good gaming business that Valve built for the Steam Deck into a general purpose distro, Arch is great for masochists, Debian is for people who want to build another version of Linux that’s easier to use. My point isn’t to anger the Linux fans in the audience, but to point out that most versions of Linux are pretty straightforward these days, if all you’re doing is booting into a desktop environment and using a web browser..

What’s Next?

Next week, I promise we’ll talk about something a little bit less Windows-y. I’ve been making new benchmarks over at PC World, so my brain is kind of stuck on Windows right now.

Thanks to everyone who filled out the survey on last week’s newsletter. I read all the comments, and really appreciate the feedback. One of the most popular requests I had was to ditch the beige and dark-mode up the website. That’s live now, so you can check it out and let me know what you think! I’ll do more surveys in the future, but in the meantime, if you have feedback feel free to reply to this email!

Thanks to the survey respondents, I’ve almost gotten everything lined up that I need to turn on a few paid supporter options for this newsletter, and I have some really fun things planned for folks who want to support the newsletter. (The main newsletter will remain free for people who get it in email though, so don’t worry about that if you’re capped out on subscriptions!)

Thanks for reading! If you enjoy the newsletter, consider subscribing or forwarding it to a friend

Editor’s note: I’m leaving this up for the DeDRM portions for folks who downloaded their books already, but unfortunately, the fast download script doesn’t work anymore :(

If you’ve bought eBooks for the Amazon Kindle family of devices or apps, you should know that Amazon is disabling a key feature that makes it easy to strip the DRM from those books. On the 26th, they’re going to turn off manual downloads for your books. That means if you want to be able to remove the DRM that prevents you from using your Amazon eBooks on non-Amazon devices, you have a couple of days to do it while it’s still easy.

(If you’re comfortable living inside the Amazon ecosystem, you’re fine doing nothing. While Amazon is closing this hole, the universal truth of DRM is that someone will find another way to bypass Amazon’s latest fuckery in the future, so if you don’t have time to fool with it before Tuesday, don’t sweat it.)

Jason Snell wrote up a blog post explaining the process here, but the TLDR is that you need an older physical Kindle attached to your account (everything but the newest Kindles work) and a computer (Windows, Mac, or Linux) with bun installed so you can use this Kindle Bulk Downloader. Once you’ve downloaded the files from Amazon, you need Calibre—a really handy eBook and audiobook management application—with the current version of the DeDRM plugin. Make sure DeDRM plugin is installed and configured in Calibre before you import your books. Calibre strips DRM on import, which was the most time-consuming part of this process, so you don’t want to have to do it twice. I was able to read the docs, get everything set up, download almost 700 eBooks, import them into Calibre, and strip the DRM from them in about 20 minutes, but YMMV.

Congratulations! Your eBooks are free, and you can use Calibre convert them to ePub or a variety of other formats, suitable for use on whatever apps and devices you’d like to.

The Rest of the Post

Why are we talking about this here? I feel like I’m beating a very old drum here, but DRM tech enables vendor lock-in. It’s just another way that giant tech companies make it more difficult for us to switch to their competition. And of course, when you buy that cheap Kindle you’re Amazon’s customer, but you’re also a product that Amazon sells to their other customers, book publishers who sell books and buy ads on Amazon platforms.

That collection of books that I’d bought inside the Amazon ecosystem prevented me from moving to newer, better products. I tend to upgrade my eBook reader when the old one dies, so my current reader is about five years old. But looking at the latest hardware compared to my old Kindle Paperwhite the “upgrades” are meaningless. The new model upgrades the already-blinding backlight to be 25% brighter, the four-week battery lasts another week or two, and the new one is a smidge smaller and lighter than mine. While I’m complaining, the new one still doesn’t do the thing I really want, which is give me fine control over the backlight brightness at the dimmest end of its range. I don’t eyeblast myself when I’m reading in my pitch-black bedroom late at night. It feels like Amazon is barely investing in the product line that I’ve bought thousands of dollars worth of books in.

At the same time, there’s more competition in the space that ever before. Kobo has a comparable line of readers and supports a slightly more open ecosystem of electronic bookstores and libraries. (There’s some support for libraries to access the Kindle ecosystem, but it’s mostly limited to the US.) Kobo also works with third-party stores, including Humble Bundles. (Independent store Bookshop.org promises Kobo support later this year.) There are also a ton of weird eBook readers from companies like Boox that just run Android, and thus can use the normal Android apps for any or all of the eBook stores.

So, take a moment to free your purchases from the shackles of Amazon’s DRM and then go about the rest of your day, knowing that your next eBook reader can come from any company you want.

What’s Next?

That’s it for this quick bonus edition of What’s Next. I’ll have something else to think about later this week, but in the meantime, I’d really appreciate it if you’d take a minute and fill out this quick survey. It will help me decide which topics to spend time on for the newsletter and will give me ideas about what kind of premium subscription bonuses you all are into. Thanks!

You know the feeling, you’re getting a lot done and feeling really productive and then your phone makes that little beep-boop-buzz sound that says “Hey you! Pay attention to me!” So you stop what you were doing and pick up your phone to see what your favorite screen decided was so important that it had to interrupt you.

“We’re celebrating February 21st with 21% off at Wingstop if you order in the next 21 minutes!”

What were you even doing again?

Editor’s Note: Hi there! This is the first newsletter I’ve posted since I told anyone aside from a few select readers that I was doing this, and I’ve been floored by the kind words, support, and general enthusiasm you all have sent my way. Thank you so, so much! I have a bit about what I have planned for the future at the end of the letter, but please don’t hesitate to send feedback, either by replying here or by dropping me a note on Bluesky.

The Story So Far

In the beginning of the smartphone age, the only software that was able to pop any notification on your device came from core OS or the handset manufacturer if you were on Android. It was a simpler time, but this was not a great solution, primarily because it locked us all into the core apps provided by the vendors. Messaging, calendars, to-do lists, and a whole host of other app categories were generally pretty useless without notifications.

When Apple and Google finally flipped on notifications for third-party app developers, they unleashed a Promethean technology on the world. Notifications opened up markets for an explosion of revolutionary apps—mobile software that revolutionized vast swaths of our digital lives—handling old tasks (calendaring, to-dos, email, etc) in new and exciting ways and creating a whole host of thrilling new apps that helped us with everything from fitness to breaking bad habits. It was truly a time of app miracles.

Given the benefit of hindsight, I’d even argue that building a centralized notification infrastructure for phones was one of those small features that ended up shifting the center of our digital worlds from the PC to the smartphone. Notifications let the under-powered phones of the time punch far above their weight.

But, as the saying goes, with great power, comes great responsibility. The new wave of app developers were making thoughtful decisions about when and how it was OK to interrupt their users responsibly and with minimal negative impact, but trouble was brewing.

The Legion of Doom from the TV show Superfriends. The blockmeme text says “Meanwhile at the Legion of Doom”.

At the same time, the CTOs of every consumer product company in the world were looking down the barrel of another expensive annual line item on their budgets devoted to building and supporting apps for these new devices, which at the time were owned by a relatively tiny number of (mostly) wealthy white people. They needed an app strategy, but were worried about pissing away a bazillion dollars on fad gadgets that were nothing more than Blackberries for soccer moms. And then someone popped up a chart that looked like this:

A typical hockey stick style chart, showing the number of notifications a user sees a day vs. the overall revenue of a hypothetical company.

And that was the end of that.

The OS people knew that notification spam would be a problem, so they added app-level flags to the OS to give users an easy way to disable notifications for spammy apps. Any app that wanted to send notifications had to ask the user for permission, and for a while everything was OK. People said no to spammy apps, notifications got fancier, and there was balance.

But over the last 15 years, the notification spam outpaced the OS’s ability to manage it. Marketers realized that if they offered something people wanted—like delivery notifications—they’d say yes to their apps again. Then they could shovel a bunch of garbage notifications in with a few good notifications, and make it all nice and legal by burying the option to disable the spam deep in menus where reasonable people would never venture.

I’ve spent a lot of time curating my notifications, and it’s still a disaster. This is what a regular Tuesday looks like for me, after pruning out most of the marketing crap:

A screenshot of the iOS Settings > Screen Time control panel, showing the number of notifications triggered on this device on Tuesday, February 18, 2025. The daily average is 189.

So How Do We Fix Notifications?

The good news is that this has been a problem for so long that the tools for dealing with it are actually pretty good. iOS and Android both give per-app, granular control over how notifications are presented, whether they’re persistent, and whether they make a noise or cause your phone to vibrate.

You’re about to use that control to turn notifications off. For almost everything.

It took me about 30 minutes to go to the Notifications page in iOS’s Settings app and turn the vast majority of notifications off. Yes, it’s the simple solution. Yes, it feels extreme. No, once you turn them off, you aren’t going to miss them. I promise, it’ll be OK.

Most of us don’t need to know when our in-transit package reached the Kansas City hub or when the takeout food we’re driving to the store to pick up is going to be ready. It’ll be ready (or not) whether the app notifies us (or not). All that all of these notifications do is add to our already high stress levels in the form of increased cognitive load. I don’t know about you, but I can do without the extra cortisol.

I went down my list and got rid of all of my e-commerce notifications, even Amazon. I turned off everything else except:

• My messaging apps - Signal, Slack, WhatsApp, Discord, and iMessage. If I ever work on a Teams team again, I’ll grudgingly turn it back on

• The two calendar apps I use - I love Fantastical and have a shared Google calendar that we use for the family

• My to-do list app - I like to use Reminders to pop—ummmm—reminders when I arrive at home or work, or when I’m in the car on the way home if I need to stop and grab something. And yes, I’m pretty basic here. Apple’s default Reminders app suits my needs

• My most important email address - The only one that gets permission to notify me is the low traffic address where I exclusively get contract work and other job offers

• The apps that perform useful life tasks - This is a catch-all category, mostly for zero-spam smart home stuff that I have full control over, navigation apps that tell me when I need to leave to get to my next appointment, and stuff like car-charging apps (which use notifications to help me know when I need to move my car once it’s done charging)

• Libby - Libby only pops notifications when an ebook I’ve checked out from the library is due or a hold I’ve placed is available. Libby seriously rips

• Safety stuff - Weather and QuakeAlert give me notifications for imminent bad/emergency weather and potentially devastating earthquakes, respectively

• Security apps - The rise of passkeys (more on passkeys in a future newsletter) means that I have a handful of sites that pop an app notification when I try to log into them

Everything else went—food apps, reservation apps, movie apps, car services, multiplayer games, streaming services. Everything. (OK, I kept Krispy Kreme. It made the cut because donuts rip and Krispy Kreme gives good discounts in notifications.)

Once I’d done the big cull, I went into the in-app settings for each app that survived and trimmed some more. I disabled all the non-essential notifications for those apps. This took a little longer, but after another 30 or 40 minutes I was done, with two notable exceptions.

Verizon doesn’t seem to differentiate between messages required for your account, like bill reminders, and marketing messages trying to get me to upgrade my phone. Add that to the list of reasons I’m switching phone providers shortly.

Because I work in games and run a large podcast community, I do a lot of work in Discord, so those notifications need to stay on. Discord doesn’t make it easy to limit notifications to DMs at a system level, so I full-mute every server when I join it.

I have 222 apps and games on my phone, and most of them were allowed to post some kind of notification before I started this process. Now there are just over 40 applications that have permission to interrupt me. While my daily number of notifications still varies based on the number of messages I’m getting on Slack/Signal/Discord, those are generally important and/or useful. And the number of spam messages that break my flow to no benefit has decreased dramatically. Notifications are useful again.

A Few More General Tips

The lack of notifications on your phone is going to feel weird at first. You have been trained to expect it to give you a list of potential ways to entertain yourself when you’re bored, so you’re going to have to make a conscious decision about what to do with your idle/bored time. I made a home screen with my ebook reader, my favorite publications, my music app, YouTube, and a couple of favorite games on a home screen to give me a handful of curated time-wasters to choose from.

Both iOS and Android provide powerful tools to let you customize the way your phone works based on your context. I’ve used iOS’s Focus mode to set up different notification profiles that are tied to my location and/or the time of day for work, family time, podcast recording, sleep, driving, and other activities. The Android tools weren’t as robust the last time I looked at them, but there are apps like Buzzkill that tie into the OS and give you similar functionality.

Notifications are designed to be transient, so using them as an auto-generating task list isn’t a good idea, even if it seems like it should be. It’s too easy to lose a notification to an errant swipe or OS wonkiness, which is the opposite of what you want from your task list.

Apple is trying to apply AI to this problem, but my experience with the early Apple Intelligence notification summaries has had mixed and often hilarious results. It provides quick summaries for chains of notifications from the same app, but it seems to have an equally difficult time extracting meaning from both casual Discord messages and business-oriented emails, and at this point I’m mostly leaving it on for comedy value. Your mileage may vary (hit the reply button and let me know if it’s working well for you!)

A notification that summarizes three different conversations with three different people about three different topics, and attributes them all to a person named Frank, who had, in fact, shared a disturbing comment.

iOS also provides a quick interface for notification configuration on the lock screen, so you don’t have to tackle this problem all at once. You can triage them as they come in, by swiping left on them and selecting the Options button. That popup will give you deep links to the app’s notifications pane in Settings, the in-app Notifications page, and a variety of per-app snooze and perma-mute options for notifications.

A notification adjustment popup that will let me temporarily mute interruptions from this app, as well as configure them in the Settings app or in the LinkedIn app itself. It also lets me turn off notifications for the app permanently.

When you install a new app or game, be thoughtful about whether to say “Yes” when it asks for permission to send you a notification. Does Chipotle really deserve the right to interrupt you when you’re spending time with your loved ones? No. Chipotle does not.

What’s Next?

This week, I’m assigning you some (strictly optional) homework. I want you to take a look at your average number of daily notifications. (In iOS, it’s in Settings > Screen Time > See All App & Website Activity at the bottom of the page.) Take a screenshot or make a note of that number and then trim your notifications some! You don’t have to be as aggressive as I was, but think about which apps deserve permission to interrupt you. After a few days, check back and let me know what your score is. You can reply to this letter or shoot me a message on Bluesky.

Since it’s the end of the letter, I want to reiterate how excited I am by the reaction to this project. Thanks to everyone for the support and kind words, you all have me very pumped up. I’ve been working on the concept and the articles for several months, and it’s really exciting to see that so many other people are feeling the same way I do about computers and the Internet. Also, I’m sorry you’re feeling the same way I do about tech.

Like I said on last week’s TechPod, I’d much rather work for readers than advertisers or other giant companies, but at the same time I also like keeping my family fed and housed, so I’m working on a paid component for the newsletter. I have a bunch of ideas (I hope you like ‘zines!), but if you have requests or suggestions for premium offers you’d like to see, please let me know! I’m hoping to have something to share with you for next week’s edition.

Thank you for reading this far, and if you enjoy the newsletter, please consider subscribing or maybe chuck me a buck or two!

I’ve known that email is broken for a long time, but I was hoping, when I started researching this week’s newsletter, that I’d find some piece of software that would just fix my email problems—a magic bullet of an email client, setup to handle multiple email addresses seamlessly, while keeping me from missing important messages from my contacts, all without requiring a bunch of tedious daily maintenance. If AI is worth anything as a technology, shouldn’t someone be able to create a client that’s smart enough to notify me about messages that are important and never bother me about messages that aren’t?

After testing a bunch of different clients and services that I found (and that folks recommended), I’m sad to say that I didn’t find a magic bullet. There’s no easy fix for this problem, and that’s a problem for me. That’s why I’m going to declare email bankruptcy and just start over. Again.

The Problem

Part of my email problem is admittedly self-created. My main email, the one that I’ve been using for 20+ years and the one where important messages get lost constantly, is just my name at Gmail. I have a famous name, which didn’t seem like an issue when I set up the account but has become a major problem. (Thousands of people a week are confronted with forms that require an email address to show them something they want to see. And before they get their car insurance quote, free estimate for pest control, prepaid cellular phone bill, or the sticker price on a car they’re interested in, many of them put my email address into that field.) I don’t feel that bad for not anticipating that this would be an issue in 2004.

But when I talk to other people this is an almost universal problem. If you’re using the same email address for all of your personal and business and household communication, it’s almost impossible to keep it off the spam rolls. If you make one mistake, and hand your address to a company that sells their mailing lists, you’re doomed. And even if you don’t, a breach will release your address and the next thing you know, you’re on the dark web and well on your way to spam city.

This is damn near a universal problem. We all get a range of messages of wildly varying importance via email every single day—bills that need paying, notes from my kid’s teachers, doctor’s appointment reminders, important stuff from work, all the crap that you have to take care of as an adult immediately, messages from friends I haven’t talked to in a decade, and more.

I also get another few hundred emails every day that sneak past my provider’s spam filters. And those messages range in relative maliciousness from “We want to sell you some bootleg sneakers that will suck if we even bother to ship them” to “If you click on this link, we’ll empty your bank account and end up owning your home”.

The enormous gap between these two extremes means that the simple act of checking email creates an enormous cognitive load for almost everyone who uses email on the reg. There was a time, in the distant past before I got mortgage and tax missives via email, that I looked forward to the little ding of a new email message. Hell, in 1998 Tom Hanks, Meg Ryan, and Nora Ephron even made a romantic comedy about how great getting emails from random strangers was. It was about someone who definitely wasn’t Jeff Bezos dating an independent bookstore owner over email. They even play modem noises and AOL’s email sound!

The other big problem with email is that big tech has already mined all the value that they can out of electronic mail. The level of investment in email from Google, Apple, Microsoft, venture capitalists, and the rest has slowed to a trickle, at least when compared to the early, heady days of Gmail, when people were constantly building exciting new products on top of email. The TLDR is that we have to find our own solutions to these problems—email isn’t profitable enough to wait for the giant tech companies to save us.

The Solution

I said there wasn’t an easy solution up top, and there really isn’t. Once the spammers have your email address there’s not much you can do. So I’m starting over, moving all of my important communication off of my ancient Gmail account and into a place that it’s more manageable.

The first thing that I had to do is let go of the unitary inbox. In the current spam and ham climate, I don’t think it’s realistic to have one email address that will work with everything—although there are services that make it easy to have one place to check your emails. More on that in a moment.

I actually started this migration several years ago, when a malicious actor started attacking accounts attached to my email with a famous username. The process of migrating my accounts was pretty straightforward, because I’ve been a dedicated password manager user for a long time. I just opened up 1Password and searched for the email address, which gave me a convenient, auto-updating list of the sites I needed to change. Then visited the important sites one by one and updated my contact email address.

I’ve setup different email addresses for personal communications, transactions and businesses I’m buying stuff from, and public-facing emails that I have to be able to post on the web and social media. I also have a standalone address for talking to potential employers, to ensure I don’t miss any important emails.

I made sure to name the public-facing email address something that was easily identified as mine (it’s will@willsmith.ws if you want to send me an email). It’s also the only email address attached to that domain, as spammers tend to attack common name variants once they find an active mail server.

The private facing addresses are more circumscript. For the address that I hook my services up to, I chose a random username that’s just a few memorable words jammed together. (There’s a convenient username generator that’s free to use on 1Password’s site.) This has been highly effective at limiting third-party spam—my account that handles these transactions gets virtually no spam after years of use.

For my personal correspondence, I chose an email that’s just will-private at a domain I own. I don’t want to have to update this on the reg, because getting friends and family members to update their address books will be a hassle, so I want to avoid spillover from spammers brute forcing attacks on my server by using a common email format like firstname-lastname or firstinitial-lastname.

Obviously, this approach demands some level of care, lest I accidentally use my personal correspondence email address to sign up for Spammers’R’Us. But there are a handful of tools that can help you along the way.

Private Relay Emails

When you’re signing up for a new service, instead of using your one good email address, you can use a private relay email—essentially a one-off email that is created on-demand for each signup. Apple gives you the option of automating this when you’re an iCloud+ subscriber and are filling out a form on your phone, tablet, or Mac or buying something with Apple Pay. There are other services that will handle this for non-iPhone users, as well as a few manual ways to get the same effect.

In Gmail, you can create one-off addresses by adding a +keyword to your username. Gmail will deliver the email to your account, but you can filter or just see who is selling your address by tracking the +identifier. So if your email address is ptoledojones@gmail.com, and you want to sign up for an account at your local Wingstop, you’d tell Wingstop that your email ptoledojones+wingstop@gmail.com. (Please don’t email Ptoledo. I’m sure he’s a nice person). Wingstop will send you emails, Gmail will deliver them, and you’ll get some middling wings delivered to your house along with the peace of mind that the fine folks at Wingstop haven’t sold your email address. The only downside of this approach is that some services know Google’s trick, and won’t allow you to add a + to an email address.

SimpleLogin and AnonAddy provide platform-agnostic email address anonymization services similar to the Apple’s Hide My Email. Rather than hosting your email, they’ll let you create one-off aliases and then forward all of the messages that come to that address to an email address of your choice. Both services provide browser extensions to make it easy and fast to generate new aliases, but you should be aware that by using these services you’re giving them access to the emails that cross their servers. Knowing that, I’d avoid using these services for critical accounts and limit your use of them accordingly.

Fastmail, which is a fantastic dedicated replacement for your email host, has a few really clever solutions for this, but they only work if you’re using their service to host your email. Masked email lets you create as many unique address as you need—one for each account—just like Apple’s solution. If you use 1Password, you can even directly connect it to Fastmail so that when you create new logins, it will auto-generate a masked email for you. Additionally, you can setup Fastmail with a general rule like Gmail’s + modifier. With Fastmail, if you Ptoledo has the email address ptoledo@jones.com and he wants to create a Wingstop account, he can tell Wingstop his email is wingstop@ptoledo.jones.com and Fastmail will deliver it to his inbox with a filterable address of ptoledo+wingstop@jones.com attached to it. Switching to Fastmail is a bit of a process, but we’ll talk about that more in the future.

What’s Next?

This week is the launch for the newsletter, so I’m going to be listening to you! Drop me an email or hit me on Bluesky and let me know what you like here and what you’d like to see more of. After working on this for a few months, I’m excited to show it to a broader audience and can’t wait to hear what you think!

As always, if you like the newsletter, please hit the Sign Up button! I try to keep them short enough that you can read them in a few minutes, and I promise that they’ll always be full of information that’s useful if you feel like your computers aren’t working for you as much as they used to.

There was a time, not that long ago, when you’d buy a new computer and the first thing you did with it was fire something up to see how fast it was, instead of clicking “No thanks” on a bazillion offers for cloud storage and anti-virus and other subscriptions you probably don’t need.

Setting up a new PC (or Mac) these days really makes me miss the days when Windows was a piece of software that you bought every few years instead of an opportunity for enormous corporations to sell you services that are confusing and often unnecessary. Business customers still get to buy a service, if they’re big enough to support the necessary infrastructure, but for people who buy computers at Best Buy, Microcenter, or Amazon, there’s no easy way to opt out.

The good news is that you can do a handful of simple things that will make Windows work for you, instead of for Microsoft.

De-Bingify the Start Menu

One of my favorite features of modern Windows is the ability to press the Windows key to open the Start Menu and type the name of the application I want to launch. It’s quick and easy, and lets me get to the applications I need without clicking through a million menus.

That was how it worked until the newer editions of Windows 10 and 11 also began to search Bing from the Start Menu. It’s a terrible feature, instead of doing the thing I want to do—open an application—the Bing search often intervenes, jumping to a web search result for that application name instead of opening the application on my computer. If I’m trying to open Excel on my PC, I already know what it is, I already have it, and I just want to start working.

The only possible reason I can imagine that the Windows team would add this terrible feature is to dramatically increase search traffic for Bing. I imagine the conversation in Redmond that started with “We need to get our Bing search volume up somehow…” ended everyone in the meeting high-fiving each other and millions of Windows users accidentally searching Bing every day.

Sometimes a search will default to the option you actually want, but if you’re too quick on the keypress, it will send you to an unwanted Bing search instead.

It’s easy to get rid of though, if you don’t mind delving into some obscure system settings using regedit. Regedit is a tool that lets you change Windows settings that aren’t exposed in the UI, like this one. This is safe and easy, but don’t go poking around in Regedit in places you don’t understand, or you can mess up your computer.

First, open regedit, but pressing the Windows key and typing regedit and make sure you launch the application instead of the Bing search one last time. Windows will ask if you want to let regedit make changes to your device—you do. Then browse to Computer\HKEY_CURRENT_USER\SOFTWARE\Policies\Microsoft\Windows\Explorer. (You may need to create the Explorer key. To do that, right click on the Windows folder in the left pane of Regedit and select “New > Key”. Name the new key Explorer.) Right click on the Explorer folder and create a new 32-bit DWORD value called DisableSearchBoxSuggestions and set its value to 1.

Once you log out of Windows and log back in (or reboot), Bing will be excised from your Start Menu search permanently. For the record, this is the first thing I do on every new computer I buy or build.

Stop Windows From Adding Crap To Your Start Menu

The Microsoft Store—Windows’ version of an app store—likes to put shortcuts to popular apps in the Start Menu as a way to drive people into their app ecosystem. Obviously, this is the kind of fun feature that most people would turn off, if it was appropriately labeled, so Microsoft didn’t do that. If you don’t want to see those apps on your PC, stopping them is as easy as flipping a switch in the Settings app, but you have to know the right one. To find it, open Settings and search for “Suggested content”, clicking the “Show me suggested content in the Settings app” option, and toggling that option to Off in the window that opens. While you’re at it, go to Personalization > Start and uncheck the Show recommendations for tips, app promotions, and more.

Turn Off the Stuff You Never Use

Windows links services you’ll probably never use all over Windows. There are widgets that go to MSN on the lock screen, search panes that go to Bing on the Taskbar, more widgets that go to MSN on the Taskbar, recommended apps in your Start Menu, and more. It’s safe to disable all of the widgets—there are options in Personalization > Taskbar > Widgets and Personalization > Lock Screen > Get fun facts, tips, tricks and more on your lock screen. You can disable the Search pane on the Taskbar by going to Settings > Personalization > Taskbar and unchecking Search. Don’t worry, we’ll replace the default Windows search with something better in just a moment.

This widget pane is filled with crap I neither need nor want. Kiss it goodbye!

This goes for Microsoft’s push into generative AI—Copilot—as well. With the exception of the code-generation and cleanup tools sold through Github, I find it to be pretty useless for my day to day. Removing it used to be a significant hassle, but on the latest versions of Windows 11, you can simply right-click the Copilot icon in the Start menu and select “Uninstall”.

Ditch Services From Your OS Vendor

This one applies to both Mac and Windows folks. If you’re paying money for services that give you more storage for cloud files or emails, consider using services that aren’t provided by your OS vendor. Operating system companies like Apple and Microsoft tend to provide better support to their platforms, as a way to lock users in. Third-party providers are incentivized to provide top-tier support to all platforms, which means you’ll be able to more easily move to another operating system, should you choose to in the future.

• This means you probably want to use Dropbox instead of OneDrive or iCloud Drive

• It’s generally pretty easy to access different webmails on other platforms, but consider using Proton Mail or even Gmail instead of Outlook.com or iCloud Mail

A Few Bonus Tools to Make Windows Work Better

Everything is a whole PC filename search tool that provides results instantly. I use it on all of my Windows machines in place of the slow, unwieldy default Windows search. It’s worth mentioning that it doesn’t search inside text files, you’ll need to use Windows search to get that, but if you’re good about naming your files consistently, it makes it easy to find a file or folder at a quick keypress.

I love the Microsoft Phone Link app. It connects to your phone and lets you send and receive text messages from your PC. It works for both Android and iOS users (iOS users have limited support for group texts and iMessages, but it works fine for 1 to 1 texting or iMessaging). It’s preinstalled with Windows 11, so you probably already have it.

Don’t Worry About Third-Party Anti-Virus and VPNs

The tools that are integrated with Windows consistently rate near the top of the pack for anti-virus and anti-malware these days. There’s no reason to buy third-party solutions for these services, unless you have very specific needs or your IT department requires them. The same goes for VPN software. Unless you have a specific need for VPNs, there’s no reason to bother with them.

What’s Coming Next Week?

I hope these tips and applications were helpful for you! I expect this type of tip to be an ongoing conversation, so if there’s something about your devices that bugs you or if you have a favorite tip you’d like to share, please drop me a line and I’ll add it to the next roundup.

Next week, we’re going to talk about the Steam Deck, and why it’s a good example of the value of a relentless focus on the end user’s experience. Til then, thanks for reading and as always, if you find this newsletter helpful and/or informative, please consider subscribing and if you’ve already signed up, chuck me a few bucks or share the newsletter with a friend!

Thanks!

I’m always looking for both big and small ways that you can recapture your computers. One of the main reasons I love the Steam Deck is that it cuts almost all of the BS out of running a gaming computer.

(For folks who don’t know, Steam is the largest online store for PC games. It’s owned by a private company, Valve Software. They occasionally make their own games, but they also take between 10 and 30% of every dollar spent on their platform. The company famously makes more money per employee than pretty much any other company you can think of. It’s a juggernaut in games.)

The Steam Deck starts at $399 and makes playing PC games dead simple.

At a basic level, the Steam Deck is just a PC shaped like a Nintendo Switch, running Linux on a custom AMD processor with a lowish-resolution screen and a decent battery. It’s packaged with a web of software that lets people play Windows games on Linux. And the secret is in that software, which gets out of the user’s way and handles all the complicated business so that playing a game is as easy as tapping a big green Play button. SteamOS is built for controller input and handles all of the system and driver updates automatically. It also optimizes game settings so they run well on the hardware, lets you know if a game won’t run well on the Deck, maps gamepad controls, and more. It turns the notoriously fiddly experience of playing PC games into an easy, console-like experience.

Contrast that with the experience you have running a similar handheld device on Windows. We’ll start with software updates. You’ll quickly learn that games update in the store where you bought them, Windows updates in Windows update, the drivers for your hardware are in yet another app, and your vendor probably wants you to update their software yet another place. All this updating is complicated by the fact that your main interface with Windows is a small, 7- or 8-inch touchscreen, and that the Windows UI isn’t built for touch, much less on such a small, low-resolution screen. It’s a frustrating, fiddly experience and the only upside is that you can more easily play games from other stores or that use DRM or anti-cheat that blocks Linux users.

The Steam Deck’s simplicity comes at the expense of it’s single-purpose nature. While you can shut down the SteamOS user interface and swap over to a traditional Linux desktop to run a browser or productivity software, the hardware really isn’t designed for that. Its screen is too low resolution for work and the machine just sports a single USB-c port, so connecting an external monitor or other peripherals requires an external dock. (The low-resolution screen stinks for working, but it’s ideal for playing games on the go. Fewer pixels means the GPU in the machine uses less power, which extends the device’s battery life.)

The Steam Deck isn’t a new idea. There have been low-volume handhelds running similar hardware out of China for more than a decade, and more mainstream fast-follows from Asus, Lenovo, Acer, MSI and more since the Steam Deck launched. Even though Valve’s a private company and doesn’t really report Steam Deck sales, we do know that none of those other handhelds have penetrated the market like the Steam Deck. The closest we have to real numbers is a vague “multiple millions” number from November 2023, and we’ve seen a minor hardware refresh since then.

How does that stack up to other handhelds from laptop juggernauts like Asus and Lenovo? According to my industry sources, “multiple millions” of Steam Decks sold likely means that the Steam Deck is outselling every other machine in the nascent category combined.

Why is the Steam Deck successful? Price helps—Valve is shipping lower end hardware and isn’t paying Microsoft to pre-install Windows. I’d argue that Valve’s imprimatur has an even bigger impact. The Steam Deck is on the Steam homepage every single day, where it’s seen by millions of potential customers. It’s almost always near the top of the Steam top sales charts (presumably based on dollars sold, not units shipped). But I’d argue that the extreme focus on usability—getting people into games with as little hassle as possible—is an even bigger part of their success.

But Valve hasn’t been able to harness their promotional power to push hardware in the past. After all, this isn’t Valve’s first attempt at hardware. It’s shipped everything from game controllers and streaming boxes to VR headsets and even whole gaming PCs, and none of them saw the same level of success as the Deck.

A big argument in favor of usability as the key is that users are actively working to create versions of SteamOS that run on more generalized hardware than the Deck itself. Because Valve’s work on SteamOS is all open source, it’s a matter of adaptation rather than reverse engineering to add support for other hardware to the platform. This is showing up in Linux distros like Bazzite, which runs on general purpose PCs as well as AMD-powered handhelds from other vendors. Where SteamOS only includes drivers for the hardware inside the Steam Deck, Bazzite supports GPUs from other vendors and also bundles in tools that make it easier to play your games from other PC games stores—Epic, Battle.net, GOG, and even indie stores like Itch.io and Humble.

Bazzite has an active dev community bringing support for the SteamOS front-end and feature set to all PCs and handhelds.

And of course, Valve announced they’re making SteamOS available to other hardware vendors too. They have committed to shipping a version of SteamOS that runs on more general hardware later this year and the other hardware vendors are already onboard. Lenovo is shipping SteamOS hardware later this year.

Lessons to Learn

SteamOS feels good and different in today’s otherwise enshittified ecosystems because Valve’s goals and our goals as users happen to align perfectly. I want to play games and Valve wants to sell more more games. It’s in their best interest to make playing PC games more accessible—in all senses of the word. At a starting price of $400, the Steam Deck is cheap compared most starter gaming PCs, but the experience for players is quite good. Its low-resolution screen and hardware configuration align favorably with the current console generation, which means developers were already building games that could comfortably run on that hardware. There isn’t a ton of required work to get most games that run well on the Xbox or PS5 to also run well on the Steam Deck. At the same time, the Steam Deck also condenses much of the backend work of running a computer into one bi-weekly update that happens automatically when you reboot the machine. And Valve’s rewarded for this hard work with a hefty cut of every sale that’s made on Steam.

What are the lessons to learn here?

Set a goal and be relentless about focusing on that goal. The goal Valve had with the Deck was to expand and extend the PC gaming market. They designed a cheap machine, built software that made it the single simplest place to play PC games, and marketed it to their core audience. They were able to do this because it’s a small company and there aren’t 50 different stakeholders diluting the impact of every product they ship.

Microsoft could have made this machine and it would only play Xbox games and use their cloud streaming service, but it would also have apps for video streaming services (that sucked compared to the ones on your phone and TV), ads that get in the way of you actually playing games, and you’d probably need to subscribe to GamePass to make the machine usable. If Google made it, they’d use some framework an engineer was hyped about and pass the cost of porting games to it to developers, which would prevent anyone from ever releasing software for it. If Netflix built it, they’d spend a ridiculous amount of money on new games, and kill 95% of them all at the second milestone because someone in the decision tree “didn’t get it”.

Minimize work to maintain the machine. This is the place Windows fails the hardest. The open nature of the Windows platform means that each vendor—the PC builder, the chip and peripheral providers, the game vendor, and Microsoft itself—wants to own their own interface for providing updates to their drivers and software so they can market other products to you. Unfortunately, those applications are often really poorly made—either they’re slow and unwieldy, are confusing or difficult to use, or just plain suck. But even if they didn’t suck, having to use multiple applications to update the software your computer needs to run is a bad solution for everyone.

Advertising the limits of your product is a feature. There’s no ambiguity on the Steam Deck support page that’s on every single game. If a game passes their certification tests, it gets a green badge and is labeled Supported. If it won’t run, it gets an Unsupported badge. And if there are a handful of problems, the page lists it as Mostly Supported and gives an explicit list of the problems encountered in testing. There’s no attempt to trick a customer into buying something that won’t give them a good experience, because if they did, the customer will just refund the purchase.

The Steam Deck Compatibility link is on every game sold on Steam. It either says it works well, doesn’t work at all, or lists exactly what doesn’t work, for games that are close to passing.

What’s Next?

That’s it for this week. Next week, we’re going to talk about why email is broken, why none of the big companies are trying to fix it, and how you can make email more tolerable, because at this point, we’re stuck with it. If you enjoyed this newsletter, please consider subscribing or chucking me a few bucks, and share it with a friend!

I have to start today by apologizing. In last week’s newsletter, I dropped some inside baseball jargon without adequately explaining it. So today, I’m going to talk about dark patterns, cognitive load, and why these things matter in the modern context of phones and computers.

Cognitive load is an idea that came out of educational research in the 80s and it’s applied to a great deal of modern game and user interface design. It’s the idea that solving different types problems takes varying amounts of mental effort, that everyone has an upper limit for their capacity to solve problems, and that each person has a varying capacity to understand, process, and tackle those problems. Put simply, you can think of cognitive load as your brain’s budget for figuring stuff out.

Different types of problems add different amounts of load to your cognitive budget. Problems you understand and know how to solve create less load, while figuring out novel problems that you’ve never experienced before creates more load. And of course, everyone’s different. Some people are better or worse at different types of problems than others, and people’s respective ability to handle varied cognitive loads varies accordingly.

Dealing with a challenging cognitive load can be extremely satisfying. This is one of the main reasons that just playing some games feels really good—the act of perfectly dropping a block in Tetris, nailing a tricky jump in Mario, or finding a clever guess for Wordle makes your brain feel good.

It isn’t a universal truth, but games that are successful deliver an ideal cognitive load for a large audience of players. Wordle is a classic example—it provides context clues that signal it’s a challenging kind of word game, but when you actually do the math, Wordle is a fairly easy game to win if you make good guesses. In order to lose, you have to either have made bad or extremely unlucky guesses. (Apologies if you’re bad at Wordle.)

So if riding a high cognitive load feels good, what happens if you get overloaded, cognitively speaking? This probably won’t surprise you, but it feels bad. When you’re experiencing cognitive overload, you feel flustered, frustrated, or even upset. At an extreme, you may feel information paralysis or anger. Even riding a high cognitive load for an extended period of time can be stressful, which causes most people to make worse decisions.

Dark Patterns and You

Which brings us to dark patterns. Dark patterns are deceitful design choices that cause negative experiences for users and positive experiences for the developers. Dark patterns are used to mislead users, often into doing things that they wouldn’t otherwise do.

When you use software that utilizes malicious dark patterns, that experience often leaves you feeling bad. In games, they often create addictive loops in players that result in players feeling worse the more they play, instead of feeling better when they play. In apps, they increase feelings of frustration and often leave you feeling like you’re wasting your time.

Modern computers, phones, games, and websites expose you to so many dark patterns that you probably don’t even realize that you’re seeing them—apps that overwhelm you with information or notifications, software that’s difficult to navigate successfully, weaponized FOMO in apps and games, monthly or annual fees that create a feeling of obligation and sunk cost, relying on psychological tricks to make users feel inadequate, and more. I could probably do dozens of posts about dark patterns in app notifications, autogenerated emails, website designs, and more. For now, just trust me that most modern computers are, at best, a mess.

It’s often difficult to differentiate between poorly designed software and software that uses dark patterns maliciously. I say it doesn’t matter. If your computers are creating more work for you instead of less work, the end result is the same.

And big companies, like Microsoft, Apple, Google, Facebook, Netflix, and Amazon, are neither adequately protecting and warning users about these malicious apps, they’ve structured their businesses around services that are designed to capture an ever-growing traunch of cash from users.

So if you’ve ever thought that you used to enjoy using computers and spending time on the Internet, but that feeling has faded in recent years, it’s not just you. Computers are getting actively worse. The rise of dark patterns means that using computers today actually feels worse than it did just a decade ago. Recognizing and adapting to those dark patterns adds to the cognitive load you feel every time you use your computer.

What’s Next?

Which brings us to what’s next. If you’ve read this far, you’re probably like me— you’ve sensed that something isn’t quite right with your day-to-day computing. As computers became a de riguer part of modern life, our tech-free decompression time evaporated. Prior to the smartphone, most people used a computer while they were in the office working and then spent time in the real world on evenings and weekends.

That’s gone now. We’re all extremely online, all the time these days. And even if you’re feeling the load, most people aren’t yet. But I’m ready to figure out what’s next. It was either that or move to a cabin in the woods and sign off the Internet for good.

If you’re interested in carving some of your time back from your phone, I highly recommend the method recommended in this video. My screen time is down so much since starting this that I assume there’s an alarm going off in Apple’s mothership.

Next up for me, is going to be talking to people who are making the software and hardware, highlighting both the little projects and products that make computing better as well as focusing on the big swings folks that folks are taking to try and tackle these difficult problems that we’ve created for ourselves.

As always, I hope you’ll come along with me. If this is your first time here, I’d encourage you to hit the Sign Up button and get a new letter in your inbox every week. If you’re already coming by every week, please talk about it with your friends and co-workers! I don’t have a marketing budget, so I need your help to spread the message.

I really appreciate all the feedback, so thanks, and I’ll see you next week!

I’m sitting here in January, looking at this year’s haul from CES and it’s the same parade of new products adorned with the same adjectives I’ve seen for the last 20 years—faster, bigger, smaller, thinner, lighter, shinier, smarter, better. The numbers always get better and the bars always get bigger, but nothing I’ve seen out of CES is going to solve the actual problem we’re all facing: time.

The modern computer has become a time vampire. I’m not just talking about infinitely scrolling videos on TikTok or endless games of Fortnite. I’m talking about almost every task that’s vital to modern life, from checking email to paying bills to getting emergency information to finding a doctor to buying a pair of shoes. Turning our attention and time into money has made trillions of dollars for a handful of corporations over the last 20 years, but as user growth has slowed the only way to keep revenues growing (and shareholders happy) is to steal an ever-increasing amount of our time.

For most of the last twenty years, being good at computers (and later smartphones) was a kind of superpower. You had access to secret information that offline folks didn’t know. Everything from travel to home goods to takeout was cheaper or easier than it was for people who still used their phones to talk to actual humans. We’ve all gotten addicted to skipping the line.

I read constantly and love saving a buck, so at the dawn of the Amazon Age, I loved buying books online. The books I bought were the same, whether I clicked a link in a browser or drove across town to Barnes & Noble, but saving 40 minutes in the car and a few bucks on the price tag made me feel like I was getting away with something.

Tasks that are more complicated or have more dependencies in a traditional context are even more complicated online. Buying groceries online makes meal planning infinitely more complex and has, at least in my house, dramatically increased the amount of food waste we create. An out-of-stock ingredient is much easier to adjust for if you’re pushing your cart through a brick-and-mortar grocery—you see that the ingredient is missing and adjust your meal plan on-the-fly. When you’re shopping online, you may not find out that your ingredient was unavailable until your groceries arrive, and then you’ve already purchased the other stuff you needed to make the meal. At that point, you have to either run into your local grocery anyway, make a clutch adjustment, or give up and grab takeout.

It turns out, Amazon making book buying easier than the brick and mortar experience was a bad signal. Most of the things we use computers for are more complicated than buying a book. They’re even more complicated than buying groceries. And we’ve reached the point where it’s not in any of the big technology companies best interests to make those tasks easier.

We have to realize that the time for skipping lines has passed.

So what’s next? We’re overdue for another huge shift in computing, and I don’t think this one is going to come from any of the existing technology companies. They’re too big, too ossified, and too entrenched in their existing revenue streams. All of their business models are locked into the attention economy, which is what’s pushing users everywhere over the edge today.

I don’t think the next revolution will be powered by AI. If anything, current generative models are accelerators that are just making the problem worse. Cheap, VC-subsidized content creation is just increase the cognitive load we all feel every time we use the Internet. Is that picture real? Is that AI summary accurate or the endpoint in a Markov chain of lies? Is that really my mom calling me or just another hyper-targeted scam?

I don’t think there’s a monolithic solution to this problem—I think we’re going to see lots of people tackling individual components—information, shopping, communication, etc. They’ll find ways to save people’s time and reduce their cognitive load. And those solutions will spread, just like every good idea has spread on the Internet for the last 40 years: folks will come up with novel solutions and apply them to their areas of expertise, then the solutions will spread to other teams working on similar problems in other areas.

And when they build it, people will come.

I’ve lived through the birth of the Internet and wrote about a bunch of it, from the early pre-commercial days to the modern hyper-monetized smartphone era. I know that there’s a better way, and I’m ready for what’s next. Are you?

If you enjoy this, please consider subscribing, forward it to a friend, and/or chuck a few bucks my way. Thanks!