Ask Paul: October 17 ⭐

Happy Friday! Let’s kick the weekend off a bit early and try to forget or ignore that AI is eating everything we care about.

?️ Shuffling the deck chairs?

SeattleMike asks:

What are your thoughts on the recent (massive) re-org within Microsoft and its impact on the future of Windows? Any feedback from employees? How’s it going over internally?

At the end of September, Microsoft moved all of Windows engineering under a single organization run by Pavan Davuluri, who had recently been promoted from corporate vice president to president. Davuluri came up out of Surface, though I had never heard of him until he took over Windows + Devices in 2023 after Panos Panay left the company. He seems like a better choice to run this business than Panay, but he has also weathered some mini-controversies, most notably the botched Recall launch in mid-2024.

To me, there are two things going on with the reorg at a high level. The most obvious is that Davuluri is bringing Windows Server back into the same organization as Windows client. And the second is that Davuluri clearly has two qualities his predecessor lacked: Technical skill and a real vision for the future of Windows.

When we first learned about the reorg, we knew that it was about remaking Windows with agentic AI capabilities, but there was also a misunderstanding about the scope of the reorg. That is, while “Windows engineering work” was unified under Davuluri, the “core kernel,” virtualization, and Linux remain in the Azure organization. So Davuluri did not take control of the low-level, foundational parts of Windows as I had thought.

No one here should be excited or care in the slightest that Windows Client and Server are or are not under one roof. But Davuluri does oversee the Core OS, Data Intelligence and Fundamentals, Security, and Engineering Systems teams now, rather than sharing much of that with Azure. And that is important because the shift to “Windows as an Agentic OS” will require a lot more engineering/computer science work than painting a thin veneer of a new look and feel on top of Windows, as Panay did with Windows 11. This is real work, not superficial aesthetics.

Tied to this, Yusuf Mehdi, who is an executive vice president now and consumer chief marketing officer, has been making the rounds doing press briefings and interviews to help sell this effort. I mention the marketing bit because he has routinely repeated the claim that Microsoft “rewrote” Windows “from the ground up” when it created Copilot+ PC and that it’s doing that again as it transitions Windows to an agentic OS. That claim is ludicrous because nothing of the sort happened. But it’s fair to say that there is now a lot of deep technical work happening, the type of thing that wasn’t the focus under Panay, and that that’s what Mehdi is oversimplifying. Between the Windows Resiliency Initiative and now these platform-level AI capabilities that go far beyond apps like Copilot and Recall, we’re seeing the type of foundational churn in Windows we’ve not seen in 15 years or so.

Pavuluri has pretty much telegraphed this work, most recently with that video in which he described how AI would impact Windows from August. And if you look at Microsoft’s public microsite for this past week’s announcements, you can see how this first milestone on the path to a truly agentic OS is unfolding: There is a vision (every PC will be an AI PC), there are new low-level capabilities, most obviously through the addition of agentic capabilities, and there are new end user experiences that will roll out over time and into next year.

I’ve not heard directly from any employees about this, but I don’t think the reorg changes much day-to-day. I also don’t think that this effort changes the fundamental relationship between Windows and the rest of Microsoft. Like any other business there, Windows has to justify its existence financially, which is easy, and it has to conform to the overall corporate strategy, which is all about AI. So this is Davuluri making that happen, and though some will quibble over specifics, like whether we will or will not talk to our PCs, this is a much smoother fit for Windows than was the case when the overall corporate strategy was about the cloud. That was the underlying cause of much of the enshittification in Windows today, and contorting Windows to make sense in that world just didn’t make sense.

In short, Davuluri is trying to position Windows for a quickly happening future, and that’s true both internally and externally. He seems like the right guy for that. And while Windows PCs will always be somewhat anachronistic in a personal computing market that shifted to mobile devices years ago and is not plotting a way forward in a sort of ambient computing model, it’s important to do what one can to keep Windows viable. We are all going to be talking with AI everywhere. Windows can’t be left behind as that happens.

✨ AI, the eater of worlds

wright_is asks:

On Windows Weekly, you and Richard were talking about AI replacing applications going forward. Richard’s example was the ERP system. But I just don’t see that happening, at least not in the near future. I can see AI augmenting parts of these systems, but replacing the whole front end? The SQL database? I just don’t see it.

I know. That’s why I wrote AI is the End of Apps ⭐ even though it was an incomplete thought with many related threads left hanging. It’s a big conversation and getting it all out would require a book-length work. I’m trying to rein things in a bit where possible.

Fundamentally, what I’m talking about is how individuals interact with personal technology, not what businesses are doing. But that doesn’t mean this discussion can’t touch on businesses. Many of those individuals will be getting things done for an employer in the context of their jobs. But when I say “apps,” I am referring to apps as we know them today, executables on computers and devices (which are really computers).

Today, we basically deal with apps and services, the latter of which require an app or some UI to interact with. I haven’t found a concise term that Microsoft or anyone else uses to describe this change, but there is a new evolution happening in app-to-app communications that I think of as apps becoming programmatic. You can see early versions of this everywhere, but a few examples will suffice. In the 1990s, Microsoft created what became COM and the central idea there was that code running on a server or PC would expose public interfaces for internal functions that other apps and services could use. And in the 2010s, the Windows team came up with contracts as a way for Windows (mobile/modern) apps to identify which capabilities they supported to the system and thus to other apps.

In Windows 11 today, we’re starting to see App Actions as a similar means by which apps can surface individual features that can be used outside the app by the system. That is, instead of “Open with …”, “Edit in Paint,” or whatever, you will see options like “Blur background with Photos” and “Remove background with Paint” in the Explorer right-click menu for an image file. The next step is more of a “just do it” thing in which we’re just removing or blurring the background and we don’t care which app does this and don’t need that app to even open visually.

This kind of work also has roots in the document-centric UI in Windows 95 that everyone forgets and the people-centric UIs in Windows Phone that everyone misses. By which I mean, in a traditional personal computing scenario, whether it’s a mid-1990s desktop PC or a modern iPhone or Android smartphone, you as the user wants to view or edit and image and the process involves remembering the name of the app you want to use, finding a way to launch that app, and then knowing how to use the app when it appears. That’s a lot of stuff to expect users to know and do.

As we move forward into agentic capabilities and these programmable app functions (again, for lack of a better term), the shift is that users don’t have to know so much. They just ask or tell the system to do it. And I know this is a line in the sand for a lot of tech experts, power users, and enthusiasts and for the same reasons as always. We’re change averse, this threatens our expertise and standing in this world, and we sometimes forget that the way we do things isn’t “normal,” meaning it’s not how the mainstream majority audience does things. So we dig in our heels.

But we all have examples from our own lives in which we have experienced why this shift will benefit us too. In my case, I have two examples that come up again and again. They’re very specific, but maybe that’s the point. They are:

• Every January, I write a PC sales overview post from the previous year (here’s the most recent one). This post has a chart as a graphic at the top that I made with Excel. And the way I do that is, every January I go back to my archives from the previous January, find the Excel spreadsheet I used then, make a copy of it, and then add the data from the most recent year. But then I have to redo the chart to include that data, and I always forget how because I don’t use Excel most of the time. So I sometimes can’t do it right, so you get things like my January 2024 post where all the bars in the chart are the same color. My argument here is, I’m a writer. I want this chart, and for it to look right. And I am not going to master Excel when I only use it once a year. But what I can do is just tell AI to make this chart. And that works really well.

• The logo I made for Eternal Spring, originally in Photoshop but then redone in Affinity Photo in more recent years, involves a font that I customized and, worse, a graphic/photo that appears inside each letter while leaving the area surrounding the letters transparent. There is a term for this effect, but I’m not a graphic artist and I couldn’t care less what it is. But there is also a process for making this work, and that is something I forget how to do, as with the Excel chart above. And so each time I need to make a new version of this logo, I have to relearn how to do it. Here again, AI could “just do it” for me. My goal isn’t to master the tool, again. It’s to get it done and move on with life.

Just as there will always be people who “need” to sign in to Windows with a local account, yes, there will always be people who need to open Excel to do whatever work or use Photoshop or Affinity Photos for graphics work. But there are many more people who do not need that, and AI will let them get results using functions and features from those and other apps and services without having to take the time to master them. In trying to speak and write plainly, I often run afoul of the overly pedantic. So when I say that no one should use a local account, I pretty much mean it literally but I (literally) mean it statistically. There are always exceptions. There always will be. But explaining that every single time a discussion like that comes up is tedious. We should all be on the same page with these things.

AI is moving so quickly right now—and this is rather incredible, given that we’re 2.5 years into this transition already—that it’s often difficult to see where things are going. But I can look at individual steps and see familiar patterns. After this week’s Windows agentic AI announcements, I got an almost unhinged text message from a friend in the business who claimed that “no one will ever talk to a PC.” And this is a weird one for me because I don’t use AI all that much, honestly, and I don’t see myself speaking to PCs either, at least for now. But this is what experts of a different era said about GUIs. And then about the Internet. And multitouch. And smart pens. And it kind of just keeps repeating. The notion that personal computing, or maybe just PCs, reached some moment of perfection and should never be evolved again is outrageous to me. The PC has to change. Otherwise, it will go away. Because like it or not, the world is moving on. It will do without us.

This is a struggle. When I write things like The iPad May Be the Perfect Computer (Premium) or AI is the End of Apps ⭐, they may appear to be deliberately provocative or whatever, but that’s rarely my intent. In many ways, these things are me working through my own change-averse nature and trying to step outside my own little world. Maybe some of it is tied to all the reviews I’ve written over the past 25+ years, where it’s important to try to understand how real people will use these things, and not just how I may do so. I don’t know. But it is something I have to work at sometimes.

Things don’t always pan out. When Microsoft was doing Windows 8/RT and tablets and full-screen mobile apps, I thought it was nuts, but I tried to make it work. And what’s sad is that there were some good ideas in there, not to mention some terrific advances on the Desktop side that were mostly ignored because of the community-wide outrage over the big problems. We got what Windows 8/RT could and should have been in Windows 10, but that was three to five years too late and the moment for cross-device convergence had disappeared.

Whether AI is a bubble or not is, to me, beside the point. There was an Internet or Dot Com bubble in the early 2000s that everyone remembers well, but that stuff is going stronger than ever and it never left. This isn’t like the videogame crash of 1983, where everything just stopped for three to four years (in the U.S. at least). AI isn’t going away. This is now and it’s the future. It’s happening. More importantly, it’s changing everything. And that kind of brings us full circle. We’re all experts of whatever kind. This threatens that.

Anyway. This in no way is a response to what you wrote, I guess, but it’s a sort of blurting out of some of the big things that are happening now and how I’m trying to work through it myself. But let me try to at least address some of your points.

A traditional application is much more streamlined, quicker and more efficient at these sorts of tasks. There is no efficiency gain moving those tasks to AI, it will just cost more money – you need bigger servers that use more energy to do the same task.

This assumes some expertise on the part of the user. In the sense that we will always have Photoshop experts, we will always have ERP (or whatever) experts too. And they may prefer or even need a traditional app. But they’re also the minority. There are bosses, and others up and down the corporate structure, who maybe today rely on that expert to deliver summaries to them in whatever forms. Energy concerns aside—I feel like that’s a solvable problem—I have a difficult time understanding how AI wouldn’t streamline that.

This argument, to me, currently, has a lot in common with the blockchain hype of the 2010s. Everything was going to switch to using the blockchain, we were going to replace our SQL databases with blockchain, blockchain was going to be in every piece of software, regardless of whether it made sense or not. Then, it turned out that blockchain was slower and less efficient than existing systems and, apart from some specialist areas, where transaction accountability was important, it just wasn’t economical to use blockchain – heck, blockchain isn’t even really economical for crypto currency, it often costs more to process the transaction than what the transaction actually purchased.

Removing people from whatever processes will always be more efficient and cost effective. This assumes a few things, the key one being that the AI or whatever that is replacing those people actually works. But that, too, is a solvable problem. People are not a solvable problem. We will always be the weak link. Even with something as non-work-related as a referee or umpire making calls in a game. People make mistakes all the time.

I have the feeling the AI bubble will burst, at least in this current “AI will replace everything” form.

To me, this is more about AI will be a part of everything and most people will use AI in whatever forms to get things done. This is the speak/write plainly thing, meaning there are always exceptions. But those exceptions are a small audience comparatively. Anyone who creates art, writes novels, makes videogames, or whatever and believes that AI won’t “replace” them is nuts. This doesn’t mean literally that these people go away. It means that AI will democratize those tasks and make them more accessible to a much wider audience. We will always need experts. But most people want to get some task done and move on. And those experts will use AI in other ways, outside of their expertise, for the same reasons.

Using the AI where it can be a genuine help and the increased processing costs still mean a time and financial saving over having work done manually makes more sense than replacing cheap, accurate systems with incredibly complex systems that can’t be properly tested for accuracy and cost much more to run.

I do think most of our current systems will be replaced, but these things can’t happen overnight. Boston used metal coins in its subway system until the mid-1990s, even though most other metros had moved on to computerized systems involving paper cards with magnetic stripes. Then we had apps. Then tap to pay. And now you don’t even need to sign up or join some system, you can just show up in a new city, tap your phone (or watch) on the way through a turnstile and it just works. So that original system was replaced. But it took decades. Perhaps in the future it will just do a retinal scan. It will change again, that’s for sure.

?️ Big Tech, Little Tech, no tech

spacecamel asks:

when you mentioned data sovereignty on WW this week, it made me think about your De-Google project. Any updates? What is your next step? I am personally using an old computer to build a OpenMediaVault to store a copy of my work and get the important things off of the cloud. I might be able to get rid of some of the subscriptions too.

When I first read this, I thought back to some efforts, about a decade ago, to see where and how it might make sense to de-Google a phone or whatever. But the more modern take on this is what I’m working through in the Online Accounts series, of course. And it’s clear to me that this is an ongoing thing. Meaning, I can’t flip a switch and be free of Big Tech. But I can keep evaluating my options and see where it makes sense to minimize my exposure to Big Tech.

And we need to be pragmatic here. I’m not going to stop using, say, Gmail because it just works and I haven’t noticed any negative impact from enshittification or whatever. I feel the same about Google Photos, Search, Maps, and YouTube and YouTube Music. But I am also open to replacing any of that if something better comes along. And these days, increasingly something better (to me) arrives in the form of what I call a Little Tech solution. Proton Authenticator instead of Google Authenticator. Typora (really, Markdown) instead of Microsoft Word. Synology Drive instead of Google Drive or OneDrive. Affinity Photo instead of Adobe Photoshop CC or Elements. And so on.

I always say something like, “it’s not like I’m going to move to a cabin in the woods.” But I do have those moments where I could almost see that. This is partially an age thing, I guess. But we all reach our limits at some point. I’ve long thought that teenagers are terrible in some deep level biology/evolution sense because that gives them and their parents the impetus they both need to let go and send them on their way as they move into adulthood. I’m starting to think the same of retirement, or when people reach a certain age. They’ve seen so much, they can’t deal with whatever new change is happening, they have a negative opinion of younger, more energetic people, and then it’s time to say goodbye. And this is a natural process perhaps.

I definitely had moments when I hit the proverbial wall. The original release of Windows Azure was like that. So was quantum computing. These were just bridges too far for my brain at the time. Today, we’re wrestling with AI per the above questions that I never really answered. But this one, at least, I feel like I can handle it. Whether I handle it with Google, Microsoft, Apple, or some smaller companies is unclear. We’re in the middle of it.

Regardless, moving away from Big Tech, or at least minimizing it where it makes sense, is both subjective to the individual and not a hard stop. It will happen over time if that’s where your head is at. There will be steps forward and steps back. And we should recognize that regulation and shifting consumer sentiments can drive change, too, and reverse some of the enshittification that triggered these thoughts in the first place. Big Tech will never do the right thing for individuals and customers unless they’re forced to. That can come from different places. It won’t come from within those companies, though, not anymore.

There is no end game here. We will all take whatever steps, or not. And we will all use whatever mix of platforms, apps, and services, to get work done, to be entertained, and whatever else. I’m not a privacy nut, but many people blissfully give up their privacy to use whatever Big Tech services, and I do think those people are nuts. But it’s a matter of degrees, and I do it too. I use Google Maps knowing that I’m feeding the beast, for example. And I’m OK with that because for whatever reason, the trade-off there feels fair to me. But I also notice that Google’s redirections while en route are starting to feel like they’re less about getting me to where I’m going as quickly as possible and more about managing traffic for the entire user base. And while that’s not necessarily enshittification, I would like this thing to selfishly work for me, not other people. And the mind starts to drift.

➕ Naming is hard

spacecamel asks:

Do you think Apple changed the name of their streaming service because they are going to change the name of the box when they release the new one later this year?

I don’t personally think they should have rebranded the service, and the ADHD part of my brain doesn’t like the inconsistency, since Apple has other paid services like iCloud+ and News+ that will continue using that naming. But I do accept the explanation, which is that most customers thought of the service as Apple TV, and those in the company referred to it that way as well. I almost respect that this happened without any market study or whatever. And the devices do have individual names like Apple TV 4K or whatever. We will survive this. 🙂

? Aced it

anderb asks:

Thoughts on Ace?

For those unclear on this, the original lead guitarist of Kiss, Ace Frehley, passed away yesterday.

I have a complicated history with Kiss. In the band’s initial heyday, during which Frehley was its lead guitarist, I was young enough that I was aware of some of the music (“Rock and Roll All Nite,” of course, “Detroit Rock City,” and so on), but they didn’t really impact me until later. I had “I Was Made for Lovin’ You” on a K-Tel album, but I didn’t really get into Kiss until the mid-1980s, coincidental to the band’s no makeup era. I have seen Kiss live many times starting at that time and through early 2020, when they were the last concert we saw, and it was incredible, before the pandemic.

Part of the issue for me with Kiss was that Van Halen happened, and that band made the original Kiss lineup seem a lot less talented by comparison. And that’s particularly true of Frehley, who never reached the pinnacle of guitar god virtuoso that, say, Eddie Van Halen did. And he wasn’t particularly well-spoken, which didn’t help matters. Obviously, he has many fans, and he is iconic for all the right reasons. But I very much preferred the 80s and early 90s Kiss lineups, and the music they made (which I know will be divisive for many Kiss fans).

So I don’t know. It’s telling to me that he didn’t play much (or at all) on the two final albums he made with Kiss before he left the band in 1982. And that he likewise had almost nothing to do with the album Kiss toured on when they reunited in the late 1990s. I guess he just never resonated with me.