Ask Paul: August 22 (Premium)

Happy Friday! We’re moving from the slowness of summer into a far busier period with tech events, travel, and whatever else. But first, a pause for another great selection of reader questions. And then it’s back to the grindstone.

❤️‍? Healthy gadgets

Red09 asks:

Hi Paul, now that you are using the Samsung Galaxy phones a lot more and that you were also, if I recall, a user of the apple smart watch for health monitoring, would you look into trying the Galaxy Watch at some point? I would be interested in your take on it. I’ve been using the older galaxy watch 4 classic which does give me some insights into certain aspects of my health and sleep and was thinking of updating to the Watch 8 with the extra sensors and monitoring it has, but not sure if it’s worth it. I always wonder how much a watch can realistically monitor and whether it’s just marketing not to mention how reliable it would be. Wondered if you had any thoughts based on your previous experience with the health monitoring you did in the past.

I don’t have any plans to test a Samsung Galaxy Watch at the moment, but I do move between an Apple Watch and a Pixel Watch 3 when I switch between the iPhone and whichever Android phone I’m currently using. And while I like the Apple Watch more overall, both platforms–watchOS and Wear OS–are roughly the same functionally, at least in how I use this kind of thing. But they’re also different in curious ways.

As I wrote in Full Screen (Premium), sometimes you don’t realize how much you use or like something until it’s gone. In that case, it was the system clock in the Taskbar, which you can’t see if you auto-hide the Taskbar. So I use a third-party utility (that I wish was still actively supported to some degree, as it recently disappeared from winget) to display the time/date above any windows, and problem solved.

With a smart watch, obviously the time is important, but it’s also one of those things I miss when it’s gone. I will typically charge whichever watch when I shave and shower, but in the time I’m in the bathroom without the watch, I often wish I could see the time for whatever reason. And I just coincidentally commented about that to my wife. I’d like to get some kind of cheap time display, even if it’s just something you stick to the mirror or whatever.

But a smart watch is a bit like the lock screen on the phone (or, back in the day, the live tiles in Windows Phone) in that it can display these other things that may be useful to see at a glance. And this is one minor way in which the two platforms differ, though I always find it a bit off for a while after switching. Depending on the watch face, I can display things like the device’s battery life, the AQI (air quality), the weather/temperature, and whatever else. And I do like having at least some of that there, though what I can put there depends on the watch and the watch face. And so I twiddle with this constantly, no matter which device it is.

I like that Apple Watch has colored backgrounds, whereas WearOS (at least on Pixel) always uses a black background, which is really no background, which I suspect is about preserving battery life. But Apple Watch switches to that type of display when you’re not looking at it. (And either could just be configured to not always be on, which would help battery life even more, I suspect.)

It may not seem like it, but that’s a curiously big difference. The others are the shape of the screen—round with WearOS and rectangular with Apple Watch–and the way they handle health tracking. Where WearOS is a Fitbit device (on Pixel), Apple Watch obviously does whatever Apple Health things. And so its focus is on those three rings, which are Move, Exercise, or Stand, whereas WearOS/Fitbit is more about steps or whatever metric. I mostly prefer the Apple approach, but I also like having an at-a-glance heart rate display, and that’s where WearOS/Fitbit is better. The sleep tracking on Fitbit is ridiculous and I hate that you can’t turn it off.

I don’t have a recent Samsung Galaxy Watch, but my sister and brother-in-law do, and we talk about this sometimes, mostly because I’m curious. The most recent models have a squircle design that I really like, though the screen is still round, of course, and some think it looks weird. The unknown to me here is the health stuff. I assume these devices tie into Samsung Health and not Fitbit (though I assume that’s at least possible, too), and that is slightly less interesting to me. Not that I have any particular affinity to Apple Health or Fitbit, but just out of having to learn yet another way to do things.

But this is also where smart watches are perhaps more functional than I need as well. I mention all the stuff above because these are the things I actually “do” with the watches, and it’s not much. Look at the time. Look at whatever at-a-glance info is there. Use it to track whatever health metrics over time, but rarely look or think about any of that beyond the hourly reminders to stand (Apple) or move (WearOS). Charge it, every single day, which is a hassle. And that’s kind of most of it. From a health tracking perspective, I pretty much just care about resting heart rate, blood oxygen (which only just returned to the Apple Watch), sleep (but only on Apple because Fitbit is terrible), and steps. But most of this isn’t a daily thing. I just check in occasionally.

This makes me wish for something simpler. Something that would last longer on battery. And something that worked on both platforms. And maybe I’ll get there someday.

Regarding Galaxy Watch and Samsung Health, I’m not against any of that. I wouldn’t get one now, not because there’s anything wrong there or whatever, but because I already have a Pixel Watch 3 and they’re very similar, functionally. And if/when I upgrade either watch I do have, it will be another round of rethinking what it is I’m doing with these things, and I’ll see what’s out there. Ultimately, these smart watches are a lot like smartphones, where there is no perfect device (for me). There are multiple choices, each with their own strengths and weaknesses (to me). None of them are exactly what I want. Each of them is fine.

Reliability and accuracy are always questionable, but whether that matters will depend on what you’re doing. If you’re actively tracking whatever–steps, perhaps, or some vaguer overall fitness score–inaccuracy isn’t a huge issue if all you want to see is progress. For me, it’s not so much about a goal as it is to just make sure that certain metrics are in the right range over time. So it’s not a daily activity for me, per se. But if you’re actively engaged, a newer device of whatever kind will of course have more and better sensors, more features, and so on. This is either critical or just too much data based on your needs/wants. I could get by with a lot less, I think.

This is probably not very helpful, sorry. I haven’t really figured this one out for myself.

? OneDrive feeds on your soul

AnOldAmigaUser asks:

When Microsoft started pushing the idea of “backing up” one’s computer to OneDrive, I remember it being mentioned that they wanted the data to feed their AI.

Yes, but let’s be careful about what that means.

This behavior came to light in 2023, ahead of the release of Windows 11 version 23H2, which was the first Windows 11 feature update to be delivered as an enablement package. This let Microsoft deliver its massive set of new features, including Copilot, ahead of 23H2, to those on Windows 11 version 22H2, preventing businesses from skipping that new release and not getting Copilot and whatever else. That October, I wrote about why Microsoft had made this change to the release schedule, and that on the consumer side, it was about getting data in OneDrive tied into Copilot as quickly as possible to create a sort of Microsoft Graph-like back-end for consumers. Copilot is more useful, and thus may be more interesting to users, especially those who may pay for Copilot Pro, when it has access to this data. And that obviously benefits Microsoft.

So, yes, this is about using OneDrive to “feed their AI.” But it’s not necessarily as nefarious as it sounds, at least once you get past the forced OneDrive Folder Backup enablement and the way Microsoft jammed Copilot down everyone’s throats two years ago. Businesses have to opt-in to (and pay for) this capability. Consumers have to opt-in, too (see below). And Microsoft doesn’t use this data to train its AI. That’s true whether you’re a consumer or a business.

So, again, yes, it’s feeding Microsoft’s AI. For your use and only if you enable it. It’s not helping Microsoft directly, it’s helping Microsoft by making Copilot useful enough that some consumers and businesses will pay for the service. This is the definition of a healthy relationship–a company offers a service and you pay for it if you see the value–with the caveat that Microsoft is using dark patterns and questionable behaviors to guide you to just using it.

But you do have to opt-in and make this choice. And you pretty much have to pay for this, though Copilot in Windows 11 for consumers now has “File search” and “File read” options so it can interact with files on your PC, which could include those in OneDrive. Not for training. But to ground the AI for your queries about those files. If you care about this stuff, it’s worth looking at the Settings app in Privacy & security > Search, too.

If [the data in OneDrive] is going into a larger model that others are using then I wonder what actual use it would be, and I would be rather upset if something that I had done came out verbatim from the AI, as has been known to happen…not that I would be likely to know. Corporate clients would probably be more upset

The way to think about this is that Copilot uses a cloud-based LLM–really, multiple LLMs, or models–and that your data isn’t “going into” those models but is rather being used to ground those models to your data set for purposes of queries and answers

Which brings me to the point of privacy. Microsoft’s own statements of OneDrive privacy state

None of your organization’s data is used or transferred by Microsoft to train AI models, large-language models, or any other models.

Are these suddenly just weasel words? Is this single prompt to Open OneDrive meant to imply consent to allow Microsoft to train AI models? Do you know if this is showing up for Business accounts? And if so, is the model staying within the tenant?. But again, you’re opting in to this. As are businesses. Which also have to pay for the service 100 percent of the time. This is only happening when customers choose to enable it.

This was what triggered Microsoft’s 2024 statements about its privacy policies and AI: Someone read a privacy statement, misunderstood what it meant, and figured that Microsoft was stealing their data to train AI. And that’s not true.

So, no. You have nothing to worry about here unless you think that Microsoft would violate a legal agreement, which would open it up to class action lawsuits, governmental regulatory action, and whatever else. And if that’s what anyone thinks, they’re not using Microsoft products anyway and have bigger issues to deal with. Like that bunker they’re going to build some day.

? Audiobooks

owllicks asks:

Do you have any new audiobook recommendations? I really enjoyed Apple in China, as well as your review of it, and I’d love to find similar titles to read or listen to. I spend most of my workday listening to audiobooks to help me focus, and I’ve been looking for something to break out of the hard sci-fi rut I’ve been in lately.

When it comes to books, I split my time between Kindle and Audible, mostly Kindle. And when it comes to listening to things, I split my time between music, podcasts, and Audible (and, come to think of it, YouTube, if it’s an interview or something that doesn’t rely on video). So coming up with audiobook-specific recommendations can be difficult, if only because that’s not the only way I “read” books and it’s not the only thing I listen to.

But in the Apple in China category, I did read (on Kindle) Careless People: A Cautionary Tale of Power, Greed, and Lost Idealism, which is about Facebook/Meta and perhaps deserves as much attention as Apple in China, especially for the explosive revelations in its final chapter. This isn’t always the case, but I got this one on Kindle in part because I listened to a preview of the Audible version and didn’t care for the narrator’s voice.

I also recently read The Valve Effect: An Independent Biography of Gabe Newell on Kindle; it’s a history of Valve and Steam, and there’s no much out there on these topics, but this isn’t well written and isn’t on Audible anyway. And I’m mostly through The Insider’s Guide to Innovation at Microsoft, which is terrible. Based on the Audible sample, that version is even worse. I can’t recommend this one.

Outside of the industry books, I just finished Never Flinch by Stephen King, which was OK but had a poor ending. (This was also on Kindle and for the same narrator issue on Audible.)

On Audible, I’m partway through The End of the World as We Know It: New Tales of Stephen King’s The Stand, which is surprisingly good so far, The History of Rock ‘n’ Roll in Ten Songs, and Down and Out in Paradise: The Life of Anthony Bourdain.

I also bought, but haven’t yet started, Armada: A Novel by Ernest Cline (of Ready Player One fame). And I’ve preordered the next Dan Brown (The Da Vinci Code) book, The Secret of Secrets: A Novel, which comes out in September. These are both on Audible as well.

? The AI bubble

jrzoomer asks:

Paul do you feel like Microsoft and Google will continue to have the edge over Amazon in their cloud business in this new AI age? It seems like Microsoft is the leader currently in new AI deployments and companies given their partnership with OpenAI and their enterprise dominance, followed by Google Cloud and last by AWS. I guess this may be the current trend but is this temporary in your opinion or more of what’s to come for the longer term?

This is difficult to summarize easily because it changes so much week-to-week. But the fundamental similarities between the three big AI infrastructure providers you mention, which is all of them, are that each is building backend services that will be used by third parties while also building first-party services that they sell directly to customers.

By comparison, Apple is only doing the latter: Its WWDC announcement about allowing developers to use its foundational models applies only to local AI models running on devices, not to its cloud. And OpenAI mostly relies on Microsoft’s Azure infrastructure today, though it can now partner with third parties at Microsoft’s discretion. So even though it’s in some ways the bigger player in this space, it’s also reliant on others in ways that Amazon, Google, and Microsoft are not (infrastructure-wise).

At a high level, you might argue that AI doesn’t really change the dynamics at Amazon, Google, or Microsoft. And that for all the tens of billions of dollars each is spending each quarter, a successful outcome in this battle will simply mean that nothing changes. Each should still control some portion of whatever we call these markets, just as they do today in cloud computing. But each is obviously also keen to grow share and revenues, too, and that’s part of what makes this AI era so fascinating. It’s moving so quickly, and these companies are aggressively trying to outdo each other–and, when required, outright copy each other–to maintain/grow their respective advantages.

In cloud computing, Amazon AWS has been the big dog since its inception. Azure started small but has grown dramatically, and it’s now a solid number two, not a distant number two. And Google Cloud, despite strong gains, is still a distant number three. Here, I’m evaluating things based on third-party usage, customers that pay these companies for their cloud infrastructure.

But each also offers first-party services, and here, Microsoft is king thanks to Microsoft 365. Google is a solid number two, and Amazon appears to be mostly retreating from this space (and it’s even adopted Microsoft 365, if begrudgingly).

These companies respective cloud infrastructures give each a nice position for AI, but there, too, it’s split between first- and third-party, and this shifts a lot and will continue to do so.

We don’t hear a lot about big customer wins at Amazon Bedrock, or whatever, but then we wouldn’t in most cases. Developers and other third parties just use AWS and it’s likely that many will simply move into whatever Amazon offers there, and some of that is third-party AI access (OpenAI, Anthropic, etc.) On the first-party front, Amazon has Q (which I assume literally no one uses), and it has Alexa+ for consumers, and … I don’t know. Eh.

Google’s core strengths are its first-party services and its ability to integrate with the online search engine anyone cares about and a dozen or more consumer services (Gmail, Maps, etc.) with billions of users. So it has the same basic advantage that Amazon has with third parties on AWS: Huge audiences that use these services and are likely, in many cases, to just use whatever AI it is that that company offers. (The caveat here is antitrust, but we’ll see.)

Microsoft is clearly still hurting from its defeats in mobile and devices, and its fall from sole ownership of personal computing, and it has long sought to be there when “the next wave” happened. So its embrace of AI is perhaps explained in part by that institutional memory, but also aided by its fortuitous investments in OpenAI, and the unique partnership arrangement it secured. Had OpenAI somehow ended up in some other company’s lap, it’s likely that Microsoft would have done well in time, but thanks to OpenAI, it could seize the day and jumpstart the current AI era. And so Microsoft hopes to leverage its strengths in cloud computing to help both OpenAI and whatever other customers as third parties, while building out its own first-party solutions, mostly Microsoft 365 Copilot, to leverage its dominance in office productivity.

That makes Microsoft unique in some ways. It has massive customer bases on both sides that it can sell AI to. Amazon is mostly third party. And Google has both sides, too, but its third-party infrastructure capabilities are smaller because its cloud infrastructure is smaller. Obviously, anyone building on Android will at least consider Gemini and Google, but I suspect a big chunk of that will go to third parties AIs too.

But no matter how one views this, I feel like these three companies will all land in similar places respective to each other when the dust settles. Or when the bubble pops, as we might more correctly view this. There are no other companies that are as well positioned and have as much money to spend, assuming you leave out Apple. Because Apple tends to do things its own way, and it will likely need to lean more heavily on third party AI to make up for lost time and missing capabilities. It’s never going to be an infrastructure provider, regardless.

There’s all kinds of uncertainty, of course. No one could have predicted the Sam Altman drama or its split with Microsoft on infrastructure exclusivity. And we all anticipate a day when that split will be complete and Microsoft will have to go it alone. It’s clearly working towards that now. But more broadly, there’s a big story about AI adoption across almost every single app and service across every single platform, this notion of “if everything is AI, then nothing is AI.” And how or if this changes things. If I can get some AI in, say, Typora or Notion or whatever, and it’s free or next to free, why would I pay for OpenAI, Microsoft Copilot, or anything else? Is there a great leveling moment coming? It may have already come. These things are in flux.

But … I think the big players will still be big. The ratios may change, but these companies, and maybe a few third-party AI providers, especially OpenAI, will likely continue to be dominant.

You know, unless something changes.

? Pixel imperfect

epsjrno asks:

If tin whistles are made of tin, what are fog horns made of? I’m kidding. Just trying to get this party started. I would have asked about your impressions of the Pixel 10 offerings but you pretty much conveyed them well in the Pixel Matters article.

As Stephen Wright once observed, the world is a small place. But I wouldn’t want to paint it.

I could write endlessly about the new Pixel devices, but I had to end that somewhere and move on with life. I give Google some credit for turning what is essentially a very mild set of upgrades to a family of products that almost no one uses or cares about into something very interesting. Though in some ways, the bigger news is that the company also indicated, via an interview, that it plans to continue forward with Pixel for the long term despite its relative lack of market penetration. This is a bit of clarity and reassurance that I think many Pixel fans need.

But I have almost too many thoughts.

There are indications that Apple will stagger the next (2026-2027) generation of iPhone releases over some period of months, with a foldable iPhone appearing late next year and then the normal (non-Pro) iPhones not shipping until the following spring. And Google just messed with the Android release schedule and in two ways: It moved it forward (backward?) by at least a quarter to accommodate Samsung’s late summer annual foldables launches, and it is staggering the release of new Android 16 features across several QPR versions. These things may or may not be related, but the old static annual release schedule appears to be over. Case in point: For some reason, the Pixel 10 Pro Fold and Pixel Watch 4 won’t ship until October. A first in both cases.

So much of the Pixel 10 Pro XL (and, I believe the Pixel 10 Pro) is identical to its predecessor. Including, oddly, the camera system which one would think still needs an annual update.

So little separates the Pixel 10 Pro XL from the Pro, aside from the screen sizes. But the Pro XL supports 25 watt Qi2-based wireless charging and the Pro is only 15 watts … for some reason. Ditto wired charging, where the Pro XL supports 45-watt fast charging and the Pro is stuck at 30 watts.

The Pixel 10 Pro and Pro XL both support Zone UFS storage, but only if you get 512 GB more of storage. For some reason.

The differences between the Pixel 10 and 10 Pro are smaller than with previous models. There is a very strange resolution difference on the displays despite them being otherwise identical. Both have 30-watt (wired) and 15-watt (wireless) charging. The Pro has 16 GB of RAM, but the non-Pro has only 12 GB (and there are no 512+ GB storage options, so no Zoned UFS storage). And though all four cameras are different, the non-Pro now has three rear lenses, like the Pro, and it even has 5x optical zoom on the telephoto lens (albeit at 10.8 MP vs. 48 MP). You lose the Pro controls and Pro Res Zoom features on the non-Pro, and some high-end video features, but most don’t need that. And the Pro is $200+ more expensive than the non-Pro Pixel 10. Google has kind of made the Pixel 10 (non-Pro) a bit of a no brainer. The phone I bought starts at $1200, which is ridiculous. But the Pixel 10 starts at $799. And that is … reasonable. (Yes, it starts at 128 GB of storage, not 256. But still.)

I have been wondering about Google Home and Google’s first party speakers and displays for a long time. I wanted to get a smart display, maybe two, but the Google Nest Displays are so out of date, I couldn’t do it. But now they’re talking up Gemini Home. And there are hints of at least one new speaker, finally. They need do more here, and more quickly. Also, end the feud with Sonos. And maybe just buy Sonos. And do all this before Apple gets its act together and releases a HomePad or whatever.

The Google Pixel Buds Pro line was never good enough for me from an ANC perspective, but now we have a new Pixel Buds 2a … with ANC. Where’s a Buds Pro 3 that can take on the AirPods Pro? Which is about to move to a third generation version.

I like the Pixel Watch, and I bought a Pixel Watch 3 last year in part because it was free or mostly free because of whatever phone purchase, and I got free cellular for two years through Google Fi. The Pixel Watch 4 won’t ship until October, and the diferences there are minor, but still maybe of interest. A domed display, which apparently makes it appear bigger to the eyes. It’s brighter. Better battery life, always a good thing. A newer processor (finally), and it’s actually new silicon. But aside from Satellite SOS and a few AI features, basically the same functionality. This is a tougher sell, to me, and I’d only get $140 in trade for my Pixel Watch 3. Plus, we’ll be away when it comes out, so I couldn’t preorder it anyway.

But the big one, the thing I’m perhaps most uncertain of, is the Pixel 10 Pro Fold. I am quite interested in a folding phone. Was waiting on this announcement so I could compare the Fold to the latest Samsung. And … it doesn’t matter. We’re away in October, I have to wait. And maybe that will give me time to read some reviews, see what others think, and then … we’ll see. I’m not sure that either will be as successful as a hybrid as, say, the iPad with iPadOS 26. But it does seem like they’re getting there.

But I will wait. Impatiently.

I think I’m aligned with many when it comes to Google and its consumer hardware. I like the idea of it. I don’t like that it seems to zone out sometimes and just give up on things. And I worry about what this means for the future. But I also perk right up every time it announces anything in this area, and I would love to see them hit one out of the park, across the board (ecosystem). And they just haven’t done that yet.

Hope springs eternal.