But what of the Steam deck? If it sells well enough then you will absolutely be buying a machine with Arch pre-installed on it from Walmart.

A big 'if', and not currently the case, but also not as impossible as commentators such as yourself would suggest.

I'm not sure what that has to do with the desktop. AFAIK the steam deck is a handheld gaming device similar to a Nintendo Switch, not a desktop PC. It seems about as related to desktop Linux as google stadia, i.e. not much. When I say machine with Arch I mean a desktop/laptop with just the stock Arch on it.

Edit: And if you really want to be pedantic, ChromeOS is technically Gentoo underneath the covers, but the only reason it works for its target use cases is because it's heavily modified with some proprietary components. Getting one of those is really different from the experience you would have with trying to put upstream Gentoo on a laptop.

>AFAIK the steam deck is a handheld gaming device similar to a Nintendo Switch, not a desktop PC

It is essentially a fully fledged desktop PC in a handheld form factor [0]. It is identical architecturally to a desktop PC with integrated controller running Arch-based Steam OS 3.0.

>When I say machine with Arch I mean a desktop/laptop with just the stock Arch on it.

Which is fair, but not what I asked.

[0] https://www.techradar.com/uk/news/steam-deck

Yeah I get that, but you still probably are not going to want to run desktop apps on that thing. It doesn't have a keyboard or mouse. It looks more suited to running tablet apps. And regardless of that, my point is that the status of gaming APIs shipped by one vendor on a proprietary bundling of Linux has little to do with the status of the rest of the platform, which by the way is also woefully inadequate for shipping tablet apps. And also: Linux still is not able to adequately run the majority of desktop software for Windows that people actually want to use, which is all proprietary. Progress has not changed or increased here, Wine is still about as far behind (relatively speaking) as it's ever been. AFAIK the only reason Proton is able to work is because support for Direct3D 11 was added, which at this point is a 13 year old API, and it still doesn't work perfectly.

While I doubt that many people are going to want to run desktop applications on the Steam Deck, it appears to have the ability to connect a monitor and keyboard/mouse. Devices like this, Chromebooks, and many others will make Linux a more consumer facing OS in the decades to come.

I agree that the Windows API, beyond those needed for games, is a thorn in Linux's side. On the other hand, I also see the Linux desktop taking a larger percentage in the years to come. It won't really be because of growth in the Linux user base. It will be because desktops will have less of a presence outside of business use.

"Devices like this, Chromebooks, and many others will make Linux a more consumer facing OS in the decades to come."

Can you see how that's the point I'm making? "Linux" is still decades off from being a viable desktop OS, and it will always be so. A new embedded device running Linux that you can technically connect a keyboard and mouse to doesn't help the general case that Windows targets, or the case of purpose-built desktop/laptops that Apple targets.

With a dock you can have a keyboard and a mouse :)

Moreover, given that the Steam Deck is essentially a PC, any investment in the Deck in drivers, APIs, desktop environments, display manager (Valve have invested in KDE Plasma/Wayland and Gamescope), native game compatibility, compatibility layers (applies to all software, not just games) is also an investment in desktop Linux as a whole. Even if we are just limiting it to games, having more games on Linux makes the platform more viable for people who want to use their computer for gaming.

This isn't even mentioning how much has moved to either web or cross-platform electron apps in the last 10 years (which includes Teams, Spotify, Discord, Youtube, VSCode and others)

>which by the way is also woefully inadequate for shipping tablet apps

I will bow to your experience on this one, I don't have any experience in this area.

>adequately run the majority of desktop software for Windows that people actually want to use

Actually you can run pretty much all Microsoft Office on Linux - the big ones that people still have trouble with are Adobe Creative suite.

All of that stuff is fine and good but it's the same incremental progress as always. I don't see anything particularly special about using this device with a dock, that seems like a really niche use case. These type of devices are not new and have never been big sellers. I'd assume most customers would just use it for games and don't care that it technically runs Linux and allows them to switch out of their game and fumble around trying to use LibreOffice or GIMP on a touchscreen until they get frustrated enough to plug in a keyboard. Unless Valve is planning to invest billions of dollars in overhauling KDE for general uses (including non-gaming uses), then I don't think anything is going to change significantly. Yes, that's how much money it would cost. This is what Google had to do to with Android to make that kind of "Linux" a viable platform, it won't happen with anything else unless similar amounts are invested.

And for windows software I'm not talking about Microsoft Office or Adobe, I'm talking about the "long tail" of Windows apps that don't get attention from Wine developers. There's still a lot of those.

> Linux still is not able to adequately run the majority of desktop software for Windows that people actually want to use, which is all proprietary.

All of this has been getting replaced by the web over the course of twenty years.

The primary exception to this is games, because they're so resource intensive and benefit from being closer to the metal. Which is why gaming APIs are significant -- AAA games don't get converted into a web page and therefore are the main thing still tying a large group of consumers to Windows.

You assume customers don't want to be tied to Windows which is not the case. Maybe you and I got sick of Windows 20-30 years ago but Microsoft's customers (the vast majority of desktop computer users) don't care. They see ads in the OS and they think "that could be our ad up there".

Sure they don't care. But that means they don't care.

If you tried to sell a PC with Linux on it at Walmart in the year 2000, people would return it because it wouldn't natively run Microsoft Outlook or Starcraft.

If in a few years from now you're on webmail regardless and it can run every game on Steam, there is no reason to return it. And it costs less by the amount of a Windows license, so there is a reason to buy it.

That cost reduction doesn't just come out of thin air. If you want to build software that is competitive with Windows in other aspects, then you will have to charge similar amounts. Otherwise you're just leaving money on the table, because we know by now that customers are extremely willing to pay for it. And as long as Microsoft is shipping new APIs on Windows, Linux will never be able to run every game on Steam or every Windows program. It will always be perpetually behind playing catch-up, just like it has been for the last 25 years. The problem with your plan (and with all of these type of plans that I've seen people post) is that it requires Microsoft to give up and cede the market, which is not going to happen.

> That cost reduction doesn't just come out of thin air. If you want to build software that is competitive with Windows in other aspects, then you will have to charge similar amounts.

The assumption here is that development costs are linear.

If you spend a thousand dollars developing an operating system and your competition spends a million dollars, theirs won't be a thousand times better than yours, it will be infinitely better because theirs might actually boot and yours won't.

But if you spend a billion dollars and they spend a trillion dollars, theirs won't be a thousand times better than yours, it will be 2% better. Maybe. Because your billion dollars already caused to exist an operating system that does everything people need from an operating system and there is a point of diminishing returns. Which is where all the major operating systems already are.

What you would expect in a market like this is price competition. 80% as good for 20% of the price wins. 98% as good for 0% of the price wins even more. We're already there -- Microsoft has been giving away the Windows 10 and Windows 11 upgrades for free. They now need that to be competitive.

They can add new APIs all they want, but if Valve is providing ones that are just as good and will allow your games to run on their consoles which have any non-zero market share in addition to Windows, developers will prefer those.

I don't think they are at a point of diminishing returns. There are still advancements in hardware and drivers, and that's where the investment really pays off. Microsoft has historically been really good at designing APIs for this stuff that developers actually want to use. The vertical integration with shipping the whole OS really pays off for them. In Linux, it's extremely difficult to make any changes because you have to plumb everything through several layers of the stack which are all in separate disconnected projects. Only a few wealthy companies can really do it well and those tend to have a narrow focus on a specific vertical (Google, Red Hat, etc).

Price competition on the OS doesn't actually matter, as you have noted the market has moved towards additional services. When you say "98% as good for 0% of the price" you're not factoring in all the additional paid services that the company wants (and needs) to sell. I can tell you that Canonical certainly does not charge 0% of the price for everything, if they are taking a customer from Microsoft then they would probably like to gain 100% or more of that price from any given customer.

I have seen nothing to show that developers will prefer any Valve APIs, a lot of the games still seem to be running in Proton which actually shows that most developers still prefer to use the Microsoft stack and tooling. I should have mentioned this earlier but Visual Studio (the original, not VS Code) is also extremely important and entrenched, game developers are not just going to all switch to Vim and Emacs because Valve sold a product with Linux on it.