As of 2022, google got chrome to work on fuchsia, completing 90% of the functionality an operating system needs to have.

As mac, android, and Linux is unix-like, Windows is Windows. What does that leave?

@DanielPugh Fuchsia, which still has thousands of people working on it iirc.

Are we classifying Plan 9 as a unix-like operating system?

@nicwest06d6 Yes

I remember making a similar bet in 2004 about 2014, and at that time we couldn't agree whether mobile phones counted or not.

How will you deal with as of yet unknown devices? Which count and which would not?

@ErwinRossen Any device that is a software platform would count, meaning a popular reason to use the device is it's third party apps. Like a mobile VR headset (but Oculus is Linux).

However market share would be a best estimate across all devices so it would not be enough for a mobile VR device running a new OS to capture 3% market share only in comparison to other VR devices.

Could you clarify what platforms count?

My prediction is that Unix-like platforms are good enough that there’s little reason to replace them. It’s possible that some future platform (i.e. not desktop or mobile) comes along with a new OS paradigm but I don’t think that’s likely.

Desktop only? What about Android?

@Stralor Mobile counts but Android is based on Linux so I'm considering it "unix-like"

@JeremiahKellick ah wasn't aware! thx

Gonna bet yes due to the following factors:

• nixos demonstrates that the right way to bootstrap an OS is to make it so that the original projects don't need to know anything about nixos for nixos to add support for them. Make packaging tools very good at patching. Eventually your packaging tooling will start having support for software from other OSes. Whatever comes next will be doing this.

• The multipolar OS ecosystem drives a constant shift towards cross-platform tooling, and web apps being capable of more, consider webassembly. Target platforms will matter less.

• IIRC linux has fundamental problems, can't remember what they are just now, but if it didn't why would it be the case that:

• Fuchsia is being developed. Fuchsia is pretty good. Google's VR OS, when it emerges, may run Fuchsia, and be popular. It's also expected to be deployed to phones as well.

By what metric?

@Jai I'm also interested to hear the intention. I wouldn't be surprised if FreeRTOS is already in double-digits, but we have no easy way to count the dishwashers, cars, and bombs that use it. The current numbers online are web-focused, so the term "market share" probably refers to web servers or clients?

@fwbt @Jai Hm, good question. Looks like these numbers are harder to find than I hoped. Most estimates are based on Web access which is probably what I'll have to go with. Numbers like this https://gs.statcounter.com/os-market-share. Though I would like to be a bit more broad and include game consoles and web servers, for example. I'm open to recommendations. Embedded systems aren't in the spirit of this, however. I'm interested in devices that are software platforms, not hardware focused.

Operating System Market Share Worldwide | Statcounter Global Stats

This graph shows the market share of operating systems worldwide based on over 5 billion monthly page views.

@JeremiahKellick How would you consider smart displays? I think some already run Fuschia, and although they don't have big app markets, I could imagine that they reach a similar status to game consoles in future. Maybe you could add some threshold like "most users these devices use them for software services from more than one vendor"?

@fwbt Yeah, I'd like to include smart displays with third party apps. The criteria you gave would be ideal, but this will be limited by whatever I can get reasonable numeric estimates of. After some cursory searching, I might have to go with web usage even though that fails to capture some things I'd like to include. But I'm open to better metrics.

@JeremiahKellick Thanks! So if 8 billion people have smartphones in 2033 and this makes up 75% of the "market", it would suffice for either Google or Apple to partially migrate to a non-Unix kernel, or for adoption of a new programmable device (like displays running Fuschia) to reach 400 million users. I guess that could be assessed on sales numbers alone, rather than web metrics that might exclude these devices.

Mmh, BSD at <0.5%

macOS is counted as Unix here right?

@2eb7 yes