accessibility has been going down for the last few years

Quick counterpoint as this gets raised a lot and I consider it disinformation.

In the Xorg -> Wayland transition, accessibility was immature as were a number of other things. And the implementations between x11 and Wayland were different (and so difficult to compare feature by feature).

Because of this, Wayland detractors made accessibility a favourite bugaboo and, even now, it is possible to find examples of things that worked better on X11 than they do on Wayland.

And there is no denying that accessibility was worse on Wayland for a while. You can say that about other things as well.

What the detractors do not tell you is that, for the major desktop environments at least, accessibility on Wayland is now better overall than it ever was on X11. Like a lot of things, whereas the poor security in X11 allows you to do many things, the capabilities have to be explicitly built into Wayland resulting in a period with poor support followed by systems that work excellently (better than they do in X11). This is a Wayland truism overall but particularly true for accessibility.

Latency and security are improved in particular. Assistive tools in X11 are a massive security hole. And accessibility in Flatpak apps is now far better as the tech built to work with Wayland sandboxing helps with Flatpak samdboxing as well.

Finally, accessibility is a greater focus in Wayland and so still improving whereas it was always an afterthought in X11. So regardless of the current state, I would say things are looking up for accessibility.

Developers need to get back to developing for Linux natively (not Steam)

No thank you.

Windows APIs are very stable. In many ways, they are better than Linux APIs for things like games. I will come back to this.

Games do not gain much performance by being native. The instructions to the GPU are the same on both platforms and this is where most of the performance stuff happens.

Linux adoption for gaming will be much faster if most titles work on Linux. A strategy of making Windows games work on Linux is going to result in a vastly larger catalog than will getting games studios to target Linux natively. Game studios do not want to create ports for small platforms.

What we need is to convince the game studios to ensure their Windows games work on Linux as well. We need to resolve the kernel level anti-cheat situation in particular. Perhaps we need these to be cross-platform.

The Steam strategy is a good one.

Now, back to Linux native…

There are many examples of Linux ports that now do not run or have problems on modern distros because of changes to the Linux userland since the games shipped. At the same time, Windows versions of these games work via Proton. Crazy but true. The Windows versions work better and keep working for longer (on Linux).

You could easily make the case that this is a problem with Linux as this instability is a major drawback of Linux for all commercial software (binary distributed is really the problem). It is not black and white though as this flux is what drives Linux forward. Over long periods of time, proprietary platforms have trouble keeping up. But this is a real problem for apps that ship as binaries.

On the non-game app side, the solution is Flatpak. Flatpak works by installing a parallel userland so that the Flatpak has access to the libraries and services that it expects.

So, one solution could be to use Flatpak for games on Linux as well. Or to create a gaming version of something that works like Flatpak does.

But guess what, we have that already. It is called Steam. Steam lets you install a parallel userland so that the game has the libraries and services it needs to run properly. It just so happens that the platform it provides is Windows. This works well for games.