An increasing number of these devices have touchscreens, so Phosh could work quite well. And for me it's more about how these devices are leaving a lot of potential functionality off the table. When hundreds of thousands or more people are all using something, you can never predict all the different use-cases they're going to have. So it's strange to me that these frontend developers are putting so much effort into turning them into dumb single use appliances, when the entire world of Linux software can just as easily be made available.

But yeah I was thinking about a hypothetical search scenario. I don't see it being useful for anything more than average local-level forces. Some of them support dual-booting, so those stripped down emulation frontends could come in handy in that case. Have one distro, ideally complete with full disk encryption. Then have another that actually is for the games. If somebody turns the display on and sees it demand a password, they'll demand further investigation. So they would need to be met with an open menu with an obvious, full list of games ready to play, with no indications there is anything else to search into.

No, that's really not how this ecosystem works. I'll give an example based on the Retroid Pocket 5, just because that's what I own and am familiar with. The company ships it with Android 13 by default. So this already has a full desktop environment, full suite of available app stores - everything that any other Android device can do, plus built in game input. No extra hardware needed. I can outright install Signal, Briar, Matrix, any of those kinds of things on it right now. Retroid never did any testing or development for that use case, it's just what happens when a device has a full, open Android experience from all of the development that has gone into Android and apps over the years.

However, they also chose to use a chipset that has support for Linux as well. At least two projects have already ported their variants of Linux to the RP5 - Rocknix and Batocera. While it's true that porting, testing, fixing, and maintaining operating systems has a labor cost, you're trying to apply standard business logic to something that doesn't operate that way. Batocera accepts donations, and Rocknix doesn't even accept any financial compensation. These are volunteer projects, labors of love. They are not subject to capitalistic incentives, and nor is it entirely accurate to call open-source projects like these "products." The processes involved are more organic, more democratized. It's a commons, and anyone can get involved.

The reasons they're designed the way they are is because that's what fits their motivations and project goals. It's the appliance model - they want the device to do one thing, and do it well - to play games, particularly retro games. So they put in only enough software to make that happen, and then try to make it as much of a polished experience as possible.

I am only trying to point out that the only thing getting in the way of more general purpose systems being available to these devices pretty much comes down to whether more people feel like trying to compile them.