Many, perhaps even most, popular mobile apps have some sort of online component that will surely stop working if the business is closed. And apps made for an OS two releases back or more regularly seem to stop working if not updated. I know I've certainly got paid apps that simply do not work at all anymore.

This is not a new phenomenon. Try running a game made for Windows XP or even Photoshop 6.0 on Windows 10, it likely doesn't work even in the backwards compatibility modes. Companies like GoG.com have made it a successful business to take old beloved games and make it convenient to play on today's machines, and I'm a happy customer of theirs even though I also own the originals in many cases.

The way I see it, the software was sold to work to work in a certain environment and period of time, and I'm free to try to keep it running best I can. But I'm certainly not entitled to demand that the software developers keep supporting me forever simply because I paid once. Therefore I strongly prefer subscription models for the software that I regularly use, as it's the only model that aligns both my and the developer's incentives in the long run.

I disagree with the subscription model helping in the long run, because with a flat fee model, you at least have a chance of keeping the old version running. If I like something and the developers add new features, I'll buy the upgrades to get new features; that's how developer incentives can align with mine.

A subscription model doesn't automatically exclude running old versions after you stop the subscription, see for example the IntellIJ by Jetbrains model.

The "available in the next version" effect of one-time purchased software creates bad incentives for the developers to exclusively focus on new features even if they're gimmicks, because that's the only thing that brings in the money. As a customer, I literally cannot give them money (except gifting copies to friends) just to keep polishing what they have even if that's exactly what I'd like them to focus on.

I've found that after Adobe switched to Creative Cloud, and Jetbrains to their yearly subscriptions, I've benefited as a customer of Photoshop and IntelliJ respectively by both more rapid releases of genuinely useful features, not in big-bang releases but constantly over time, and that they've stepped up their general polish in the products. Photoshop CC is SO MUCH BETTER these days than the always buggy yearly releases of the past, because they don't have to cram 51 new crap features in every 12 months! A developer with a subscription model doesn't have the constant feature pressure, and can work on what the majority of current customers values most, and some times that is just bug fixing and polish.

A subscription that works basically like paid software you can keep plus a maintenance fee is much more palatable to me. I place a high value on being able to maintain access to my tools and creations, and prefer models (including FOSS) that permit that.

> Try running a game made for Windows XP [...] on Windows 10, it likely doesn't work even in the backwards compatibility modes.

This is the exact opposite of truth. Windows 95? Sure. But XP? Rubbish.

Don't I wish that were true. Off the top of my head, I know for a fact that a few of the Splinter Cell series games are impossible to run nowadays, and other games that used the same Ubi DRM too. And I just can't seem to get the original Fallout 3 working for either love or money.

As someone pointed out elsewhere in the discussion, app subscriptions mean DRM that's basically guaranteed to kill your ability to run the software once it gets old and everyone loses interest. If there aren't enough people buying old games for the publishers to keep them working, it's unlikely enough people would be willing to subscribe to make it worth their while maintaining the subscription processing.

I'm sure "someone" knows how to run them, arguably justifiably if you have paid for them.