> "Cloud services and mobile rule the world now, and RMS never came up with a good solution to either besides "don't use them.""
I generally agree and I think the problem is bigger than SaaS. Consumers are now becoming accustomed to software "running" on low power hardware that simply cannot actually run on that low power hardware, which requires some sort of "SaaS" system, which in turn requires that they either trust somebody else (inevitably a for-profit corporation who very often has incentives misaligned with the user) or become their own system administrator (an unreasonable ask for the typical user.)
A concrete example:
Suppose a user wants a FOSS photograph organization application that supports organizing large numbers of images through tagging. A user coming from the proprietary world might reasonably expect this application to perform face detection and recognition to automatically tag all of their friends in the application. There are FOSS implementations of this kind of ML tech.. but what are the chances any of that heavy duty stuff can run on the users low power smartphone? Snowball's chance in hell. Even if you get it to run, it certainly won't scale as well as the SaaS alternative from Facebook/Google/Apple.
What's the answer though? I don't have a great answer to this, RMS obviously doesn't, I don't think any RMS supporters or critics do. Not that I've heard anyway.
Apple's on-device ML at least proves it's possible even if the FOSS world would take a while to catch up. A bigger problem may getting access to training data.
It's only possible to a limited extent. A low power smartphone will simply never be able to do as much ML as a high power server, so there will always be a capability asymmetry here. And that's not even getting into the matter of FOSS drivers for such hardware.
And as you say, training these models is a different matter than merely using them. The data and power requirements involved are a huge problem.
I generally agree and I think the problem is bigger than SaaS. Consumers are now becoming accustomed to software "running" on low power hardware that simply cannot actually run on that low power hardware, which requires some sort of "SaaS" system, which in turn requires that they either trust somebody else (inevitably a for-profit corporation who very often has incentives misaligned with the user) or become their own system administrator (an unreasonable ask for the typical user.)
A concrete example:
Suppose a user wants a FOSS photograph organization application that supports organizing large numbers of images through tagging. A user coming from the proprietary world might reasonably expect this application to perform face detection and recognition to automatically tag all of their friends in the application. There are FOSS implementations of this kind of ML tech.. but what are the chances any of that heavy duty stuff can run on the users low power smartphone? Snowball's chance in hell. Even if you get it to run, it certainly won't scale as well as the SaaS alternative from Facebook/Google/Apple.
What's the answer though? I don't have a great answer to this, RMS obviously doesn't, I don't think any RMS supporters or critics do. Not that I've heard anyway.