As for microcode: IMHO Stallman is just wrong about it. He's okay with the logic being in hardware or non-volatile ROM, but uploading the code to volatile RAM when the hardware starts is a huge problem? I don't follow that logic. Of course it would be better if this was open, but it's much less of a deal than he makes it out to be.
Stallman will be Stallman in all his weird and egocentric ways.
This is a case of the ship choosing to go down with the captain.
The same way someone distributing a program that makes use of an LGPL library is only obligated to release the changes they made to the LGPL library, not the entire program. You are allowed to build your proprietary GUI on top of the LAGPL base and offer it as a service over the internet but you are not allowed to hold any changes to the base for yourself.
The line of what is considered a derivative and the line of what is considered distribution are orthogonal and one of the 4 quadrants is currently missing.
The goal is to effectively prevent the introduction of incompatibilities through Embrace-Extend-Extinguish. Weak copyleft like the LGPL has been effective at preventing EEE without discouraging all corporate use. Strong copyleft is actively avoided in corporate environments but weak copyleft is not. The goal would be to achieve the same but in a cloud environment.
The space it would be useful in is preventing cloud providers from introducing incompatibilities between their offering and the original FOSS project yet allowing them to build services on top (GUI's, billing, etc.) which allow them to differentiate.
How this gets encoded in legalese is their job as stewards of the GPL license suite, holders of the copyright over the text of the licenses and social gatekeepers of license proliferation.
"Another significant change in this release is that it was pointed out that there were error messages in Linux suggesting users to update x86 CPU microcode. Since such microcode is non-Free Software, such messages don't belong in GNU Linux-libre. We now have patterns to detect and clean up this sort of message. A number of them were introduced recently, relying on microcode changes to mitigate Spectre and Meltdown problems, but there might be others that go farther back. I haven't yet made my mind on whether to go back, check and possibly respin such earlier releases."
https://lists.gnu.org/archive/html/info-gnu/2018-04/msg00002...
2. Nobody stops anybody from creating new licenses, in fact many many projects tried to do just that (MPL etc). Most of them fail simply because you cannot have your cake and eat it too, you either allow some things or you don't. You either require source code for programs users interact with on the network, or you don't; I'm not sure what middle ground is there to find. If you want to release certain parts of a website but not others, you can do it already in many areas - any SPA client-side code can be covered by GPL 3, for example. AGPL in many ways was a proto-GPL3, and it's to the FSF's credit that they didn't drop it, but actually updated it when v3 came about. They even try to explain the differences with dedicated documentation [0].
[0] https://www.fsf.org/bulletin/2021/fall/the-fundamentals-of-t...
2. The popularity of FOSS licences to some degree depends on endorsement by FOSS organisations like FSF and OSI. Custom licences are actively avoided and criticised. The GPL3 does nothing to protect against SaaS because that is not a threat that was envisioned by Stallman at the time. He was more concerned (and rightfully concerned) about Tivo-isation. Tivo-isation is a Right to Repair issue and it is really painful today. But Tivo-isation and SaaS are just different sides of the same coin called erosion of ownership. By failing to more strongly endorse the AGPL and by failing to provide a LAGPL, Stallman and the FSF (Stallman's fanclub) neglected the threat of SaaS. The LAGPL would have probably had more impact than just complaining about SaaS at conferences. Eventually some projects like MongoDB switched to non-FOSS licenses and others like CockroachDB switched to the BSL which eventually reverts to a FOSS licence after a predefined amount of time. For all of these projects the AGPL would have resulted in almost no adoption, but the LAGPL might have been a viable option. The FSF failed at their main job of license stewardship by failing to provide the licences society needed while being one of the two gatekeepers of licence proliferation.
Note, by SaaS I am currently referring to PROVIDER HOSTED SaaS which implies a service provider doing the computing on behalf of the user. SaaS can also be just a licensing model like it is for Photoshop.
And to support that argument I bring three anecdotes.
1. He is much more attached to his own creation, the GPL, as a tool for driving software freedom and as a "hack" on top of copyright. As such, he is pro-copyright and pro-IP (intellectual property) in general (but against just software patents). Because without copyright the GPL could not exist. But without intellectual property, everything becomes reverse engineer-able and decompile-able and any available source code is effectively free. Copyleft and the GPL are not enforceable in a world without copyright but they are also not as necessary. I confirmed this position directly with him and also through the articles he wrote.
2. He is more attached to the GPL as his brainchild than any other variant. As such the AGPL (not created by him) has always been a bastard cousin and his organisation never bothered to create a LAGPL (like the LGPL and AGPL combined) even though it was requested several times. The GPL has allowed the rise of SaaS and the AGPL is unpopular because of how strict it is and therefore it was rarely used. A LAGPL would have been a middle option that would have forced SaaS corporations to contribute back to the community and could have had a grater impact than the AGPL.
3. With regards to microcode his position is asinine. He would prefer hardware for which the user is simply unable to change the microcode (even if it is broken) just to avoid blobs. He is not pushing for opening up the blobs. He is not pushing for Free as in freedom microcode. He is not pushing for reverse engineering the blobs.