Is mutually alligning self-interests the best way to help free software adoption?
There are collectors of old things and a lot of them like new old things. There are people that want a DeLorean even though it was an an engineering mess, but some want a new DeLorean even though that's not their favourite pop culture icon car. So, there's a company that makes new DeLoreans and in the 50's, people bought old Model-Ts and suped them up with after market kits and called them "hot-rods".
I think there could be something similar with late Windows XP-Era computing, I think ReactOS would be great for people locked into devices that only had XP drivers and save them from legacy hell. Another target for ReactOS would be that hot-rod crowd I mentioned, I've looked into that crowed and they did a retrospective on Skulltrail, Intel's legendary Dual Socket 771 Core2 Extreme board and it actually holds up well with more contemporary APIs that spilt graphical draw calls on multiple cores. Those people love the Socket 775 platform because it supports years of legacy with some Socket 775 Boards having Windows 98 Drives and Windows 8 Drivers. I think people like os could piggy back off the legacy hot-rodders. What if somebody could construct a libre Socket 771 chipset IP Core (you know, a northbridge bus replacement) and support single, dual and quad sockets and support more contemporary and less expensive RAM so say if we integrated New-Old-Stock x4500HDs, it would more than likely perform better than original hardware due to faster memory. I think both us and the hot-rodders would eat that up. They would especially be interested if somebody can implement a clone CPU that would be a "Core2 Octo Xeon" later down the road and especially if we can make it 40 watts like the low power xeons. And of course have blob-free software.
I think Power is cool, but not the way to go and RISC-V is cool too, but I don't think it appeals much to average joes or hot-rodders. I like the idea of a total re-implementation of the socket 775/771 platform because old CPUs are cheap, but the RAM isn't. I don't think Dell will ever sell a RISC-V Laptop or a Power 11 Workstation and I think it's best to stay with mutual niches and they could possibly expand from there.
GNU was a clone of of Unix 5, ReactOS is a clone of NT 5. (5.2, to be specific) and ReactOS could not only be the new FreeDOS where it appealed to legacy and hot-rodder, but it could be a new GNU project where GNU does thins UNIX V could only dream of doing and with the Linux Foundation being in bed with MS, this could be our moment where we make presentations with a slide that says "GNU <3 libre NT Clones, but not Microsoft".
For hardware, it could go into Tesla style Phases:
Phase 1: Libre Chipset
Phase 2: Libre CPUs
Phase 3: Libre Laptop based on what we learned from Phase 1 & 2
(so, a drop-in replacement Thinkpad Board)
I think aligning with legacy and hot-rodders is our best bet because like it or not, the software controls the user in the form of legacy binary-only software, so there would be some appeal and maybe it could evolve from there. What do you think?
I took a deep-end dive into Free Software by purchasing a pre-librebooted machine. Although I absolutely LOVE the x200, the ports, no trackpad, a 'real' keyboard (not some chiclet), and excellent build quality. I was using a quad-core i7 before and the CPU speeds found on libre machines is somewhat lacking. If a x86 compatible deblobbed CPU were available, I would love to swap out the Core 2 Duo! If a different architecture like RISC-V or ARM hell even PowerX (Idk what number it is now) I would be happy to swap out the motherboard if it could be substantially more powerful than Core 2.
I think for another architecture to go mainstream then it actually has to perform better than what Intel and AMD have offering. I believe that Apple is slowly transitioning away from Intel to have a vertically integrated production chain and they are moving towards the same type of CPU that they make for iPhones (I believe I watched a Youtube video a while back or read it somewhere, I'm not certain of this).
Heres to hoping Intel/AMD get dethroned!
I'm thinking we can only rely what we want and find people that will use what we want but for what they want.
I think outside of Software like a new GNU Project, a good place to start would be to build a foundation on something old and evolve from there like implementing AVX-512. We don't care about backwards compatibility with non-free binaries, but others do. You have to get others from outside your camp to care.
I hope Intel and AMD get dethroned as well, but by what is the important factor.
I like the creative and strategic thinking evident in your post about legacy PC hardware and "hot rodders". I don't know enough about the specific technologies you mention to comment on those (hardware is not my strong suit), but it might be relevant that at least some of the patents on these older technologies have probably expired. So it may be possible to reverse-engineer and clone some of them as libre hardware designs.
I agree with you that making free code software relevant to users' specialist needs is a more effective strategy for growing the software freedom movement, than trying to make users care about software freedom (in the abstract) to begin with. I myself was brought into the software freedom movement through my role in Indymedia, where free code software served our need to create open-publishing websites. Another anecdote that's relevant concerns a person I met through the fediverse who is an actual "hot rodder" - a DIY car enthusiast. He got involved in free code software because he was looking for software to work with custom car parts and utilities he wanted to use. He nows uses a lot of free code apps and is well on his way to using GNU/Linux and learning more about the theories behind software freedom.