My energy is running out today but I'd love to :3 I should mention I'm pretty new to the field of computer architecture specifically and I don't really have the money to pursue it academically. But I've been working ("working" not in the sense of wage-labor, I've mainly lost hope in actually being employed in working with these systems.) with operating systems and computers in general for most of my life.

The situation in the computer industry as a whole (at least in the west) can be summed up as stagnation and monopoly. There are some new approaches and technology being deployed, of course (not counting the obvious grifts which there are a lot of), but only those technologies not threatening to the AMD/Intel/Microsoft/TSMC/etc complex and the system of individualized computing in the service of capital as a whole. As it is rn, the direction of actually existing computer architecture is dominated by a few monopolies more interested in turning out a profit by keeping people dependent on their technology, which only they can maintain due to patents and highly complex design and infrastructure, than actually building reliable, secure, and frugal machines that respect the privacy and creative energy of the masses. We have been brushing up against the limits of the conventional amd64/arm64, von Neumann, binary, register machine, instruction-set-having, etc architectures which can be seen in the diminishing performance returns in spite of rapidly sprawling complexity of designs and shrinking of feature sizes (think of transistor sizes) to physical limits and the deluge of errata, bugs, and especially security problems (Spectre, Meltdown, Rowhammer, etc) which threaten to make meaningless all the security guarantees people have taken for granted for years now (memory protection, isolation of tasks from other tasks, isolation of users from other users, really anything tbh). There is still some interesting research being done in the universities (some, the free money has been drying up for years) and in R&D departments (which often get their budgets slashed to fill the pockets of executives, of course) but most of it never sees wide deployment or implementation. Until this field is unshackled from profit, capital, and the market, the mainstream will probably continue as is with minor changes (goes without saying: minor changes, sold by the monopolies!), regardless of the technical merit of the alternatives. This is why the direction of computer architecture is inextricably interrelated with the development of political economy in the society it exists within.

But communism needs computer architecture too. The socialist states of the future have the great task of coordinating economic activity and human creativity on a scale never seen before. Luckily, we have access to techniques the USSR or the PRC never had. The calculation part of the "economic calculation problem" proposed by von Mises was capable of being solved in the 60s and 70s. The real problem is developing a system that can integrate all of the information from something as large, complex, and made up of so many disparate parts as a world economy and coordinate the activities of billions of people in carrying out a plan. It's honestly impossible for me to imagine fully and I think it will involve more changes than even the dreamiest of Soviet cyberneticists, who spent decades imagining such a system, could think of. I hope we live to see it. We already have the Internet and wide deployment of computers throughout industry so we have a base to work with but I don't think it will be enough. The internet is a highly "top-heavy" and centralized system that was added as an afterthought to our fundamentally isolated and individualized (like literally the "Personal Computer") computer systems. Any coordination between nodes on the internet (if they can even reach each other because of things like NAT or ISP firewalling) running conventional operating systems and architectures is an exception to the rule that tasks usually stay on the system they started on (think process migration or sharing of resources, even if you can do that its usually programmed specifically for the task). I think we will need to develop more social computer systems that naturally integrate better with other systems if we want to build socialism and carry out something as ambitious as a collective computerized economic plan.

Sorry if I'm being a bit vague, because there are many approaches and strategies we can try to do something like this. Talking more concretely, there are some systems that I think are ahead of their time. Most people (at least computer-people :3) are familar with Unix or Linux but they usually aren't familar with what the people who developed Unix did after. There was a time at a place called Bell Labs (as in AT&T) where researchers were allowed a lot of freedom to experiment with things using the vast amounts of money that came from a telecommunications monopoly. This is where Unix was originally developed. After the development and success of Unix, the developers of Unix forced themselves to start over and develop a system that allowed computers to integrate more naturally with each other over networks. They developed a kind of "universal protocol" that allowed computers running their operating system to share resources and use other computers' resources locally, as if they weren't remote at all. A Plan 9 system doesn't care if your keyboard, screen, mouse, hard drive, filesystem, printer, etc is attached to another computer or not. All resources on the system are represented as files. No special handling (ioctls) or protocols are needed to access any system resource! Through careful design and clever use of these abstractions Plan 9 is able to accomplish what far more complex systems can with a tiny fraction of the code size and development team. Tools were included in the system to make it easy to talk to and coordinate with other members of the team on development. Plan 9 is another very unique, innovative, and (at least in my opinion hehe) technically superior system that never saw wider use. This is an example of the kind of technology I think will be necessary to carry out the task of building socialism in the 21st century.

More or less, computer architecture needs a revolution in the same way that political economy does. I hope any of that made sense lol. Thank you for reading. Oh, let me know if any of it was unclear or if you want me to elaborate. I'm not so good at writing lol.