[Cialug] 10yr future of Linux and open source

[Nicolai]

It's fun to write down predictions and a good learning experience to
revisit them later. What are your predictions for Linux, open source,
and the tech world in the year 2035?

I think "Linux on the desktop" will continue growing and at a slightly
elevated pace, maybe even reaching 10% marketshare by 2035. That said, I
believe that if the community popularized a concerted effort to displace
closed systems, Linux COULD become a leading desktopish OS (along with
Windows, iOS, macOS, Android). Ubuntu and some other distros have been
sufficiently user friendly for years, and I think Linux could have much
more marketshare than it does. The days of struggling with xfree86 are
long gone. Whenever I see a Windows system now the navigation is
completely bewildering and apparently there are even ads now? If you
want to popularize Linux on the desktop, you have everything you need
today.

I think systemd has won and more distros will adopt it, and some that
don't will come to an end. Developers will increasingly write code
dependent on systemd, being therefore less portable, requiring other
operating systems to provide expensive workarounds. This will harm the
ethos of open source making it a little more superficial (a general
trend unfortunately).

Currently, we're on the verge of getting 192 core AMD Epyc CPUs running
at up to 3.7GHz. What will be available in 2035? What if this is a
standard high-end consumer workstation in 2035, with high-end servers
doing kilo-core computing? What will people do with unlimited power?

RISC-V seems to be advancing while ARM64 hasn't progressed as much as I
thought it would. Only Apple is doing interesting work there and it's a
closed system. Asahi Linux has done miraculously cool work on the
M-series CPUs, allowing other operating systems to follow in their
footsteps, but IIRC they've lost a gifted developer who did most of the
work. And if only one computer maker (say, Gateway 2000) had seriously
produced x86 machines back in the day then x86 would have stagnated,
which is where we are with ARM64. Currently, most ARM "boards" require
some janky work to get them going and the ecosystem doesn't favor people
who want to ice-skate uphill. I used to think ARM64 would be the next
dominant platform but now I think it will mainly pressure AMD/Intel to
advance PPW (performance per watt). So maybe open-source RISC-V could be
in 2035 where ARM was around 2020, with the potential of being the next
great thing. Hope so but I doubt it, as RISC-V seems also to be building
a culture of tinkering with SBCs. That's cool but not for 99.9% of
computer users.

I want to believe Ladybird will grow and succeed. https://ladybird.org

Memory safe PLs will continue to make agonizingly slow progress. I hope
"Rust in Linux" will try again. I think memory safe languages will need
to take over userspace before operating systems make significant moves
in the kernel. But that seems to be slowly happening. One of the
blockers in OpenBSD is the fact that Rust doesn't support nearly all of
the 14 hardware platforms, but I'd be surprised if OpenBSD still
supports Alpha (among others) in 2035, and luna88k is supported entirely
by a single developer who is almost certainly its only user in the
entire world. Super impressive, but not what creates marketshare.
There's a broad trend to shrink hardware platform support and I don't
see Rust expanding to obscure platforms like landisk and hppa etc.

Thoughts? Go ahead, share your own 2035 hopes and predictions.

Nicolai