I wonder if it was a veiled reference to the debate between Linus and Andrew Tanenbaum, who wrote MINIX based on OS research that, according to him, showed that microkernels are better than monolithic kernels.

https://en.wikipedia.org/wiki/Tanenbaum%E2%80%93Torvalds_deb...

In a funny twist of history MINIX is running at ring -3 on most modern Intel CPUs. So it's probably running on more desktops than MacOS or Linux.

Here's the mailing list entry referenced: http://lkml.iu.edu/hypermail/linux/kernel/9906.0/0746.html

My point was that as a newcomer in either operating systems (Linux late last century) or a new search engine (Marginalia, Kagi) you cannot always win by repeating the same stuff incumbents do.

Especially not when the incumbents are busy digging deeper holes.

Yeah I'm very much betting on the null hypothesis when it comes to fancy machine learning-based methods and implementing the most bog-standard keyword search you could imagine.

It's not perfect, but it's scary how well it works sometimes.

I assume they're talking about all the OS research done at places like XEROX and Bell Labs that never really came to fruition, but formed the basis for modern OS's.