Why would you need a wrapper, of a wrapper?

Researchers, at the level that he was (note in terms on Turing or Nobel - but in terms of org underneath you) are not only supposed to do pure research, but also to help the company who pays your fat salary, to find a way to use the fruits of such research in products that are able to generate revenue.
That was the failure to launch ,,,

I bet the reason of his "retirement" is due to things getting hot within Google, with management asking the research branch why, with the Transformer architecture being out since 2017, it was left to OpenAI to capitalize years later.

Too little, too late.
The world has already moved on to different frameworks.

Yeah, let's give him 5 years until he actually understands AI.

The Transformer architecture came from Google way before OpenAI did ChatGPT.
True.

I am not sure Google held back in releasing big LLMs, I believe it is more of Google Research failed to push models to the scale (in terms of parameters and input tokens) that OpenAI did.
Sometime you observe a ML model and it seems to have hit an asymptote in terms of performance metrics, while if you push forward that, more gains can be achieved.

I am not sure what Linus had in his head, when he decided to slap in a language du-jour inside the kernel.
When, soon, we will realize that is going nowhere, someone will have to clean the house of the added crud.

Cannot agree more!

I never understood the folks whining about ad tracking and the money Google (or Meta, or any other companies which provides stuff for free).
That Stuff costs money (you know, there are actually people working on it, data centers hosting the content, ...), and you either have to pay a subscription fee, or you accept targeted ads.
I much prefer the latter for one.

True that, but even a Unable to acquire target, storage space exhausted. Please insert a new floppy ... would be pretty bad :)

In my country (or better, my other country) cyclists swerve left and right mostly due to potholes and bad road conditions.
Of course, there are also the ones that happily sits in 2..3 lane wide chatting, but in my groups this is more rare, at least in heavy-traffic roads.
All the recent cyclists death I know of, are not due to them cutting through traffic, but cars running them over (especially at roundabouts) because they simply do not notice them.
I would definitely use such device if available ...

Linus better talk about things he understands. A spinlock maybe.
Machine Learning is not within the grasp of someone who only did system programming his whole life.

I had been using Linux for decades, but in the last 8 years or so I started using OSX (MacBook Pro) with Parallels VMs.
It was nice since it integrated with the other Apple devices I have (iPhone & iPad), from moving things around, and especially making/taking/transferring phone calls among them.
With the M* chip migration I was not able to decently run x86 software (thanks Vivado!) within the Linux VM (yeah, Rosetta is fast only in ads benchmarks blog posts), so I went back to an x86 workstation running Linux.

Plan 9 was something which set up to justify itself by making things different.
It didn't matter that it was actually better, but it had to be different.
The everything-is-a-file and pass-everything-as-string was brought to the limit, and used it even in cases where it really did not fit.
Performance was terrible when compare to Linux in doing apple-to-apple comparisons.

Plan 9 at the end deserves to be where it is now ... the graveyard.

Code does not need to be changed only when there are bugfixes.
The most frequent changes happen when there are internal code refactoring, and developers really enjoy go fixing dead (or on life support at the least) code.

Yeah, a git-blame with my name will likely pop up a few thousands LOC...