People who think 60% is too high haven’t been to any mid-sized business in a while, I think.

The high of cracking an intractable problem (pre-AI) was just unreal. I’d struggle to sleep, awaiting the morning so I could get back at it.

It’s the challenge of it though.

We’ve been doing it for a long time. Linus is from Finland.

I’m going to get downvoted to oblivion, I’m fully aware.

There are many things I don’t know anything about. But I do know a fair bit about this area.

And with great respect to the worriers, this just will not happen.

The investment team that’s bought EA have many other “western” investments they’ve never interfered with, even though they do a great deal of things that the orthodoxy in Saudi Arabia would disagree with. The layers between “what I [claim] to believe” and “make me the most amount of money” are many and includes people from all over the world. If anyone thinks Saudi royalty will sit and make determinations about what games EA will ship they can stop worrying right now. The Saudi monarchy is worrying about two things only: As much money as possible and staying in power.

This is an investments like so many others. It will be ordered to profit-maximise - and if that requires one-upping Doom in gore, and shipping a new edition of Gay Paradise 2000 every 6 months - they will do exactly that.

This is a complete storm in a tea cup.

The function is deterministic, agreed. The implementation is almost always not. Hardware floating point addition is not associative. So a GPU kernel that splits a reduction differently (ie interleaving it with anything else, like running your graphics, or sharing your work with other users on the same hardware) will produce different results over different runs even at temperature 0.0.

Determinism is almost always impossible when dealing with floating point on a multi-process/multi-user system.

There are attempts to create batch invariant language models (https://github.com/thinking-machines-lab/batch_invariant_ops) but all the major ones are not.

No. Floating point arithmetic and ordering of operations won’t make 0.0 deterministic.

Exactly this. If you made an LLM that had deterministic code output then I’m all up for saying “AI is a compiler for human language”. But until then AI is most definitely not a compiler.

95% AI text agree. It reeks.

Ok, good to know. Now please relate your statement to TFA.

Also, it’s possible to have nuclear running with very little hot water discharge, you just need to build more cooling towers. Where running water exists, it isn’t seen as necessary.

Guys, guys, it’s not an OR, it’s an AND.

We need both nuclear AND renewables.

mcedit is fine. I personally find micro the perfect blend of features and simplicity.

Obviously Google install their own connectivity.

I seriously doubt 75% of password strength checkers for that very reason.

To be fair, having gone through a couple of GDPR transitions, deleted in reality is often a full row in the PII table. All your actions and behaviour are still logged, there’s just longer any way to tie your ID back to PII.

(I’m not a US citizen; I believe guns should be controlled and there’s no reason for a citizen to own a gun)

Don’t US school run shooter exercises? Darwin, even here in the UK where I live, the secondary school (12-16) ran twice yearly active shooter exercises - although they called them something else, like “lockdown exercise” and we’ve only ever had one school shooting (1996).

My point being that surely any exercise would train the whole response system, including pepper spray drones (still can’t quite believe I’m typing this - anything to avoid actual gun control, I guess)?

Feel like we’re missing an overflow joke here!

Or something like: // initialise a glass using max water value as indicator that glass is unused

I’m not a US citizen but do you have any rights at a border checkpoint? I was under the impression that no warrant is necessary, nor lawyer required, to answer questions.

I’ve worked in many large enterprises. They always cut too deep - thats the tactic: Cut more than needed to remove the “fat” (in the eyes of the decision holders) then rehire a bit where you’re not holding up.

In this case it’s two steps back and one step forward. Net, it’s still a reduction of the org size.

Lots of people leave the tag in for transparency. The world must feel like such a conspiracy for you.