amoss

On point 4, I don't use zig but a two second google leads to http://ratfactor.com/zig/stdlib-browseable2/math/atan.zig.ht...

I wonder about the accuracy of the rest of the article.

It reminds me of the Hamming quote:

"I noticed the following facts about people who work with the door open or the door closed. I notice that if you have the door to your office closed, you get more work done today and tomorrow, and you are more productive than most. But 10 years later somehow you don't know quite know what problems are worth working on; all the hard work you do is sort of tangential in importance. He who works with the door open gets all kinds of interruptions, but he also occasionally gets clues as to what the world is and what might be important."

Each of those little interruptions is a clue about the wider state of the world (codebase / APIs etc). AI offers a shortcut, but it does not provide the same mental world-building.

Mistake in headline: tge article says 54% rate on China. 34 looks like a typo.

Nuclear weapons are not self-improving or self-replicating.

I can't remember the last time I worked on a team without automatic linting enforced on check-in. Why would people choose to waste time arguing over that can be automated?

thanks, i'll try that Nd see if it helps.

> A limitation I see with AI for coding is that your problem must be mainstream to get decent results.

Or to phrase it another way, there must examples of the technique that you want to "generate" in the training set otherwise the horrifically overfitted model cannot work.

Probably a little bit too cynical but this is my experience asking LLMs "unusual" questions.

Nothing in the kernel log and the steam log shows a trace, but no debug symbols, only raw addresses.

Wayland has been "nearly ready" for so long that it is easy to miss progress. I installed an arch desktop three months ago and setup kde with wayland, expecting that I would revert to xorg when something broke.

So far everything seems to work - although the machine has a strange glitch. Processes randomly die, but only if they have a render surface. Typically it happens to factorio or to the steam ui component, which are both incredibly stable.

interesting take.

the aaa games industry is struggling (e.g. look at the profit warnings, share price drops and studio closures) specifically because people are doing that en masse.

but those 90s games are not old - retro has become a movement within gaming and there is a whole cottage industry of "indie" games building that aesthetic because it is cheap and fun.

We prefer to call it "engineering"

In a different place a long time ago we were approached by a company that wanted a collaboration. They had a very specific project in mind, with requirements already set. It was a humanitarian project to "find earthquake survivors". If you read the requirements in a very positive way it was an ethically clean project.

But the company is a major defense contractor and it was very unusual for them to be involved in humanitarian work. So it made sense to question the requirements after a less generous reading. Detecting mostly occluded humans in a highly cluttered and destroyed urban environment has other use cases. They got as far as "highlighting the survivors with a laser designator" before I started to ask some very pointed questions and start discussing the ethical considerations of us helping to build a system to kill people.

I was asked to leave.

After that meeting the potential collaboration was rejected and it never came up again. The company continued shopping around other places to get help building their "earthquake rescue system".

Never get distracted. Always question the purpose of work. Consider ethics.

Getting another job can be a pain. Living with knowing what you helped to build would be much harder.

return false

My heart goes out the cryptographers. All that code, written over decades, that assumes coin flips are 50:50. So much updating and rewriting to do. Quite a few algorithms that will need a rethink to remain fair.

Ah! That makes much more sense now :) Thanks for the link

Who is Barbara?

Publish or perish made manifest.

If you solve chess then you have a tree that is too large for us to currently compute (about 10^80 although my memory may be way off). Annotating that tree with win / loss / draw would allow an optimal player without search. The two obvious approaches to compression / optimization are to approximate the tree, and to approximate the annotations. How well those two approaches would work depends a lot on the structure of the tree.

This result seems to tell us less about the power of the training approach (in absolute terms) and more about how amenable the chess game tree is to those two approaches (in relative terms). What I would take away is that a reasonable approximation of that tree can be made in 270M words of data.

Is the continuation not an activity in the present time, or that starts in the present time? Characterizing this as being in the futures seems to be an incorrect boundary case.

"Keep at it!", "Hold the line!" appear to be orders to be doing something right now.

ah ok, so it would not be pointer alignment inside the pages but instead the assumption that page +4k is a page.

I'm missing something here. Assuming there are pages at 0k, 16k, 32k etc - all of those pages are aligned on 4k boundaries as 4k > 16k. So code written with the assumption that its pages are 4k aligned should have that assumption met when running with 16k pages. It is still early here and I have only had one cup of coffee. Am I misunderstanding something really obvious?

I'm slightly confused after reading about page alignment. Why would a 16k page be less aligned than a 4k page causing assumptions about pointers within those pages to break? The 4k pages on x86 are aligned on 4k boundaries, are the 16k pages on M1 aligned on <4k boundaries?

I wonder how model collapse would apply to the AIs created by applying the results of AI-generated papers?

One easy workaround here is to keep the hardware modern, but install Teams on it. Then you can feel the ancient hardware vibrate for all other applications.

They have played us for absolute fools.

Part of the build up said that they had deliberately weakened / thinned some of the tiles in order to test what the tolerance was. It seems that they must have gotten some incredible data about the mode of failure.

That was my initial reaction, the word scam is overused these days. But then I reached the comparison section between the photoshopped/ generated images and the real photos.