HN user

woggy

199 karma
Posts3
Comments68
View on HN

Asking it to apply the YAGNI principle also sems to work well for trimming codebases down. Generally ask it to review, generate a list of review points, then we go through each one together and I make a decision yes/no on each one (or suggest further modifications).

Straight from IDF soliders:

One incident did leave a scar. It happened at an army outpost in southern Gaza. "I was sitting there in the command room," Maya says. "Suddenly, the soldiers on watch noticed five Palestinians crossing the line they weren't allowed to cross, heading for northern Gaza.

"Everybody went crazy. There was a big mess. The battalion commander gave an order to overwhelm them with fire, even though they hadn't been confirmed as armed or anything like that. A tank started shooting at them with its machine gun. Hundreds of bullets."

She says four of the five Palestinians were killed. "A few hours later, a D9 [Caterpillar armored bulldozer] buried them in the sand. When I asked why, they said that it was so the dogs wouldn't eat them and spread disease. The one who survived was put in a cage at the outpost, and they said we had to wait for a Shin Bet man to interrogate him."

The Palestinian sat there, cuffed and blindfolded, and seemed to be freezing from the cold. "Suddenly, one of the soldiers took out his penis and started pissing on him. He told him, 'This is for Be'eri, you asshole, this is for Nova'"– Kibbutz Be'eri and the Nova music festival, two of the sites attacked by Hamas on October 7, 2023. "Nobody could stop laughing. I might have laughed too."

Very interesting, seems like a good framework to test and experiment with memory. I am curious why it wasn't able to solve it considering it is a well known game. Would be interesting if puzzle games like this could be generated so we know it's not already been trained on it.

I wonder if the improvements due to different memory system approaches apply in a similar way to tasks that are in its training history vs those that are not.

I've written a few terminal tools on top of Roslyn to assist Claude in code analysis for C# code. Obviously the tools are also written with the help of Claude. Worked quite well.

I was thinking about this the other day. If we did a plot of 'model ability' vs 'computational resources' what kind of relationship would we see? Is the improvement due to algorithmic improvements or just more and more hardware?

What's interesting to me is how little this whole situation is being talked about here. Not a single topic on the front page while ICE agents are routinely pulling people (including US citizens) out of cars, breaking into houses and shooting people on the street.