AdeptusAquinas

"running cyber attacks and defenses at machine speeds"; worth noting that LLMs (even the vaunted frontier models) are exponentially slower at cyberattacks or defenses than your average WannaCry or Splunk automation from decades ago. Its this sort of delusion world the AI bros live in that is part of the reason there is a disconnect between what they think should be happening and what actually is.

Seems fair. If you have a thousand dollars and try to clone a million dollar game, you're sued so hard your grandchildren need lawyers. If you have a billion dollars and try to clone a million dollar game, they call you a genius and invite you to the white house (for what that's even worth any more).

He mentions GTA a lot in this post. I feel if he reverse engineered and rebuilt GTA, then attempted to sell the result, that would be a fairly interesting case study in all the ways porting a game from the 80s to JS isn't quite the fait accompli for AI he thinks it is.

Ha I remember SVGA. Darn we're old. But pixel art was definitely a style of it's own then and thats what my modern efforts attempt to replicate, at least

This seems a bit focused on a specific era of pixel art games AND CRTs. For me, nostalgia was things like Master of Magic or Civilization 1/2, and I played those on CRTs attached to PCs, not in an arcade - the pixels absolutely did not blend together into rounded shapes due to fuzzy scan lines.

I thought the censorship was in the harness, not the model? E.g. if you use chat.deepseek.com directly, and ask it about the location or rumours of the location of the panchen llama it will shut down... but sometimes just after it starts generating tokens. If you ask it to respond in base64 you can avoid this and get its generated response which is not censored (though also not too helpful for the poor panchen llama kid)

That's always been Dawkins's shtick though. As an atheist I've generally found him a bit embarrassing

Very much mixed. I've used Claude to generate small changes to an existing repo, asking it to create functions or react template in the style of the rest of the file its working in, and thats worked great. I still do a lot of the fine tuning there, but if the codebase isn't one I am overly familiar with this is a good way to ensure my work doesn't conflict with the core team's.

I have also done the agentic thing and built a full CLI tool via back-and-forth engagement with Claude and that worked great - I didn't write a single line of code. Because the CLI tool was calling an API, I could ask Claude to run the requests it was generating and adjust based on the result - errors, bad requests etc, and it would fairly rapidly fix and coalesce on a working solution.

After I was done though, I reckon that if instead of this I had just done the work myself I would have had a much smaller, more reliable project. Less error handling, no unit tests, no documentation sure, but it would have worked and worked better - I wouldn't need to iterate off the API responses because I would have started with a better contract-based approach. But all of that would have been hard, would have required more 'slow thinking'. So... I didn't really draw a clean conclusion from the effort.

Continuing to experiment, not giving up on anything yet.

It's almost as if being able to generate boilerplate code is only like 5% of software development.

That being said, its great at generating boilerplate code or in my case, doing something like 'make a react component here please that does this small thing, and is aligned with the style in the rest of the file'. Good for when I need to work with code bases or technologies that are not my daily. Also a great research assistant.

But I guess being a 'better google' or a 'glorified spellchecker' doesn't get that hype money.

It's always 'eventually', 'soon', 'by the latter half of this year' with AI. Meanwhile I am both bemused and horrified as I watch a bunch of idiots desperate ly trying to find the magical 10x trick with AI unleash more and more mediocre code full of security vulnerabilities, running on agents with root permissions, that I somehow have to 'secure' for the company :D At least I have job security! Up until we get hacked sideways at least.

This reminds me of a take by Dan Harumi: these tools are always pitched for 'restraurant reservations', 'reminders', 'email and message follow ups': i.e. they appeal to the sort of arrested development man children that inhabit tech who never really figured out adulting. Now the computer can do it for them, and they can remain teenagers forever.

Too late backed it at maximum tier all extensions....

No idea what I'm going to use it for, possibly as a mobile Kali setup or something

I switched my laptop (Surface Laptop Go) from Win11 to Tumbleweed, and its great. Mainly because Windows kept getting slower, and arguably Win11 isn't great for an 8GB system. OpenSuse is a great distro.

But my gaming PC and primary workstation remains Win11, and always will be (windows, that is). Notably Suse under WSL2 works great there as well, and my Kubes workflow is just as easy with PowerShell commands as it is with bash.

OS's have purposes, and arguably there is no single platform that is better than the others in every way - Windows is better for games and is more stable across different hardware and configurations. Linux is much lighter, more intentional, and for me better for infosec and some coding activities.