HappySweeney

The hacky string matching is what I started with, but there was always new ghost text. I turned suggested prompts off which helped, but idle hooks plus ensuring the terminal doesn't change over a half second has done the trick so far, and I don't have to keep chasing new ghost text.

I have a more primitive setup where I use Gemini, Claude, and Codex at the same time and cultivate expert contexts. Each of them lives in their own tmux session so they can read each other's terminals and prompt each other. One issue that I keep stamping out is that another instance will paste a prompt and hit enter in the middle of my typing.

this site has some popup that hijacks the page and tries to trick you into installing an antivirus with fake infection reports. Closing that popup sends you to walmart.com

In my experience, audit findings decrease in frequency as a codebase matures. Was Fable doing greenfield work?

They measured density, not the probability of getting one.

A bunch of links open a hardcoded browser (like help links within Visual Studio). Executives have OKRs to meet.

I find that when an LLM jumps into tasks it was not told to do (or even worse, doing things it was explicitly told not to), it is a good sign the context is too full, and you should do a controlled hand-off to a new instance.

I have an optane and lots of ram, so I tried full-fat models for writing some function overnight, as I get about 0.7 t/s. My current go-to test is to update a scalar function to transpose a bit-matrix to one using avx512. the cloud models all play with that like its nothing. Kimi 2.6 and GLM 5.1 both failed miserably.

I think its common to develop an adversarial-collaborative approach to getting some semblance of quality out of AI. I personally favour using multiple models for different roles, having a bunch of continuity documentation maintained, and having the plan surface human-verifiable deliverables as soon as feasible. It does involve more attention than most people would tolerate probably.

Code review is the main thing I use LLMs for. I have found it to be remarkably candid when you tell it the code came from another LLM (even name it). I was running Kimi K2.6 Q4 locally, seeing if it could SIMD a bit-matrix transpose function, and it was slow enough that I would paste its thinking into Gemini every few minutes. Gemini was savage.

They are redeveloping much of that electron-based code back to Win32 as part of their quality improvement initiative.

Haven't there already been a few cases, each of which found that mechanically-produced works are not copywritable?

If I had to guess it would be "campaign contributions".

Are buckyballs organic?

Agreed. I believe it's well within Google's ability to auto-edit the sponsored segments out within an acceptable error margin.

fwiw Penn Gillette no longer calls himself a libertarian.

If you don't exclude your build folders from the scan it will slow everything down tremendously.

Don't legitimate IQ tests top out at 160 for adults?

Do investors ever say "ok, time to panic"? Aren't they always just going to shill their bags? Why should we listen to them?

It's in the article. I'm happy to read the bill and watch your linked video once my workday is over.

In the sense that you can sue to have the order challenged?

That's a good question. The article doesn't say and I haven't read the bill.

Were the bank accounts seized before or after the conviction?

Before, of course. That was one of the justifications for invoking the emergency powers, and it wouldn't have been controversial otherwise. This is a digression, though, as there is no mention of any legislative changes to bank account seizures in the article.

While I agree with most of this and oppose this bill, your last two lines are a mischaracterization. There is judicial oversight, but only after the order is implemented. Second, the bank accounts seized did not belong to protestors, as the leaders of that siege were convicted of mischief, two of which are being sentenced today. In general, protests do not engage in torturing the local populace with 95db of air horn for 16 to 20 hours a day. The account seizure also required emergency powers.

Headline buries the lede, which is that the DOJ shut it down for being a "deep state" probe. Basically, officials working for this administration can now legally receive bribes.

I bought one of the Iocrest ones from Aliexpress so I didn't block my gpu's intake, with the AQC113 chip, and it worked perfectly right out of the box - much better than the Asus XG-C100C which made me configure it to get max 2.5gbps. I thought it was windows. Note it gets super hot (like 90°C even with the heatsink).

I would caution applying our current understanding and limitations onto alien civilizations. I don't believe they will break the light barrier, but they may develop telescopes that can detect life from light-years away, and the stupendous travel times are possible with suspended animation or by stopping the aging process.

I'm a bit surprised the entire thing isn't in a faraday cage.

The stable isotope of gold is produced by the collision of two neutron stars, which is unlikely to happen in our stellar vicinity any time soon.

There is no way this is a coincidence.

To me it appears to be a way to pay his lenders without liquidating or collateralizing more Tesla stock.

It's so space can expand without increasing the Planck length