knowaveragejoe

tiktok videos themselves are often just ads.

Same.

How are you using Claude Code with the $20/mo plan? Aren't you still paying the API prices on top of the $20/mo?

> This could then be combined with some software on the machine to turn a MacBook into a difficult to detect recording device, bypassing protections such as the microphone and camera privacy alerts, since the MacBook would be closed but not sleeping.

Isn't this already possible if the MB is connected to a power source like a portable battery?

llama2 is pretty old. ollama also defaults to rather poor quantizations when using just the base model name like that - I believe that translates to llama2:Q_4_M which is a fairly weak quantization(fast, but you lose some smarts)

My suggestion would be one of the gemma3 models:

https://ollama.com/library/gemma3/tags

Picking one where the size is < your VRAM(or, memory if without a dedicated GPU) is a good rule of thumb. But you can always do more with less if you get into the settings for Ollama(or other tools like it).

With chutes.ai, where do you see a one-time $5 for 200 requests/day?

It's unintuitive how, but it probably would cause prices to go haywire.

> METR also conducted an in-depth study that asked experienced developers to code with a popular AI assistant. After they finished their tasks, the developers claimed that using the AI had made them 20 percent more productive. But independent evaluators in the study actually concluded that using AI did the opposite: it increased task completion time by about 20 percent.

This somewhat reflects my experience... I can totally see the back-and-forth dance taking longer in some cases.

But I also think there is more than efficiency being unlocked here. Sure, a developer might have cranked out a rough MVP in less time, but with this they're also often continuously updating a README, tests and other infrastructure as they go.

One could argue about whether that's its own footgun - relying on tests that don't really test what they should, and let critical bugs through later.

If you have a HuggingFace account, you can specify the hardware you have and it will show on any given model's page what you can run.

To steelman the crypto side: I think the stuff that is most interesting is found _around_ crypto and not necessarily within most cryptocurrency projects.

For example, there have been a lot of novel or interesting projects centered around cryptography, consensus, decentralization, anonyimity. On a sociological level, there are a lot of interesting social/economic experiments unfolding in the DeFi world as we speak.

But I think that is where the truly interesting or valuable contributions to humanity as a whole end, and the vast, vast majority of it otherwise is just speculation or scams. It is hard to find that signal in the noise, but it is there, if you look.

Small nitpick - the axes on the varying alloy proportions graph say "Sonnet 2.5" and "Gemini 4.0"

How can you look at this and think "druggies"?

> similarly doesn't work with ollama since they need to bump their llama.cpp runtime

Just curious, how frequently does that happen?

Modern LLMs frequently do state that they "don't know", for what it's worth. Like everything, it highly depends on the question.

I'd love a kit car that wasn't ridiculously expensive and/or designed to be raced. But I can understand that the kind of people who would invest the time in building a car would be okay with those propositions.

I wonder if there's a market for building something purely utilitarian, like a little hatchback or something, as a kit vehicle - with the express purpose of learning a lot of automotive principles along the way.

I've found the best knowledge on that front are youtubers who buy beater old cars, get them running again and maybe even restore them to resell at a profit.

Their videos are between 20-60 minutes and run through the general process of going from no-crank-no-start to running, which is basically the same for all gas-powered cars. Diesels are a little different.

You also need air and compression! And if you want the engine to run longer than a few minutes, cooling.

For what it's worth, headlight alignment is not something a lot of people think about or even realize is a thing - it's just changing bulbs, right?

With any DIY car repair, you always run the risk of things like this, where you don't-know-what-you-don't-know. But it's still worth the ride and the lesson, imo.

Who is they? Why would they "push for" that?

Why can't we get these in the US? This kind of simple utilitarian vehicle is exactly what I'm looking for.

> You should also consider normalizing the number of injunctions to the number of executive orders per term.

That paints an even more stark picture for the Trump administration.

It was never "always listen to the experts", but that's the strawman given by the contrarians who have decided we need to throw out all expertise.

What is `jinja` in this context?

The venv is created and then discarded once the script finishes execution. This is well suited to one-off scripts like what is demonstrated in the article.

In a larger project you can manage venvs like this using `uv venv`, where you end up with a familiar .venv folder.

Ah yes, it's a conspiracy and you spotted the critical oversight on their part. Real astute.

It's not actually that unlikely for other people to possess the name.

This is a really nifty CLI tool that lets you fit an LLM into a shell environment. Check out examples like piping context into and out of the `llm` command.

https://github.com/simonw/llm

What was Software 2.0?

It's not a lie, it's pretty close to the truth. The vast majority of immigration is a net benefit

Yes, that was my point.

This reads like post-hoc justification for what's going on, based on a tangled web of misconceptions about publicly funded medical research. The therapies in the article originated from government funded research that took decades to mature. "Huge swathes" of this kind of research are not plagued by reproducibility issues.