sambaumann

Even after the June changes there was some allowance to use agent SDK on the pro plan. This will move me to codex tomorrow if agent SDK is really blocked on pro

This reminds me of elevator saga (https://play.elevatorsaga.com/) - little programming game where you have to program elevator schedules.

This works if only 'trusted' actors have access to frontier class models that can search for vulnerabilities. With a near-frontier model available with open weights then attackers will be able to do plenty of damage even with 'fast remediation'

The CursorBench plot, for example, shows that fable does have slightly better performance, but Opus is pretty close, and is less expensive per task

In their video today, they said they asked Valve the original pricing, and they said (paraphrasing) "we can't tell you exactly - but the increase we recently had on the steam deck is about how much the pricing for machine increased" - which is how they came up with the $800 number

We did this once at a traditional hackathon - we took a food dehydrator and wired up a temp sensor and raspberry pi, wrote PID controls and turned it into a 3D printer filament dryer.

There's a big difference between a well considered high quality contribution made using AI as a tool and "claude find and fix an issue in this repo so I can put on my resume that I contributed to a high profile project". The problem is that it takes considerable time of the reviewers to filter out the high quality contributions vs the low quality ones.

I do think closing off contributions is a big step and would rather most projects find a middle ground, but it's definitely understandable why they did this.

This is what I do every time I want to read an article.

I still get "this model requires macOS" when trying to pull that one

I also do this, but I find the terminal ergonomics are a little awkward on mobile. I do think termius is the best version of it I've used though

Cool!

FYI the arrows on both photos only actually control the top photo.

Looks like this was a newsletter by the author, not a blogpost

Couldn't you just spend those $100 on claude code credits yourself and make sure you're not shipping insecure software? Security by obscurity is not the correct model (IMO)

Related: https://www.jbecker.dev/research/prediction-market-microstru... (previous discussion: https://news.ycombinator.com/item?id=46680515)

Sqlite’s source code is public domain. Surely if you dropped the sqlite source code into Linux, it wouldn’t suddenly become GPL code? I’m not sure how it works

This article is convincing, but ultimately still no true evidence, it's all circumstantial.

After reading this, Back does seem like a pretty likely candidate, but maybe you could run the same kind of investigation on every other candidate and find similar matches. The filters they used for the text analysis did seem pretty arbitrary to match up with Back's language

This video is linked in the article

So is it an actual moving bus or just a simulation of one? I have not heard of the concept before

Yeah, but all the models live in chatGPT have reasoning (iirc) - they could use reasoning tokens to do the 'compute', and still show the user a succinct response that directly answers the query

because for API users they get to charge for 3x the tokens for the same requests

I feel like this has gotten much worse since they were introduced. I guess they're optimizing for verbosity in training so they can charge for more tokens. It makes chat interfaces much harder to use IMO.

I tried using a custom instruction in chatGPT to make responses shorter but I found the output was often nonsensical when I did this

For reference when this was written in 2018, a P/E of the S&P of 24 was considered inflated. It stands at 29.34 as of now.

I didn't have a ton of time to play with this but on first try this is awesome. Looking forward to spending more time playing with it

10c yes seems really high for a tie. NCAA rules don't allow for ties in football. I know prediction markets have very long shot bets but I would expect that to be closer to 1c

Edit: it looks like the tie market is only for if the game is tied at halftime, which makes much more sense

I paused and wrote out all the probabilities and saw no way to improve beyond 80% - I scrolled down hoping to be proven wrong!

Turns out, at least in my area, for the grocery items you need to buy at least $25 worth to qualify for the free shipping.

I also don't agree with the argument you replied to, but a counter-argument to your point is that we don't mandate individuals to wear name tags while in public

been using vim for years, just did the tutorial and learned several things I did not know

According to the status page services are being restored