Uh, skribbl.io is definitely not a Scrabble clone lol (And also not really a 1v1 game)

Especially considering that these models seem to more or less just be quantized variants of Qwen3 with custom kernels and other inference optimizations (?) rather than fine tuned or trained from scratch with a new architecture, I am very surprised (or suspicious rather) that they didn't do the obvious comparison with a quantized Qwen3.

Their (to my knowledge) new measure/definition of intelligence seems reasonable, but introducing something like this without thorough benchmarking + model comparison is even more of a red flag to me.

[0] https://github.com/PrismML-Eng/Bonsai-demo/blob/main/1-bit-b...

Re: anomalies - I think this is just unavoidable in any sort of category system, and I don't have a good solution for it except to consider frequency and severity.

Just in case you're referring to Zhang Shan winning Gold in 1992: the decision to bar women from competing in the 1996 Olympics was made before Zhang had won her medal. [0]

> Until men with genetic anomalies are equally banned from sports (for example, being an outlier in height for basketball)

We don't have height categories, we have categories based on sex. We have categories based on sex because there are physical difference caused by difference in sex that lead to advantages in sports competitions. As such, people who have physical advantages over others based on their difference in sex (e.g. going through male puberty vs. female puberty) shouldn't be able to compete in the category created to protect participants from precisely those differences.

[0] https://en.wikipedia.org/wiki/Zhang_Shan#cite_ref-nyt_4-0

I agree with the sentiment, and many talks do this really badly ("Here is our outline, we start with an introduction, and end with a summary"), but it is worth mentioning that the alternative isn't no structure at all, but trying to convey a bigger picture to your audience for them to anchor each section in once you actually start your talk. This could be done like the OP suggests ("Just tell people the key idea upfront"), but there are other ways: instead of telling people the end result, tell them the question you set out to answer, and present your talk as this journey. look at the same thing/topic through different lenses/perspectives. Present a rough outline of a proof you are going to go through, or a case study you are about to present before going through the details sequentially.

> How does one become a good collaborator? The golden rule: Do not block.

Not only is this great advice for effective collaboration, it is also a very nice habit to have in any place where people's impression of your ability determines your future (career) trajectory

https://news.ycombinator.com/item?id=47420619

https://github.com/robotocore/robotocore

I'm a director at the Max Planck Institute for Intelligent Systems. Prior to joining the institute, I was Associate Professor for Electrical Engineering and Computer Sciences at the University of California, Berkeley. My research contributes to the scientific foundations of machine learning and algorithmic decision making with a focus on social questions.[0]

Also simply knowing of him doesn't answer the question.

[0] https://mrtz.org/

I do like the idea of saving prompts for projects like these (Which is also where the above question is answered: "Creating an MIT-licensed wrapper around Moto that has 100% feature parity with Localstack." [0] Which (i assume) is motivated by the recent changes to Localstack's distribution model [1])

[0] https://github.com/robotocore/robotocore/blob/main/prompts/2...

[1] https://blog.localstack.cloud/the-road-ahead-for-localstack/

If you just rerun and don't go to find out what exactly caused CI to fail, you end up at the author's conclusion:

> (but it could also just have been flaky again).

We already mostly know what makes people happy/healthy: personal connections, physical activity, healthy diet and some sort of purpose/goal in life that goes beyond day-to-day activities. The problem is that these things generally require (hard) work and can be unpleasant sometimes, so humans do what humans do and spend unreasonable amounts of time doing the more pleasant things such as reading and gathering info rather than applying these and what they already know. (That's not to say that a project like this can't be fun or lead to insights, especially across longer time spans, but i feel like all of the questions in the first paragraph have fairly obvious answers if you know yourself at all, that don't require extensive tracking of stats to get)

I agree that the delays are unacceptable, but the official app is great w/ digital tickets + seat registration, you don't need the PDF at all (it's even optional during checkout, so if you don't like them you can just uncheck the box lol)

The other points were always true for hackathons (For some more than others, depending on judges/audience), even before AI coding was a thing.

> I've been working on this since last September at a slow burn (no code reused for HackEurope though) and the goal is to have a running startup by May.

If anything, this is pretty much the opposite of what a hackathon is supposed to be: A place where you meet people you might not even know, come up with an idea on the spot and develop an MVP + pitch it on a tight (time) budget. Taking an idea you've already been working on for months and using it for a hackathon submission feels... odd

> A solid 90% of the projects there were just vibe coded slop. Even the ideas were AI. You can tell when multiple people implemented the exact same idea with the exact same title, description, and implementation.

The first is probably true, but to really judge the impact of it (Did AI generated ideas actually win?) we'd have to see the results

> A lot of cool ideas are out of distribution from the training data, and those rarely show up at hackathons anymore. The AI says they're "too hard" and people simply avoid these.

Probably true, but again, not a new phenomenon.