aoeusnth1

Your comment assumes that security issues are only discovered because someone downloads the package. This premise is incorrect.

Maybe they're slow rolling the announcements to be in the news more

Their scores on SWE bench are very close because the benchmark is nearly saturated. Gemini 3 beats Sonnet 4.5 on TerminalBench 2.0 by a nice margin (54% vs. 43%), which is also agentic coding (CLI instead of python).

Which app are you referring to?

This seems like a dubious conclusion. I think you missed this part:

> in accordance with Google’s relevant terms of service, privacy policy

16M = 10^7.2 ~= ↑7

No, it's extremely easy and routine to fix cell phones. You could train teenagers to do it.

Others would beg to disagree that we should be build a machine which can act as a human.

I think the before and after screenshots are pretty stark.

So 0.9T of interest and 1.9T of new debt overlap. You can't add them together to get 6.2% of GDP because interest is a subset of new debt - you're double counting.

Future corruption is less likely if punishment is expected. Not that complicated.

Not feasible.

You don’t own your marriage either? What exactly is the distinction you’re trying to make here, that you can hang onto your marriage even if your spouse doesn’t want it?

You used the word reinforcing, and then asserted there's no reward function. Can you explain how it's possible to perform RL without a reward function, and how the LLM training process maps to that?

Should companies never pay dividends?

If training is cheaper, then costs go down and models become both cheaper and more profitable. Training be scaled up to make the models smarter and larger. Distillation can bring those benefits to smaller models.

Training is a pure cost - inference is wildly profitable if you ignore training. Reducing training costs improves the economics of LLMs significantly.

Statistically, senior engineers are less likely to accept AI suggestions compared to juniors. This is just a tidbit of supporting evidence for the suggestion that juniors are not properly reading and criticizing the ai output.

All of the top upvoted comments are calling or implying this is a bubble. It seems to me the majority loves to imagine themselves persecuted in their home turf.

That's not a stable equilibrium if the hits have a large enough effect with respect to the movement of the balls. The internal circle will create disturbances against both sides of the inner circle, but the outer ball will have to travel a longer distance to move from one side to the other to counter them.

Wow, look at the crowd of NN doubters in the comments there. I see the quality of foresight in the commentariat hasn’t improved given the state of this thread, either.

Does a 20% decrease result in a -50% discount? Why would it be nonlinear?

Yes, Google uses trunk based development for almost everything. Even Android is now on trunk based development (previously a branchy mess) and heavily uses sets of feature flags for promotion / demotion through quality lifecycles.

I’m not sure why TBD should have anything to do with code reviews. You aren’t able to submit to main without first uploading a change or stack of changes to code reviews, then you submit the changes after they are approved. That has absolutely nothing to do with your branching model. Just think of it as a rebase request instead of a pull request. PR was always the wrong terminology anyway.

I don’t understand why PRs are being discussed here at all – doesn’t anyone serious use trunk based development with code reviews? How is that different from a PR? Seems like we need to disentangle the discussion of whether code review is good from whether feature branches vs trunk based dev are good. Those are completely unrelated questions to me.

I think there are a lot of interesting questions about using feature flags (a baby branch) vs actual branches. Personally I’m pro flags and anti branches, after a lot of experience in developer tools and CI.

When forced to point to a concrete issue by tooling, I often just end up rewording it - "Ideally this should X, but it Y". Comments are fine, lower overhead and don't require re-triaging later, and have all the context to immediately understand.

Yet, richer people have fewer children. It isn't clear that reducing costs (making people richer) would actually increase birth rates

Probably because lichess.org exists?

The standard model is computable, so no. Physical law does not allow for non-computable behavior.

Can you explain in what way the incentives are misaligned here? Or did that mis- prefix slip in there because of how you feel overall about the law?

Are you intentionally sandbagging the LLMs to prove a point, or do you really think 4o-mini is good enough for programming?

Even 2.5 flash easily gets this https://imgur.com/a/OfW30eL

No, it's always strictly worse because you could have bought bonds or deployed the capital in some other way with that money.