amoss

Explaining what a heuristic is does not address why one should be used in this case.

Why would the identity of the author relate in any way to the correctness of their claims?

The claims:

* The existing exploits don't work. * Reasoning chains producing an exploit have cheated the test.

If the claims are wrong then rebutting them is easy:

* Ask an LLM to produce a real exploit. * Test it.

Given the usefulness of this to anybody who knew about it between the ceremony and now - assume malice.

You have almost understood the joke.

what kind of a Culture would that lead to?

You seem to be confusing your quantifiers a little. (Not all SWEs must understand all CS) does not imply (there can exist parts of CS that no SWE understands).

> I think the next step is to demand that proofs either be human-scale or they prove that a human-scale proof is impossible and the machine proof is as good as it gets.

> I suspect that's possible without tripping over the halting problem. (But I can't prove it.)

I doubt that it is. For the language of proofs to be powerful enough to be able to express an arbitrary proof it would have be Turing Complete. Proving that a proof is the smallest proof of a given concept (i.e. there is no smaller human understandable proof) would then be proving the minimality of a program in a Turing Complete language.

oh no that sounds really "fake". nasty word. we are calling it "super intelligence" now. much better.

Keeping the tech proprietary so that it can only be used on these problems by internal teams is the very definition of gatekeeping.

If we want to call it what it is: Copyright Laundering.

Layering and placement of copyrighted materials to evade copyright.

No it does not, but it does imply that if they broken system was willing to pursue one person to his death then it should fairly uphold the same standard for executives who are doing the same thing.

Somewhat tiring that as alway any criticism is reduced to "but have you tried this on the latest model".

Sometimes this also happens because the design/spec never existed, but hopefully not in the industries under discussion.

In many codebases when you see this happening it is because someone reported a bug and the dev fixed the narrowest instance they could. Typically if(some specific case) do_fix() else do_normal(). Over time this fragments meaning in the source and makes it harder to reason about until it is a giant pile of slop.

The thing the dev should do that often doesn't get done is to take the information about the undesired behavior back up to the design/spec level and rework the definition to account for it. This doesn't normally get done because it takes longer and it is conceptually more difficult than the patch.

This is something that LLMs can accelerate significantly and the result is a shallower gradient of technical debt over the long term.

Your tool looks interesting but the readme file is very "wall of text" heavy and example light. In comparison the site for this tool shows exactly what it does. Perhaps that explains the difference in response. This is a classic case of the need for "show don't tell".