It doesn't make the economics any different. In a browser environment, you're maybe looking at the acceptable lease being 100MB for 1 second. Much more than that, and you start hitting limits of what browsers will let you do on low-end phones. Longer than that, and we're back to the user-observable latencies being too long.
100MB for 1 second just is not much of a deterrent.
Proof of work does not scale. It trades something fungible and incredibly cheap (CPU) for something incredibly expensive (user-visible latency). There is no set of parameters where the cost is going to be a meaningful deterrent to any kind of abuse (even something as low-yield as scraping) without adding crippling amounts of latency to real users.
> The dilemma for bots: when tokens are bound to the connecting ip, scrapers must limit the connecting IP pool for each site they want to scrape, becoming much more obvious and easy to block, or they have to use massive amounts of compute.
There is no dilemma. They get a token, they maybe do some automated multi-armed bandit per-site to figure out how to maximize the extraction rate they get from a single token, and then they use an IP for that many requests / that amount of time before ditching it.
My point is that Kelley did not argue that what Bun does isn't really fuzzing. He wrote that the post's claim is a fabrication. But that claim is really specific, and to evaluate whether it is true it doesn't matter what Kelley's unstated definition of fuzzing is.
So an argument about definitions doesn't seem super valuable here.
I don't understand what distinction you're trying to draw here. The very specific claim[0] in the Bun blog post that Kelley is calling a fabrication was:
> We fuzz Bun's runtime APIs 24/7 using Fuzzilli, the JavaScript engine fuzzer used by V8 & JavaScriptCore
It does not look to be a fabrication, and is very explicit just about what they meant by fuzzing.
[0] I mean, that sentence doesn't actually match Kelley's paraphrase, but it is literally the only claim in the post related to what fuzzing was done on the Zig-based bun codebase. So it has to be what Kelley was referring to, and his paraphrase is as sloppy as his fact-checking.
> Contrary to the amount of times "But honestly" or "genuinely" is mentioned, nothing about having your LLM speak for you feels honest or genuine.
"Honestly" is used once in that post, in a way that's pretty much the core, self-deprecating human use for it ("It would have been possible to do X, but honestly I didn't want to"), rather than the filler word use-case.
"Genuinely" is not used at all.
> I know it's not cool to leave responses like this, but I'm really tired of all of this at this point.
I think it is cool to flag AI-generated slop and either leave a comment or upvote an existing comment about it being slop. But only if you are sure it's AI-generated. And sorry to say, you don't seem very well calibrated on this. If you can't actually tell the difference and back up your opinion but are just guessing, then it indeed isn't cool.
Nice! Have you considered doing a Show HN for that?
That's valuable in at least three different ways: public education, showing that most of the articles are still human-written which can be easy to forget about sometimes, and as an easy way to cross-validate my intuition when flagging something as AI-generated without having to manually run Pangram.
I despair a little bit about how many HN voters either seem to want to read slop or don't understand when they're reading it. This post is obviously AI generated from the first paragraph on, and still has 480 votes.
The post I was replying to said "performs strictly better at the same cost per task". That claim was obviously not true, there are costs where Opus cannot do the task and Sonnet can, so Opus can't be performing strictly better that the same cost. It seems that you agree that it is not true.
You could make it true by artificially dropping some of the data points, but, like, why?
(Again, this is moot given the updated graph.)
> Of course if you go beyond those x-values where only one of the two are defined, then trivially the one that is defined constitutes the Pareto frontier in that region.
Not so! It's only sound to do that at the low end of the cost axis (x) or the high end of the performance axis (y). You can't do it at the low end of the performance axis or the high end of the cost axis.
I really don't get what you're proposing. The cost ranges do not overlap at the low end. You can't (by definition!) interpolate outside of the range.
If you mean extrapolate, at that point you're just making up data. The available effort levels are discrete and covered totally by the benchmarks. You can draw on the monitor with a sharpie to show a "ultra-low" effort level for Opus that scores better than Sonnet "low" at the same price, but it doesn't magic the ultra-low effort into actual existence.
(Anyway, the blog post now has an errata and a graph that shows substantially better relative performance for Sonnet 5.0 than the original graph.)
But they don't show "strictly better" performance at cost per task!
The graphs show parts of the cost/performance pareto frontier occupied by Opus 4.8 and others occupied by Sonnet 5.0. If Opus 4.8 was strictly better at cost per task like you say, by definition the entire frontier would be occupied by Opus.
So neither is pareto-dominant over the other. In contrast, Sonnet 5.0 is Pareto-dominent over Sonnet 4.6 on those graphs.
Only speaking for myself, not my employer.