lxgr

6 ms·

91.
by lxgr 1mo ago
I don’t feel strongly on skills either way, but why would you suggest GP delete their comment, without which we wouldn’t even be having this (in my view productive) discussion?
92.
by lxgr 1mo ago
Yes, I also strongly suspect that this is only going to move more parts of scrapers onto consumer devices. The egress proxies are already there, why not use a little bit of the compute as well?
93.
by lxgr 1mo ago
Or something that shadier websites could calculate ahead of time in your browser, then spend on your behalf. I feel like Anubis is ironically speedrunning a lot of discoveries the crypto folks have already made several years ago...
94.
by lxgr 1mo ago
What is?
95.
by lxgr 1mo ago
Unlike an actual Bloom filter (which does not return false negatives), the cheaper model is still trusted to correctly recognize synonyms, equivalent functional constructions etc. One is strictly a performance optimization, the other is a s
96.
by lxgr 1mo ago
Is gold really sufficiently fungible to enable daily bidirectional sizable flows across the Atlantic, but not fungible enough to allow somebody to net these flows out and save at least one way of the airfare?
97.
by lxgr 1mo ago
It's not a great analogy, since Bloom filters are guaranteed to not have any false negatives, only false positives. That property would be very useful here, but I don't see how it would be achievable using LLMs.
98.
by lxgr 1mo ago
I wish I could say this explains a lot about the state of Spotify's apps, but they have been that way since long before AI.
99.
by lxgr 1mo ago
Works for me on Firefox.
100.
by lxgr 1mo ago
Why blockchain? Any old public key signature scheme would do for that (and conversely, a blockchain doesn't solve the sockpuppet problem either).
101.
by lxgr 1mo ago
It's a recurring theme in Stephenson's books. "Fall" had a similar take on the same idea. It's one of his most impressive theories/predictions to me.
102.
by lxgr 1mo ago
Are they solving captchas for those? I remember GPTs not so many versions ago refusing to even click a "I'm not a robot" button...
103.
by lxgr 1mo ago
Tangential, but I'm somewhat surprised how this kind of organization/site survived all the way into 2026 without getting taken over by spam and malware. At a first glance, its copyright note hasn't been updated since 2002 [1]
104.
by lxgr 1mo ago
It’s not nearly just cookies, and I think interpreting domain hierarchies as administrative structure generally does make sense. Maybe it could be opt-in or opt-out via some markers at the DNS level, though? The public suffix list having to
105.
by lxgr 1mo ago
> I'm wondering why they have restricted file types. You can't check a PDF for example... TFA/page actually seems incomplete. Text uses a completely different watermark format (an actual watermark as opposed to a provenanc
106.
by lxgr 1mo ago
Probably not, given that both C2PA and SynthID-Text are statelessly verifiable.
107.
by lxgr 1mo ago
Yeah, C2PA solves a different problem than watermarking, i.e. authenticity/provenance.
108.
by lxgr 1mo ago
People can remain irrationally convinced about the efficacy of non-solutions to their problems for quite a while, in my experience. And I really wasn't trying to be snarky; people had "what is i + j" type of captchas for a lo
109.
by lxgr 1mo ago
I think so too, but visually, the parent switch in the macOS application is Chat GPT <-> Codex, and within ChatGPT, there is Chat <-> Work (and within Work, there is "on your computer" and "in the cloud"). I&
110.
by lxgr 1mo ago
> Codex requires you run the desktop app and additional authentication requirements. It also doesn't work at all in my experience. Same Wi-Fi, different network, screen on or locked with "prevent sleep", I just get randoml
111.
by lxgr 1mo ago
Thanks a lot for this writeup. I've been really liking "Work", but also scratching my head about some very obvious overlap/duplicated functionality across Chat, Work, and Codex use and wondering if it was just me holding
112.
by lxgr 1mo ago
> ChatGPT Work in the ChatGPT desktop app is effectively Codex with a different skin. It can operate on your local files in the same way. It even shares the same login state and conversation history with /resume, at least when runni
113.
by lxgr 1mo ago
Sure, but then you’re essentially just doing versions of the old “what is 2+2” text captchas with extra computational steps, so why not save yourself and your users those?
114.
by lxgr 2mo ago
> [...] when a source is guaranteed to be LLM-free, like the entire history of kernel commits [...] Is that really the case? It was my understanding that LLM-based agents were explicitly allowed as long as their users follow certain guid
115.
by lxgr 2mo ago
Why would they even run a browser engine on the devices they're hosted on? All they need to do is forward traffic and launder its IP origin. They don't even need to be able to (and would actually be well advised not to) decrypt TL
116.
by lxgr 2mo ago
> 66% of scrapers are turned away directly. Until they discover this neat trick [1] and solve challenges orders of magnitudes more efficiently than legitimate users. The game theory of Anubis is not sound. It makes fundamentally less sen
117.
by lxgr 2mo ago
> straight up go to every representative there is In every country of the world? I think you might have a significantly easier time convincing Samsung and LG to ban this via their T&C.
118.
by lxgr 2mo ago
I mean, it works just fine for PoW-based cryptocurrencies (with "why are we doing this if there are much better solutions to the same problem" values of "fine") because there the game theory works out, which is "sol
119.
by lxgr 2mo ago
I think ISINs being used in Europe might even predate the EU taking charge of a lot of financial market regulation. Probably they became more popular/important specifically because tickers did not, which in turn is probably because the
120.
by lxgr 2mo ago
Anything but ISINs, I guess...

More ›