lxgr

46 ms·

451.
by lxgr 5mo ago
Especially without being able to look at the rendered output! (At least I'd be surprised if modern server-side tool calls regularly include an SVG renderer that can show a rasterized version to the model to iterate on it.)
452.
by lxgr 5mo ago
It's mindblowing to me how modern cell phones have worse broadcast and P2P capabilities than what we had decades ago (when feature phones often featured FM receivers and could share photos via Bluetooth across manufacturers and OSes).
453.
by lxgr 5mo ago
That's what I meant by "effectively": Reflections, diffraction, and (moderate, i.e. think doors and windows, not walls) light obstacle penetration make some indoor reception possible, but the radio horizon remains the strong
454.
by lxgr 5mo ago
I agree, it is a fantastic job! The title is still misleading to me. Not terribly so – hence my not quite serious reply – but I was expecting something like https://timestation.pages.dev/ , which actually transmits using inc
455.
by lxgr 5mo ago
VHF is effectively line-of-sight, and no antenna size can change that (although it does improve efficiency for both sending and receiving), so for two handheld radios, you are limited to about 10 km. The only thing that really helps extend
456.
by lxgr 5mo ago
"Turn your phone into a nuclear reactor (by plugging it into a wall outlet served by a nuclear power plant)"
457.
by lxgr 5mo ago
Isn’t that a good argument for needing a more capable model for assisting in it then?
458.
by lxgr 5mo ago
I'd definitely give a Bitwarden client alternative a try, but I really hope this isn't the start of client fragmentation like it happened for Keepass, especially given that a server is involved here.
459.
by lxgr 5mo ago
Of course they will, just like humans are still knitting sweaters by hand even after the invention of the power loom.
460.
by lxgr 5mo ago
Why not? People are already holding their phone running Google Translate up to strangers to communicate with them.
461.
by lxgr 5mo ago
No, I just don't think that "technology x sometime/often/exclusively gets used/co-opted/... for evil" and "technology x is evil" are equivalent, and I think that the distinction matters.
462.
by lxgr 5mo ago
> no, I never found it difficult or had a problem with it. If you don't even notice the dark patterns, they're working exactly as intended.
463.
by lxgr 5mo ago
They can consider hundreds of pages of text (including terms and conditions, public sentiment, local laws etc.) in a matter of seconds in a way that I, tired after an international flight and just wanting to get a ride to my hotel, usually
464.
by lxgr 5mo ago
I’m not terrified of learning how to design a good website, I would simply not take up a design-heavy project, as I don’t enjoy it at all. I do like certain other aspects of projects like that, which is why it’s great to have LLMs to collab
465.
by lxgr 5mo ago
The dark patterns and generally options not presented in the app are exactly where I would expect a competent assistant to shine. Are local taxi services cheaper and known to be more reliable? Am I missing an obvious public transit option?
466.
by lxgr 5mo ago
> [...] broadcast that they're so important and busy that they can't take the 5 seconds [...] It takes a lot more than 5 seconds to make an informed decision these days. Apps and websites are throwing abusive fine print and dar
467.
by lxgr 5mo ago
I know how to set up a static HTML site in about 15 minutes. Building a website to host there usually takes me the better part of a weekend, and usually ends up looking absolutely terrible.
468.
by lxgr 5mo ago
Nuclear bombs yes, but what about nuclear physics?
469.
by lxgr 5mo ago
Everybody wants to do it, but doing it in a way that's survivable to a company with a brand image to preserve and potential legal liability for the consequences is not nearly as easy.
470.
by lxgr 5mo ago
If the site works, continues working, can be extended and maintained etc., it’s not slop.
471.
by lxgr 5mo ago
It’s clearly also faster than its developers can handle.
472.
by lxgr 5mo ago
It’s much worse than that. Openclaw cosplays security, which is much more dangerous than just outright defining the security model as “this is new and risky, better sandbox this thing thoroughly at a different layer”. It sucks at both secur
473.
by lxgr 5mo ago
That’s an excellent way to lose to your competitor that knows and respects your users’ resource constraints.
474.
by lxgr 5mo ago
For the few weeks in which I’ve been using it, it has brought down the Raspberry Pi it’s running on several times with extreme resource hogging, local history/memory search is broken due to a trivial bug for which all issues are auto-c
475.
by lxgr 5mo ago
The ambitious idea is actually giving a chatbot/agent access to a bunch of personal data and having it self-modify its harness and context to some extent. The execution in case of Openclaw is a hot mess.
476.
by lxgr 5mo ago
The idea is lean, the execution is slop.
477.
by lxgr 5mo ago
The fact that the OpenClaw creators seemingly missed that parallel tells you everything you need to know about the project.
478.
by lxgr 5mo ago
Not really, given that it can’t represent even basic formatting such as bold or italic text, chapter markers etc. As an output format it’s ok, but as an input format, it’s almost as bad as PDF.
479.
by lxgr 5mo ago
That's unfortunate, and almost certainly a Regulation E violation on their side. They're supposed to provide a provisional credit within 10 business days.
480.
by lxgr 5mo ago
To be honest, that's on the bank then. Doing re-2FA for every outbound transfer, and mentioning the consequences of entering the 2FA code out of band (e.g. "enter code 123456 to confirm transfer of x$ to y" or "press OK

More ›