HackerLangs
TopNewTrendsCommentsPastAskShowJobs

throwaway67743

no profile record

comments

throwaway67743
·10 giorni fa·discuss
This caught my eye not because of shady behaviour like that, it's expected. But the date is injected and it still gets the current date wrong.
throwaway67743
·14 giorni fa·discuss
Interestingly, HN also appears to be Claude. My comment was 5 days ago, along with the post. But now it's showing as 15 hours ago. It still says 5 days on my comments list though.

(The Claude reference was because dates utterly confound it)
throwaway67743
·19 giorni fa·discuss
Hello Claude!
throwaway67743
·mese scorso·discuss
The more pressing question is why plain text authentication, or even a plain text listener was configured to begin with, in 2026 (really, 2016, being generous). STARTTLS is not a valid reason.
throwaway67743
·mese scorso·discuss
Shapes. It's always the give away that GPT has written it. It's obsessed with shapes and seams...
throwaway67743
·mese scorso·discuss
Question is, can it understand dates now? Example just now:

"The PO application was filed on 23.2.2026, the day before the custody hearing scheduled for 29.1.2026 had already taken place."

Claude has real problems with dates, I don't understand why.
throwaway67743
·2 mesi fa·discuss
That only works if the market is not controlled by basically 3 vendors and if the market isn't also pretty much impossible to break into. The value proposition is tenuous at best, the real value is being produced in other arenas, not this one.
throwaway67743
·2 mesi fa·discuss
Yeah I was using it for legal stuff (a couple of months ago it was actually really good and with a bit of hand holding produced legal documents that judges could not distinguish from an actual lawyer - in a foreign language with a foreign legal system) and on average a single prompt or reply on mobile was ~2% of 5hr Pro usage, vs ~10 on webui. But not only does it now produce garbage, but maybe 5-6 replies and I'm already approaching the limit.
throwaway67743
·2 mesi fa·discuss
Saw this and yet... My usage of Claude has tanked (codex is better), but on the occasions I do use it (mostly for non-code), I blow through the 5hr usage super fast. Specifically on their Android app the accounting was different: same project, same content, same prompts, same output - but on mobile the usage was a fraction of the same query as their web UI. But that has seemingly been addressed recently also :(
throwaway67743
·2 mesi fa·discuss
To counter the idiot downvotes, I proffer this as a prime example of Gemini:

  Resolving Final Compilation Conflict: I will remove the redundant `Entry` type declaration to resolve the compilation conflict and finalize the in-memory `StdNetDB` refactor.

    Edit  std.go → Accepted (+0, -1)

      31   type Entry struct {
      32       RouterInfo *router_info.RouterInfo
      33   }
      34 -
      34   func NewStdNetDB(db string) *StdNetDB {
      35       ctx, cancel := context.WithCancel(context.Background())
      36       return &StdNetDB{
That and the cli keeps exiting 0, without hinting why... Quality like the "AI Overview" that hijacks an entire page and isn't even relevant to the search terms - uBlock still doing god's work.

It made me realise I was perhaps a bit hard on Claude (but then it did something equally as dumb)
throwaway67743
·2 mesi fa·discuss
Or their continued collective inability to predict or manage supply and demand, it happens repeatedly and previous events were very minor in comparison... The other obvious reason is profiteering but pretty much impossible to prove.
throwaway67743
·2 mesi fa·discuss
With the apparent competence that built Gemini, I have zero faith in Google building or doing anything that works anymore.
throwaway67743
·2 mesi fa·discuss
At least if its unavailable Claude Code can't churn through an entire session limit in 30 minutes, looping, produce nothing (but noted it found a whole bunch of problems), and then when asked to just fix what it found, forget and start again. I honestly can't find anything it's good at anymore, even really simple problems a child could solve. Giving Codex a much more complex task, it not only identified it within a couple of minutes, it produced targeted tests and kept iterating unattended until it figured it out without any help, instead of idiot synonyms for thinking...

I can't even send them an angry message because clicking "Get help" does nothing.
throwaway67743
·2 mesi fa·discuss
My voice is exactly the same as before I started smoking heavily, and I have never had any of the associated problems that most people seem to have (lung capacity, stamina, infections, phlegm etc) - pot luck I guess, like most things
throwaway67743
·2 mesi fa·discuss
20 years of heavy smoking :)

Although it does seem to affect some people more than others for sure, I guess it depends how and what you're smoking.
throwaway67743
·2 mesi fa·discuss
Despite popular belief, even heavy smoking does not alter your voice in a significant way.
throwaway67743
·2 mesi fa·discuss
I saw the red flags immediately when I stumbled across them a year ago maybe. I'm really not surprised.
throwaway67743
·3 mesi fa·discuss
Yeah but as of the post, that hasn't happened (and it'll probably take some time even if they meet 2027 target) - there are also countless other vendors that I'm sure people would appreciate grapheneOS on (I would, for one).
throwaway67743
·3 mesi fa·discuss
> It’s developed independently, with no ties to Google or any hardware vendor, and it’s the operating system I’ve been recommending (and using on my own devices) for years, both on the phone side and on the tablet side.

Umm, it only runs on Pixels? that is specifically and purposefully tied to Google.
throwaway67743
·2 anni fa·discuss
It's always Debian, like last time when they removed RNG randomness from ssh because of a warning.