IanCal

41 ms·

811.
by IanCal 2y ago
Have you never understood the solution to a puzzle much more easily than solving it yourself? I feel there's literally a huge branch of mathematics dedicated to the difference between finding and validating a solution. More specifi
812.
by IanCal 2y ago
Prodigi are a company I've used for print on demand stuff before (a snapshot of the sun in different spectra given a specific time like a birth), though I've only done art prints they do have temporary tattoos. https://
813.
by IanCal 2y ago
I understand what it's doing, and could easily set out the cases I needed.
814.
by IanCal 2y ago
It did fix it, I iterated passing in the type and linter errors until it passed all the requirements I had. > If you merely know how to check, would you also know how to fix it after you find that it's wrong? Probably? I'm capa
815.
by IanCal 2y ago
I know this is a take on the internet but it's a bit on the nose to also have it request to send your data to over a hundred different companies.
816.
by IanCal 2y ago
I used to build these too, my example was one in actionscript (always fun on restricted devices) and it was a bug in a library I wrote. I was actually testing the property tester and found the bug then, after realizing it wasn't a bug
817.
by IanCal 2y ago
Half of the problems are people treating them as searchers when they aren't. They're absolutely not ngram indexes of existing data, either.
818.
by IanCal 2y ago
That seems like a wild restriction. You can give them more latitude for things you know how to check . I didn't know how to setup the right gnarly typescript generic type to solve my problem but I could easily verify it's correct
819.
by IanCal 2y ago
I disagree. I've used it for things like "regardless of where you are on the page, tab n times and shift tab n times leaves you on the original item". I found a bug in our tv ui library which was actually a bug in the spec. R
820.
by IanCal 2y ago
> 1: I understand your concern in terms of google auth. I always refused to implement it because i did not want google to have information about my users, but every developer told me i was missing out on so many signups, so i tried it th
821.
by IanCal 2y ago
I really don't think that's the definition of a program. > You want it to run until you tell it to stop, No? Many programs I don't want to run until I tell them to stop. Even then, this reduces it to irrelevance.
822.
by IanCal 2y ago
It depends how you define reliably. The halting problem isn't so relevant in most development, and nothing stops you having a classifier that says "yes", "no" or "maybe". You can identify code that definit
823.
by IanCal 2y ago
> But to me saying Google's advertising product should enforce how the individual websites work [fingerprinting], is to me more in the direction of "an electricity provider should enforce how people live their lives in any home
824.
by IanCal 2y ago
> and would therefore prevent people from doing fingerprinting without ICO having to get involved (which is what the person you originally replied to was focussing on). But that simply isn't true in the broad sense. It would stop
825.
by IanCal 2y ago
Too late to edit - "decision at this layer "
826.
by IanCal 2y ago
Yes, they call it out as a bad change and are making a statement that it will be hard for companies who use this to justify it.
827.
by IanCal 2y ago
I really don't understand this comment. They're not expecting google to enforce anything, and they are talking about going after individual companies. > If governments enforce policy by bullying HSBC/Google/E.ON to
828.
by IanCal 2y ago
I've found having a cascade of things helps. Trying to split things into "decision at this later, or pass on to the next" with increasingly complicated models/approaches. Start with ifs, then svms then something else for
829.
by IanCal 2y ago
I think you're talking about something related but not discussed here, which would be making car modifications illegal. Let's say BMW make heated seats an optional extra, which it already is. What if they also install them in ev
830.
by IanCal 2y ago
Is that true? I thought there was a desire to move towards byte level work rather than tokens, and that the benefits of tokens was more that you are reducing the context size for the same input.
831.
by IanCal 2y ago
Devils advocate here If this made it more expensive, because having cars that are physically different is harder, would you prefer that? Why can't I simply not care that I don't get something I've not paid for and there&#x2
832.
by IanCal 2y ago
Although if you drew that out you'd have about the same. Cron trigger. Process. Gpt API. Database for persistence. Email sender. Which part of that wouldn't you have?
833.
by IanCal 2y ago
Using openrouter, a bunch of models fail on this. Sonnet 3.5 so far seems to be the best at saying it doesn't know, other than perhaps o1 pro, but once that has said "no" (which can be triggered more by telling it to respond
834.
by IanCal 2y ago
O1 pro. The "thinking" part explains it seems to be about a custom web server, and tries to think what ggh might be, saying maybe something like "go gprc hub" and it needs more info. The response is: I’m not aware of a s
835.
by IanCal 2y ago
maybe I get your point, but there's an important difference between "I don't know what they are" and "they don't exist".
836.
by IanCal 2y ago
Sure, I'm interested in where the boundaries are with this. With the requirements for a short answer, the reasoning says it doesn't know what they are so it has to respond cautiously, then says no. Without that requirement it says
837.
by IanCal 2y ago
> Whereas LLMs today will simply give you a definitive answer even if it doesn't know. Have you not seen an LLM say it doesn't know the answer to something? I just asked "How do I enable a scroflpublaflex on a ggh connecti
838.
by IanCal 2y ago
I find this kind of argument comes up a lot and it seems fundamentally flawed to me. 1. You can set a bar wherever you want for a level of "seriousness" and huge swathes of real world work will fall below it, and are therefore att
839.
by IanCal 2y ago
Have you tried using an llm to say whether the puzzles are good or not?
840.
by IanCal 2y ago
I don't think it's worth continuing then. I have previously got into long discussions only for a follow-up to be something like "it can't have purpose it's a machine". I wanted to check things before responding

More ›