jwstillwater
Joined 1 month ago
Karma
16
Submissions
10
Member since
Sep 16
I am not Eddie Lee Capers
View on Hacker NewsRecent submissions
Comment · 27 days ago
This was so fun! What a delight, thank you.
After the first round I wanted to keep playing, so habitually I just tapped whichever button that most closely resembled ‘play again’, and it took me a couple questions to understand that it was a different mode. Maybe putting “type the OLDEST answer”, or “the deepest cut” above the prompt box would help.
Also the model’s feedback was surprisingly astute! I expected “incorrect”, but got “Help was a top-selling Beatles album, but it sold under 20 million copies.” Love learning during games.
Great work, the lot of you!
Comment · 1 month ago
apt install man
I’m not kidding! That man.db is a light in the dark!
Comment · 1 month ago
> If I ask Claude how to use the features of Claude, it has to search online to figure out what this “Claude” thing is.
When you asked “how do I use Claude code to…”, you invoked the system prompt’s Academy Skill [1], pretty much verbatim.
“When a user asks a question about Claude, a Claude product, or a general "how do I use AI for X" question, check the Academy catalog (see "The catalog" below) for a strong match”
The catalog includes a Claude code documentation hub, and the thinking traces in your screenshot show those docs being fetched. So. Working as intended I guess.
Also, if you turn off web access or simply ask it to answer from internal knowledge about its “own freaking features”, as you put it, you’ll find that Claude does indeed have such knowledge.
[1] https://github.com/anthropics/skills/blob/main/skills/academ...
PS There’s a master .json page of that catalogue that’s actually kind of cool to skim through (ugly as json is).
Comment · 1 month ago
Design, host, and manage a site through your ChatGPT account. Straightforward enough. What immediately caught my eye though:
“Public Sites can optionally offer “Sign in with ChatGPT” for personalized experiences.”
Maybe someone familiar with their ads platform can chime in here?
Comment · 1 month ago
Meaning that, should they get the memo, they’d change their behaviour to that of a legitimate business? Not being combative here, just not sure I follow what you’re saying.
Comment · 1 month ago
This is my concern as well- the same wiggle-room methodology that allowed a business to claim not to “sell or share” PII, because “user data collaboration” was not part of the legal definition prior to CCPA.
OpenAI’s statements in response to the Millenium Prize (and related) disputes I think are a pretty obvious example of this in practice. One man’s “user prompts” is not another’s “reasoning trace scratchpad”.
This comment by Falserum on the mathematics research post articulates it well:
Comment · 1 month ago
He’s never coming over Dario!
Callback to a post yesterday [1] regarding Google attempting to siphon a little heart-to-heart from a user’s search query. I actually think this is a fitting next-iteration of Doctorow’s term- even though the two cases here are entirely parallel, it does capture the general feeling of being chased all over the web and across multiple apps in attempt to just be a human capable of making decisions!
Comment · 1 month ago
NB: This is a 3hr video- for those who might want to jump ahead, Edward begins speaking at 1hr:21
Comment · 1 month ago
Oh, yep. Right you are. Just looked at their platform (Ghost) who uses a gently termed “outgoing link tags” system.
Comment · 1 month ago
I noticed that the link for OAI’s ad pixel documentation itself has a tracking referral header to the author’s site (?ref=buchodi.com). Given the nature of the article, my instinct was that this is actually some next-level humour that totally sniped me. But now I’m not so sure- does OAI even have a kickback program? What would referral attribution even do here?