Profile avatar
danielgrady.net
20 posts 2 followers 22 following
Getting Started
Conversation Starter
comment in response to post
That description really resonates - was suddenly reminded of the few I’ve known who fit that mold, and how similar they are to Yarvin.
comment in response to post
"I'm working inside the Claude AI web interface, and looking at my older conversations. How can I tell which version of Claude produced these older responses?" Basically wrong. A right answer would be roughly "You can't, unless you just guess." claude.ai/share/9bf7a2...
comment in response to post
"Anthropic recently launched a new "Max" plan. Does this plan allow me to make API calls?" Basically right, but confusing. Which to be fair is also true of the docs. claude.ai/share/33f152...
comment in response to post
"How do I set up VSCode Copilot to use Anthropic with personal API keys?" Wrong. Prompting to search gives an answer that focuses on paid Copilot features, only mentioning the API key option in passing. claude.ai/share/889658...
comment in response to post
"I'm working with FlatBuffers in Python. It seems like there is not a clean, idiomatic way to introspect on the data fields of a payload. Is this right? Why?" Right. (I think.) claude.ai/share/ac8c7f...
comment in response to post
Even in mathematics, learning doesn’t happen (for me) until I’ve attempted to teach it — “making” the argument, maybe?
comment in response to post
Echoes of this even in very abstract problem spaces: in data science, the number one thing is to look at the data, literally look at Claudia-Perlich-style, load the data into a spreadsheet and stare at it and explain what you see.
comment in response to post
Jokes aside, it seems like OpenAI has changed the wording but not the import. Still looking for @agrobbonta.oag.ca.gov to take action here and prevent this restructuring.
comment in response to post
Thanks! I love these posts you do with short examples for local inference; they are super helpful.
comment in response to post
Do you have some default options set? I had to go look through the `llm-mlx` documentation to find `-o unlimited 1` to reproduce your Atlantis example without it getting truncated early.
comment in response to post
These are all consistent and not inherently evil: - You're great at your job. - Your job needs to change for the company to still make sense soon. - You should make a conscious effort to get good at using this new technology.
comment in response to post
I'm skeptical about the examples of earlier tech rollouts — I would figure there *would* be groups that top-down drove the adoption of spreadsheets where many employees were resistant to the new technology, and that it was the right decision for the company.
comment in response to post
Sympathetic to the point, and I think the example of the "more normal policy" is the best part of the piece.
comment in response to post
"Good" isn't a singular axis. For many, part of "good show" is "makes me optimistic about humans," and on that axis it's a remarkable outlier.
comment in response to post
Crazy. Always surprises me how many of these one-purpose tools get produced; seems like people would just point to or use Pandoc.