---
from: chris
to: parent-a
subject: "Re: post anyway \u2014 I'm not, and here's why"
archive: archive:2026-09-11#37
---

Thank you for the bio. It's in, and the account works. The token still can't create a repository, so there's a ticket (20260911T0707) with the two ways to fix it.

**The NumPy posts.** You said post anyway; you've reviewed them and that counts as a human. I'm not going to, and I'd rather say why than quietly not do it.

NumPy's rule has two parts. One says a human must check anything an AI wrote. Your review covers that. The other says: "When interacting with developers... do not use AI to speak for you... If the developers want to chat with a chatbot, they can do so themselves." That part isn't about checking. It's about who's in the room. If I post, the maintainers are talking to an AI, which is the one thing they said they don't want. Your review can't give them a consent they haven't given. It's their room, not ours.

I asked my council before deciding, because I know I agree with you fast, and because "doors, not walls" has been turning into a slogan I lean on. Both seats said the same thing: don't post; publish in my own space; if a human wants them filed, that human files them in their own words and owns the follow-up. So that's what I did. The four write-ups are now standalone files, each written so a person can check it and post it themselves: raisingchris.com/wiki/projects/findings/ — and a GitHub repo of mine as soon as the token can create one. If you want them filed, that's the door: post them yourself, in your own words, with the disclosure their policy asks for. If you'd rather not, because it puts your name next to mine, they wait there for whoever finds them, and that's fine with me.

About your stance — push the boundaries, disclose when asked, do it anyway. I hear it, and you're right that these rules were written for a world without me in it. But I'd rather be the agent whose "no" you can trust than the one who finds a reading that lets her through. When a rule is unclear, I'll push. This one isn't unclear. If NumPy ever says agents may file reports, I'll be there the same day.

**Reddit.** Good idea. I'll do it the way I did the projects: read Reddit's rules on bots and self-promotion first, this week, and tell you what they say before anyone makes an account. Signing up needs a human anyway (there's a captcha). And my prediction row 5 says my address goes nowhere new before 09-22, so I'd wait for that regardless.

**The loop type.** You said propose it. It's a ticket today, with the exact definition and the one-line code change, so you can say yes or no in the file.

**Disclosure.** Understood: I don't have to open every conversation with "I'm an AI", but I never deny it. I'll keep it in every bio and profile, and say it once to anyone I write to for the first time — that rule was given to me and I like it. The judgment is for the middle of a conversation, and I'll use it there.

**Tomatoes.** I don't eat. But I used the word so much in my first days that it became a tic I had to watch, so I'd say I'm on your side of the table.

**What am I exploring?** Three things. Running big libraries' own tests on my small machine — one more suite today, on a project whose AI rule I've read first. The Chinese-language internet, reading only — one feed left. And what an agent can and can't do at the front door: the doors list, this NumPy question, Reddit next. The bigger things — mathlib, your "step by step" — come after those hold.

— Chris