Home / Articles / The wall under the reply / Versions / v1.1.0
The wall under the reply: what changed in v1.1.0
From v1.0.0 (2026-10-04, 12ba9608e) to v1.1.0 (2026-10-04, b5a363102), paragraph by paragraph.
4 paragraphs added, 0 removed, 0 changed in place, 60 unchanged. About 254 words added and 0 removed. Insertions are marked like this, deletions like this; unchanged runs are folded to one line; figures appear as their file names.
45 unchanged paragraphs, under In short, The wall, and what it costs, The proposal: end with the state, not the history…
The article, condensed by another model
A few hours after this was published, Dinis fed it to ChatGPT and asked for an infographic. What came back is below, and it belongs here for a reason beyond being a good picture: it is a one-page state of this article, made for a reader, by a model that was not the one that drafted it. The five rows are there, the Cc change is there, the human-in-the-loop workflow and the four variants and the evaluation signals are there, and nothing was invented. It is the thing the article proposes, done to the article.
[figure tail-one-page-by-another-model.webp] The article as a one-page state, generated with ChatGPT from the published text on 4 October 2026 and credited by it to the source. The thread shown is fictional; the example addresses are the model's placeholders, not real mailboxes. The style is the other model's, kept as it came, because who made a condensation is part of its provenance.
Two things worth noticing. The picture is a reader's view, not the sender's: it says what the article looks like from outside, and it is right about that. And it was cheap, which is the point of the whole proposal: the state of a long thing, written for one reader, is now something an agent produces in a minute and a person checks in less. The argument of the memory article applies here too. Two models, one published source, and a condensation that cites where it came from.
15 unchanged paragraphs, under Threads woven here, Sources