https://hackerpaper.kaipfister.de/p/2608.13867/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMThttps://news.ycombinator.com/item?id=49345660urn:hackerpapers:mention:2608.13867:49345660Tue, 18 Aug 2026 13:52:31 GMT<p>Hi, author here! This essay came out of a much larger review I&#x27;ve been working on on coding-agent reliability, and a ton of time operating these systems and watching them break (a lot). I&#x27;ve seen agent + fleet failure show up in lost or ambiguous state, duplicated retries, stale workers, weak ownership boundaries, non-authoritative observations being treated as truth, bad admission control, etc etc. In other words, many of the failures popping up look like distributed systems engineering problems. The longer technical review arXiv document is here: https://arxiv.org/abs/2608.13867 and I put…</p><p><a href="https://hackerpaper.kaipfister.de/p/2608.13867/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.SE