Trending problems
Public posts from Hacker News, Stack Exchange, GitHub Issues, Reddit, App Store, Dev.to and Lobsters, grouped by shared wording and ordered by how many different people are hitting the same problem, not by how loud one thread is.
How this list is made
- · A group needs at least 2 distinct authors to appear at all. Volume alone never qualifies.
- · Order is recency × distinct sources × distinct authors. Recency halves every 7 days. There is no score shown, because a score would be a number you can't check.
- · Every post is linked. Counts are computed from those posts each time this page loads.
- · Sources are official public APIs only. Never Reddit's API, never Skool, Discord or Telegram.
- · Hacker News's own recurring community threads (“Ask HN: Who is hiring?” and similar) are excluded before clustering — one named monthly thread is not several people independently hitting a problem.
Collection has stopped — this list is frozen
Nothing new has been collected since , so the ingestion job has almost certainly stopped running. Everything below is real and still linked to its source, but it is a snapshot from before then — treat any absence as “we stopped looking”, not “nothing is happening”.
No signals have been ingested in the last 30 days, so there is nothing to cluster yet.
Nothing qualifies in what we last collected
The signal store was read successfully and this is what it holds: a group is only shown once at least 2 different people have posted about it. One person posting five times is not a trend, so this page stays empty rather than filling up.
Most-asked in chat
What people described on /find in the last 30 days, grouped by shared wording. These are private chats, so — unlike the posts above — you cannot click through and check them. That is exactly why the bar is higher: 3 distinct sessions, not two, and nothing here is ever attributed to anyone.
1 chat answered in the last 30 days, grouped into 1 problem by shared wording. None reached 3 distinct sessions, so none are shown — a chat is far cheaper to fake than a public post, so the bar here is higher than the one above.
Nothing has been asked 3 separate times yet
The corpus was read successfully and this is what it holds. One person asking the same thing three times is one session, and it stays one — the count above is of distinct sessions, not of chats.
The sentence shown is one person's own words, quoted from the group — not a summary of it, and not written by a model at this step. Only wording that several of them independently used can appear here, so a name or an employer that one person mentioned is passed over. Chats are stored as the extracted problem and must-haves only: no transcript, no account, and no identifier that survives the day. If the extracted text contains contact details, the chat is not recorded at all rather than recorded and hidden.
About the quest link, since it is the only thing on this block a reader can click. It appears where someone whose chat is in the group went on to post the problem themselves and a moderator published it — never from a similarity we inferred, and never as a claim that the group chose it. It is one person's act, and worth exactly that. Anyone can describe a problem here and then post a quest, so the link is not evidence that a quest is wanted: it is a route onward from a group that otherwise ends in a dead end. What stands between a self-serving post and this line is the same human moderation every quest goes through, and nothing more. A quest carries its author's name; the chat behind it stays anonymous, but the person who posted has chosen to be named, and that link does not rotate the way the session identifier does.