Hacker News “Who is hiring?” — how to actually search a 600-comment thread

Once a month, a single Hacker News thread quietly becomes one of the best job boards on the internet. Founders post directly. Small teams post directly. The listings are unfiltered by an ATS, unpolished by a recruiter, and — more often than any job board you've used — they actually state the salary and whether remote is real remote. Then you open the thread and realize the interface is a wall of text with no search box, and the best job board on the internet is also one of the worst to use.

The best job board has no search box

The monthly "Who is hiring?" thread is exactly what it sounds like: a top-level Hacker News post where anyone can reply with an opening, and anyone can reply to those replies with questions. No login wall, no application form standing between you and the person who wrote the post, no keyword-stuffed job description written by a template. You're reading the actual words of someone trying to hire, which is rarer than it should be.

That's also exactly why it's unpleasant to search. It's hundreds of comments on one page, sorted by nothing more meaningful than posting order. There's no filter for "remote," no filter for "backend," no filter for "posted in the last hour" the way there is on a real board. Threads inside threads carry follow-up questions, "is this still open?" replies from three weeks into the month, and the occasional repost of a listing that already appeared once. You're not browsing a job board. You're reading a very long, very unstructured group chat and trying to extract postings from it — and because there's no applicant count shown anywhere, you have no reliable way to infer your place in the pool or a review advantage from when you respond. Use the post date as a freshness signal, not as a promise of recruiter preference; the same evidence boundary matters when interpreting public applicant counters on LinkedIn.

Ctrl-F patterns that actually work

Before reaching for anything fancier, the plain browser find-in-page (Ctrl-F or Cmd-F) gets you further than it should, if you search for the thread's own conventions instead of your own words.

• REMOTE — the single most useful search in the whole thread. Posters have converged on writing REMOTE in caps near the top of a listing specifically so people can find it this way. Searching lowercase "remote" catches it too, but also catches every mention buried mid-sentence in an onsite posting explaining why they don't offer it.
• ONSITE — the inverse filter, useful for ruling postings out fast if you've decided remote is non-negotiable.
• A location string — city or country name, if you're anchored somewhere and want to skip past the flood of US-only listings.
• $ — catches the postings that actually state a salary or range, which is a meaningfully smaller set than the postings that don't, and worth reading first.

The other habit that saves real time: read top-level comments only, at least on the first pass. Replies nested under a posting are almost always questions, clarifications, or someone else's "same, but for our team" — rarely a second independent listing. Skimming every reply chain in full is how an hour disappears without you finding more postings than you would have in ten minutes of top-level-only scanning.

Tools people built to cope

Enough people have hit this same wall that a few tools exist specifically to make the thread searchable, and they're worth knowing about.

Hacker News's own Algolia-powered search can be scoped to a single story, which turns the thread into something with an actual query box — search for a keyword and get matching comments back instead of scrolling. A handful of community-built "hnhiring"-style trackers go further, pulling each month's thread into a searchable, filterable list with tags for remote, visa sponsorship, or tech stack pulled out of the free text.

These fix the part that's genuinely broken: finding the comments that match a keyword. What they can't fix is relevance to you specifically. A filter can tell you a posting mentions REMOTE and Python. It can't tell you the role wants eight years of experience against your two, or that the "small team" is actually a 400-person company's new AI division, or that the stack they buried in paragraph three is one you'd rather avoid. Filtering narrows the pile. It doesn't read the posting the way you would.

Scraping it yourself

Because the thread is Hacker News comments under a story, it is public and structured enough to pull programmatically instead of scrolling by hand. The official Hacker News API represents stories and comments as items with IDs, timestamps, text, parent links, and child IDs; clients must traverse the comment tree themselves. Once the relevant comments are structured, every Ctrl-F pattern above can become a real query. Deleted or changed items and inconsistent free text still need explicit handling.

Or let the reading be done for you

Scraping solves the finding problem. It doesn't solve the reading problem — someone still has to open each matching posting and decide whether it's actually worth your time, which is the same judgment call a keyword filter can't make for you.

That's the gap Oink is built to reduce. Hacker News can participate alongside broad boards, specialist sources, employer pages, and open-web discovery when those paths are enabled and healthy. Viable candidates are ranked on Oink's 0–5 scale with fit reasons and concerns. Source participation and enrichment depth can vary, so the job-source guide avoids a fixed count and documents the failure modes instead of implying that every source is searched every day.

The jobs are in there. You just shouldn't be the parser.