Hacker News in 2026: a launch strategy for technical products and writing

Every story starts in /new with a single point and has to claw its way out. It needs roughly five upvotes to escape the "new" sandbox into the live ranking pool, and from there a time-decay formula — about (points − 1) / (hours + 2)^1.8 — fights gravity for it. Empirical analyses of the rankings keep landing on the same conclusion: escape comes down to how many of your early readers actually upvote, not how many people you can drag to the page.

That distinction is the whole game. Because position decays with time, you can't win by slowly accumulating votes over a day — you win by converting a high fraction of the first wave of readers, fast. And the only thing that makes a skeptical HN reader upvote in the first thirty seconds is a title that promises something real and a piece that delivers it. There's no growth hack underneath; the hack is the work being interesting to this specific crowd.

The official HN guidelines set the frame for what "interesting" means here. On-topic is defined generously —

"anything that gratifies one's intellectual curiosity"

— and off-topic is most politics, celebrity, sports, gossip, and anything clickbait. Titles should use the original article's title unless it's misleading or linkbait, and you're told plainly to drop the hype and the exclamation points. The ranking rewards curiosity; the guidelines are just that rule written down.

A "Show HN" is only for something people can run on their own machine or hold in their hands — a tool, a library, a runnable demo. Everything else — an essay, a benchmark, a post-mortem — goes up as a plain link with a neutral title. Post a blog post as a "Show HN" and it gets corrected fast; it's explicitly off-topic.

The Show HN rules are specific: the thing must be something you personally built that others can try, it should be "non-trivial" and "deeply personal and interesting to you," and ideally easy to poke at without a signup wall. Blog posts, newsletters, landing pages, and lists are out. If it isn't ready for strangers to actually run, the guidance is to wait. A plain link submission, by contrast, is for articles, research, and notes — original source, flat title, no "Show HN:" prefix.

Here's how the moves trade off, drawn from 2024–2026 HN launch write-ups:

One thing the table understates: even when an HN link is nofollowed or sits on a noindex page (more on that below), a link from a DR-90+ domain still drives real referral traffic and tends to prompt second-order dofollow links from the blogs and newsletters that pick up a front-page thread. The direct link is the least of what you get.

Run every candidate through five gates before you let yourself think the word "submit." Would a senior engineer learn something concrete? Is there a real, arguable delta — a new pattern, a benchmark, a hard post-mortem? Is the tone non-promotional? Is it specific enough to fight about? And can you sit with it for three to six hours after it goes live? Miss any of the first four, or fail to protect the time for the fifth, and it is not an HN post.

1. Would a curious, senior engineer learn something concrete? It has to teach a pattern, show a benchmark, or give real operational detail — original research, technical deep dives, and firsthand experience, not thought leadership.
2. Is there genuine novelty — a clear delta? You're in good shape if at least one is true: it describes a new pattern or architecture (an orchestration pattern, a failure-mode taxonomy, an infra design) with diagrams, pseudo-code, or concrete rules of thumb; it publishes numbers or benchmarks others can reuse or argue with (latency, reliability, cost curves, scaling behaviour); or it's a detailed post-mortem or build log for something genuinely complex.
3. Is the tone non-promotional? No pricing, no sales CTA, no "why you should hire me" baked into the body — that belongs in a tiny neutral byline or your profile, nowhere else. And no superlatives. Launch HN's own guidance is blunt that marketing, sales, and PR-style copy "doesn't work on HN." Drop "fastest," "best," "first."
4. Is it specific enough to spark a fight? Clear, falsifiable claims, named tradeoffs, stated limitations — HN rewards posts where people can argue with the details, not the vibe. You want a couple of obvious question hooks built in ("why this architecture over X?", "how does it fail at scale?").
5. Can you be present for three to six hours after it goes live? A Show HN or a high-impact link needs you answering questions in real time. If you can't block the time, the piece isn't ready, regardless of how good it is.

If you can't answer yes to the first four and block the time for the fifth, keep it canonical on your own site and seed it through GitHub or your other channels instead. There's no shame in a piece that isn't for HN — most aren't.

Title plainly, post on a weekday morning US time, and put your real story in the first comment — not the title. The submission itself shows only a name and a link; the work of persuasion happens in a 200–300-word memo you post the moment it's live.

Before you submit. Confirm it actually fits Show HN — personal, non-trivial, runnable. Strip any unnecessary signup wall for launch day; the guidelines say plainly you get more and better feedback when it's easy to try. And check that your HN username isn't a pure company name — Launch HN advises against it and will even rename you to a personal handle.

The title. Start with exactly Show HN:, then the name and a flat description of what the thing does. Clarity beats cleverness every time. From the source playbook, two that work:

Show HN: Pravda Agents – workflows for long-lived AI operators Show HN: A 728k-line agentic automation system I built solo

No "AI-powered," no "industry-leading," no "10x." That language reads as marketing, and marketing reads as someone to ignore.

The timing. Post on a weekday morning, US time — Tuesday through Thursday, roughly 06–08 Pacific / 09–11 Eastern, when the US and Europe are both awake and the front page is churning. Weekends are slightly easier to reach (less competition) but thin on the founders and decision-makers; if you're after B2B attention, take the weekday.

The first comment — your launch memo. The page itself shows only the title and link, so the first comment is where context goes without you cramming it into the title. Two to four short paragraphs, about 200–300 words: what it does in concrete terms, why you built it, what's technically hard about it (architecture, infra, the tradeoffs and failure modes), and what you actually want feedback on. Close with a plain disclosure — you're the solo builder, the full write-up lives on your site, happy to answer anything. No upvote requests, no "please share," no funnel talk; soliciting upvotes is explicitly banned.

Initial velocity — the line you don't cross. Healthy early velocity looks like your existing followers seeing the link in their normal feeds and some of them clicking and upvoting because the thing is genuinely interesting. What's not okay: asking friends or a private group to upvote, running a vote ring, or telling a cohort to upvote inside a window. HN detects coordinated voting and penalizes both the account and the domain.

🔒 authorize for free to reveal 93 words

This is where the real value lives — the thread, not the traffic spike. Block at least three hours and be hyper-responsive for the first two, when ranking velocity is decided. Answer every good-faith question with something concrete, take criticism head-on, and link almost never: one link to your canonical note in your top comment, then argue on the merits inline.

The mindset sets the tone for everything else. Talk to HN as peers — not "users," not "leads." Launch HN's line is to write as if you're having a drink with a former coworker, not pitching investors: modest, factual, cheerfully self-critical. When someone's trying to learn, help them learn more.

On cadence: three hours minimum, the first two the most important, then check back a few times over the next 24–48 hours to catch readers in other time zones. On substance, technical beats vague every time — answer with the actual config, the actual tradeoff, the actual failure mode; a "we're working on it" non-answer gets downvoted. When someone finds a bug or a hole in your thinking, thank them, say whether you agree, and explain what you'll do or why you chose otherwise. That exchange, done in public, is worth more than the upvotes.

On links, Buska's 2026 HN lead guide is the right discipline: don't drop product links unless the thread is explicitly about tool recommendations, because repeated promo links get flagged and banned. One link to your docs or canonical note in the top-level comment, and then you answer inline like everyone else.

If you can't block three to six hours to sit in the thread, don't post at all. A drive-by Show HN reads as pure self-promotion and goes nowhere.

There's a quiet GEO payoff here too. Make your answers quotable — short rules, clean "if X then Y" statements a model can lift as clean text — because HN is scraped so heavily that a sharp comment on someone else's thread earns you AI-trust with zero links attached.

Roughly one part promotion to nine parts everything else. The guidelines allow self-promotion "part of the time" but ban using HN primarily for it. Moderators have long cited a 9:1 rule, and a 2025 strategy write-up puts it at 10–20% of your submissions. In practice: at least ten substantial comments on other people's threads for every time you post your own work, and one or two self-posts a quarter — not a week.

The official position is short and worth quoting:

"It's ok to post your own stuff part of the time, but the primary use of the site should be for curiosity" — and "please don't use HN primarily for promotion."

Around that, the community has converged on numbers. Moderators have historically pointed at a 9:1 ratio (one self-promo per nine non-promo posts or comments), and a 2020 comment from a YC partner literally described 1-in-10 as the right order of magnitude. A 2025 HN-promotion strategy article lands in the same place at 10–20% of submissions. So the working target is: ≥10 substantial comments on others' threads per self-post, and 1–2 self-posts a quarter if you're otherwise active. Don't fire several submissions from the same domain in a short window, either — per hacker-news-undocumented, accounts and domains that look spammy or like a vote ring get shadowbanned, with everything silently marked [dead].

There's no published karma threshold, but accounts with a normal participation history are far less likely to get auto-killed or quietly buried. So before any first big launch, spend four to six weeks as a citizen: 10–20 thoughtful, technical comments on threads in your area, plus a couple of non-self submissions of things you genuinely admired. You want to arrive as a person who's been around — not an account that materializes only to post Show HNs.

Anything that isn't for the reader. Landing pages and signup walls, thin opinion and generic AI listicles, "v1.3.1 is out" updates, and anything that's really about your consulting offer — all of it gets flagged or sinks. The harder filter underneath: most of what you write isn't for HN at all, and that's the correct outcome.

• Pure landing pages or signup walls. Show HN wants something people can try, not an email-capture form or a fundraiser.
• Thin opinion pieces and generic "10 AI tips" listicles. HN is for material that rewards careful reading; surface-level advice underperforms hard.
• Incremental updates ("v1.3.1 is out"). Only a major overhaul justifies a new Show HN.
• Anything primarily about your services or consulting. It collides head-on with "don't use HN primarily for promotion."
• Re-posting the same note repeatedly. Duplicates get buried, and the FAQ asks you not to delete-and-repost outside rare exceptions.
• Anything you can't be around to defend. A post you abandon reads as a drive-by and dies.

The honest version: most of your output — field logs, small refactors, incremental pattern tweaks — should stay canonical on your own site and GitHub, never going near HN. Reserve it for the rare piece that introduces a genuinely reusable pattern, a benchmark, or a whole system design. Realistically, maybe three to five of your notes will ever clear that bar. Treating that scarcity as a feature is the entire discipline.

Less than the legend says, and more than the cynics think. HN discussion pages now carry noindex, nofollow, and many outbound links are nofollowed, so don't expect raw PageRank. The real payoff is indirect — the front-page spike gets you cited by blogs and newsletters that do link dofollow, it builds your name as an entity, and HN is scraped heavily enough that it's become a high-signal source for the answer engines.

On direct SEO, be sober. With nofollow and noindex in play, and Google treating nofollow as a "hint" rather than a hard block since 2020, there may be some marginal ranking value — but you shouldn't bank on PageRank from HN alone. The indirect path is where the value actually accrues: a thread that hits the front page gets picked up by blogs, newsletters, and X accounts that link dofollow to your site, and the whole episode raises brand and entity awareness — the kind of signal that feeds E-E-A-T-style reputation over time.

For AI search specifically, HN is crawled, mirrored, and re-discussed across the web, and the AI-search players treat it as a high-signal read on developer sentiment and serious technical discussion — Buska's 2026 guide leans squarely on that "high-quality dev audience" framing. The honest caveat: the exact weighting inside any AI ranking system isn't public, so treat the AI-trust benefit as probable, not guaranteed. The right mental model is that an HN launch is an authority spike — one that throws off secondary links and citations when the work is strong, not a reliable ranking lever you can pull on a schedule.

Warm up the account first, hoard your few genuinely novel pieces, and launch rarely. Four to six weeks of real commenting, three to five HN-worthy notes in the bank, one Show HN around something runnable — then at most one self-submission a month, each with three to six hours blocked to live in the thread. Everything else is comment-lab time.

1. Warm up for four to six weeks — 10–20 thoughtful comments on threads in your niche, plus one or two non-self submissions of work you genuinely admire.
2. Curate three to five truly HN-worthy notes using the five-gate test — likely a system post-mortem, a pattern taxonomy, and one or two benchmarks.
3. Plan one Show HN around something runnable — a minimal open-source slice of your system, an orchestration library, a reproducible workflow kit.
4. Schedule rarely — at most one self-submission a month, ideally one every one or two months, mixing Show HN and plain links.
5. For each launch, block three to six hours — follow the title, timing, and first-comment checklist, then work the thread like it's the point, because it is.
6. The rest of the time, treat HN as a comment lab — answer hard questions on other people's threads, and only once people already know you as the person who builds this stuff, let your canonical note show up where it genuinely answers the question.

The people you most want as clients are already in those threads, and they can smell a pitch from the first sentence. The only move that works is the slow one: be useful, specifically and concretely, in a room full of skeptics, until being useful is the thing they associate with your name. Most builders won't have the patience for it — they'll post once, get flagged, and decide HN is hostile. It isn't hostile. It's just honest, and it's the rare place left where honest still wins.