Show HN Teardown: Front-Page Launches by Title, First Comment and Timing

What do the sources say a front-page Show HN looks like?

The sources describe a Show HN that does well as a project other people can try, with a neutral title and a short comment from its maker that explains the problem behind it. They agree on that shape and disagree on nothing important, but none of them ties title, first comment or timing to measured outcomes.

• A Show HN is for something you made that people can try. Blog posts, landing pages and sign-up pages are regular submissions.
• Titles are plain: no capitals for emphasis, no exclamation points, no praise.
• The maker adds one calm comment about who they are, the problem and what is unfinished.
• No hour is magic. Post when you can stay and reply.
• One analysis of 10,000 sampled Show HN submissions puts reception's predictive power for survival at an R² of about 14%, which leaves most of the variation unexplained.

The strategy for this page planned a reproducible dataset from the public Hacker News API. That dataset was not part of the evidence supplied, so this page does not pretend to have one. It offers a sourced reading, the arithmetic on the figures that exist and a method you can run.

What counts as a Show HN?

A Show HN is a submission about something its maker built that readers can try. Syften's guide reads the official rules as excluding blog posts, newsletters, lists, sign-up pages and landing pages, because none of them can be tried.

The same guide says a Show HN does not need polish, but it should be non-trivial, personally made and ready for feedback. It works best, the guide says, for dev tools, apps, hardware, open source projects and experiments.

The label has a documented start. The Show HN: Show HN thread, dated 3 July 2014, shows Show HN being introduced as a site feature, with 1302 points and 166 comments. For the exact current wording, read the official Show HN guidelines, which this page treats as the authority over any summary.

This answers a question that recurs in Hacker News threads: whether lists and reading material belong in Show HN. Per Syften's summary they do not, and a blog post about your product goes in as a regular submission.

What data does this teardown rest on?

It rests on three written sources read on 20 October 2026 and on the titles of three Show HN threads. The sources are the Show HN: Show HN thread, the Analyzing 10,000 Show HN Submissions write-up and Syften's Hacker News posting guide.

It does not rest on a pull from the Hacker News API, a sample of front-page posts, or any first-hand test. No points, rankings or first comments for individual launches were supplied, so none are reported. The table below shows each claim the page makes, where it comes from and how strong it is.

Syften offers Hacker News alerts, so its guide is a practitioner's view and is labelled that way throughout. The Show HN thread and the data write-up are the only primary or near-primary material here.

How many Show HN projects are still alive?

The only survival figures in the evidence come from the Analyzing 10,000 Show HN Submissions write-up, which checked the top 100 by hand and a random sample of 10,000 by script. Its author found 11 of the top 100 submissions dead and the other 89 still delivering what they promised, so the top of the list is mostly alive.

For the sample, the author requested each submission's URL and treated a 200 response as alive and every other result (301, 404, a timeout and others) as defunct. The write-up gives no date for the analysis in the text read here, so treat the figures as undated.

The author then regressed alive or dead status on variables including Hacker News reception and reports an R² of about 14%. That leaves about 86% of the variation in survival unexplained (100% minus 14%), even though the write-up describes the predictive power as good.

Two limits follow from the method, and both are inferences, not statements in the source. A redirect counts as defunct, so a project that moved to a new domain looks dead. A live URL also does not show a live product. The write-up adds that Stripe and Dropbox started with Show HN, and that front-page appearances are a small fraction of submissions.

Who holds the top Show HN scores, and can you repeat it?

The top scores are held mostly by people with one big project, not by repeat launchers. The write-up says that in the top 100 the most common number of projects per author is one, and that even for the few authors with more than a dozen projects most of the score comes from a single submission.

The write-up lists the top 10 authors by total score. The table keeps its submission counts and totals, drops the usernames, and adds one derived column: total score divided by submissions.

Adding the columns gives 63 submissions (7 + 9 + 9 + 1 + 4 + 5 + 1 + 1 + 1 + 25) and 15,304 points (3589 + 1472 + 1464 + 1386 + 1318 + 1306 + 1302 + 1215 + 1172 + 1080). That averages 242.9 points per submission (15,304 / 63).

Four of the ten authors have a single submission, and each of those averages above 1,170. The author with 25 submissions averages 43.2. The reading is ours: a Show HN score looks like the result of one launch that landed, not of a repeatable routine, so planning around a repeat of the top score is planning around luck.

One cross-check is possible. The rank 7 row (one submission, 1302 points) matches the 1302 points shown on the Show HN: Show HN thread. We infer it is the same post, because the write-up does not name the submission in the text read.

How should a Show HN title read?

A Show HN title should be plain, close to what the thing is and free of promotion. Syften's reading of the official guidelines says to avoid uppercase, exclamation points, praise, site names, gratuitous numbers and editorialized wording.

Syften gives two pairs. A bad title, “10 Amazing Lessons From Our Product Launch!”, becomes the better “Lessons from launching our database migration tool”. A bad “We built the best AI agent for developers” becomes the better “Show HN: A local agent for reviewing pull requests”.

The guide adds that titles are shared labels for a discussion, not ad copy, and that users or moderators may change a title that is misleading or too promotional. Expect your wording to be edited if it reads like a pitch.

To see what plain looks like, the table reads three real Show HN titles from threads in the research and one illustrative title from Syften. The research does not say whether any of the three real posts reached the front page, how many points they earned or whether the maker wrote a first comment, so they are title examples only and prove nothing about what works. Character counts are ours, taken after the “Show HN: ” prefix.

What can be observed: none of the three real titles uses capitals for emphasis, an exclamation point or a leading number. What can only be inferred: the third title states a claim without saying what the tool does, which is the pattern Syften's “best AI agent” example warns about. The threads are the HN Simulator post, the AI powered reader post and the self-learning AI post.

What should the first comment on a Show HN say?

The first comment should give the thread a place to start, not repeat the landing page. Syften advises adding one calm comment after posting that covers four things, and the list below turns them into a checklist.

1. Who you are and why you made the thing.
2. The specific problem, constraint or odd observation behind it.
3. What is new, incomplete or technically interesting.
4. The kind of feedback that would actually help.

Syften says a bad first comment asks people to support you, explains how big the market is or turns the thread into a sales page. Its reasoning is that readers do not need to be told the market is large, they need a reason to care.

This section has no examples of real first comments, because the research contains none. Anyone claiming to tear down first comments needs a sample of them, which is the gap the method section below is built to close.

When is the best time to post a Show HN?

No source here finds a best time, and Syften says plainly that there is no magic hour. Timing can help at the margin, but title fit, topic fit, early discussion, flags and moderator action matter more.

If you must choose, Syften suggests posting when the US technical audience is awake and you can spend the next few hours replying, which for many founders means a weekday morning in US time. Its practical default is 9am to 12pm Eastern, and it calls that availability planning, not a ranking trick.

For a team in India that window falls in the evening. The conversion below is our arithmetic, using UTC-4 for Eastern daylight time, UTC-5 for Eastern standard time and UTC+5:30 for India: 9:00 at UTC-4 is 13:00 UTC, which is 18:30 in India.

Syften also says not to post right before a meeting, a flight or sleep. The point is to be present for the first hours of discussion, since the guide treats early discussion as one of the things that matters more than the clock.

What gets a Show HN killed, flagged or ignored?

A post can be killed by software, by user flags or by moderators, and Syften lists the habits that invite it. Using the site mainly for promotion, asking for upvotes or comments, deleting and reposting after a failed first try, posting a duplicate of a story that recently got attention, disguising a landing page as a Show HN and pasting AI-generated or AI-edited comments are all on its list.

Syften reads the Hacker News FAQ as saying ranking is affected by votes and time but also by user flags, anti-abuse software, demotion of overheated discussions, account or site weighting and moderator action. Moderators can change titles, kill posts, switch unfair flags off and move submissions. The guide also describes a second-chance pool, where moderators and a small number of reviewers pick older submissions that fit the spirit of the site and software can place them on the lower part of the front page.

The practical consequence, in the guide's words, is that two posts with similar points and age can behave differently. It adds that karma does not rank a post higher, and that emailing the moderators is for moderation mistakes, not for asking for promotion.

The guide also warns against old advice on lining up helpers, using several accounts or writing catchier titles, because those ideas conflict with the current rules. Its blunt summary is that many good submissions die and that Hacker News is not a predictable launch channel.

How did the first Show HN thread set the tone for comments?

The first Show HN thread set a tone of respect: the thread opens with a call to hold feedback to the Show HN guidelines, and the guideline text was revised during the discussion. This matters for a first comment because the maker is inviting that kind of reply.

In the thread, one guideline about weak work was reworded that day. The original advice to consider saying nothing was replaced with the line that you need not pretend something is good but should not be gratuitously negative, and the post's author explained that substantive criticism is about the work, while personal or snarky remarks are the problem.

The thread also shows what the front page looked like on launch day. Commenters noted that it was crowded with Show HN links and the author predicted it would settle down, which is a reminder that a single day is a poor basis for a conclusion about how the site works.

For a maker, the lesson from this thread is modest and practical: ask for the feedback you want in your first comment, and expect direct criticism of the work. Do not expect sugar-coating, and do not answer personal remarks in kind.

What is documented and what is analysis?

Documented facts are what a named source states, and analysis is what Kanso concludes from them. The two lists keep them apart.

• Documented: the Show HN thread of 3 July 2014 shows 1302 points and 166 comments, and a guideline reworded that day.
• Documented: the data write-up found 11 of the top 100 submissions dead, treated only an HTTP 200 as alive, and reports R² of about 14%.
• Documented: the write-up lists the top 10 authors with their submission counts and total scores.
• Documented by Syften, a practitioner: the title rules, the four-part first comment, the 9am to 12pm Eastern default and the list of habits that get posts killed.

• Analysis: score per submission, the total of 63 submissions and 15,304 points, and the 242.9 average are our arithmetic on the write-up's table.
• Analysis: a top Show HN score looks like a single launch that landed, not a repeatable routine.
• Analysis: a redirect counting as defunct and a live URL not proving a live product are limits of the survival method.
• Analysis: the India Standard Time conversion and the comparison of three real titles.

How do you run your own Show HN teardown?

You build a sample of Show HN posts, record the same fields for each, and keep what you observe apart from what you infer. Hacker News publishes its data through a public API, which the strategy for this page named as the intended source, but this page did not query it, so check the API documentation for field names before you start.

1. Write the front-page rule first. For example, a post counts if it appeared on the front page at any time in its first 48 hours, and write down how you will check.
2. Draw a sample that includes posts that did not reach the front page. A sample of winners alone cannot tell you what separates them from the rest.
3. Record the fields in the sheet below for every post, using the same time zone and the same reading day.
4. Count the titles, comments and times before comparing anything, and show the arithmetic.
5. State the sample size and the date of the pull beside every result. The write-up read here drew a random sample of 10,000 submissions, so treat a sample of a few dozen as anecdote.

The result you are after is a comparison between posts that reached the front page and posts that did not, for each of title style, first comment and posting hour. Until someone publishes that comparison with its data and date, the claims in this page remain a practitioner's consensus plus a thin set of figures.

Where does the Hacker News Agent fit?

Kanso's Hacker News Agent finds relevant Hacker News conversations and recommends what to do next. It never posts, so a Show HN title, first comment and every reply are written and submitted by you.

For the wider decision on whether to post, comment or stay out, read Hacker News for B2B founders: when to post, when to comment, when to stay out. For the reply habits that carry over from Reddit, read writing comments that earn trust.

Show HN examples: frequently asked questions

Sources

• Analyzing 10,000 Show HN Submissions, antontarasenko.github.io
• Hacker News Posting Guide: Rules, Show HN, and Timing, Syften
• Show HN: Show HN (discussion thread, 3 July 2014), Hacker News
• Show HN: I built an interactive HN Simulator, Hacker News
• Show HN: AI powered Hacker News reader, Hacker News
• Show HN: I built a self-learning AI without an LLM, Hacker News