A 2013 Post on How HN Ranking Really Works Is Trending Again. Should You Launch on Hacker News Before Applying to YC W27?

Today's Hacker News front page has a strange guest at #3: Ken Shirriff's 2013 deep-dive on how HN ranking actually works, sitting at 176 points thirteen years after it was written. A few slots above it, CUA-S1, a computer-use model from Cua (YC X25), is running a Show HN. Supabase (YC S20) is occupying the hiring slot. And if you scroll the last few weeks, W26 companies like Skillsync have been using front-page launches to announce themselves mid-application-season.

If you're preparing a YC Winter 2027 application right now, the obvious question is the one founders keep pasting into ChatGPT and Perplexity: should I launch on Hacker News before I apply to YC?

The short answer: launch if it gets you users you can measure, not because you think YC partners are counting your upvotes. Here's the longer answer.

Does YC actually look at your Hacker News launch?

YC built Hacker News. Partners read it. But there is no field on the application that asks for your karma, and no reviewer is cross-referencing your Show HN score.

What reviewers do look at is the traction section, and this is where a launch matters. "We launched on HN, hit the front page, and 40 of the 3,000 visitors became weekly active users" is a real sentence with real numbers in it. "We have a waitlist" is not. The launch isn't the credential. The retention curve it produces is.

There's a second-order effect worth naming: several recent YC companies were noticed partly because their public launches demonstrated they could ship and attract developers before anyone funded them. OneCLI (S26) walked into interviews with 300K downloads. That kind of number usually starts with a public launch somewhere, and for dev tools, somewhere is usually Hacker News.

What the ranking mechanics mean for your launch

Shirriff's article, the one trending today, is worth 15 minutes before you post. The mechanics haven't changed much since 2013, and they explain most launch outcomes founders find mysterious:

• Gravity beats points. Score decays with time to the power of 1.8. A post that doesn't catch votes in its first hour or two is effectively done. This is why reposting a flopped Show HN a few weeks later is both allowed and common.
• Penalties are invisible. Posts can be flagged, and heated comment threads trip what Shirriff called the flamewar detector: more comments than points is a bad sign. A controversial title can sink a good product.
• Show HN is the right door. It has its own rules, its own audience, and more tolerance for early, rough products than the main feed.

The practical takeaway for a W27 applicant: a launch is a few hours of variance stacked on top of whatever your product actually is. Treat the front page as a distribution lottery ticket, not a milestone. Plan for the version where you get 12 upvotes, because that version tells YC nothing bad about you. Nobody at YC has ever rejected a company for a quiet Show HN.

When launching before you apply helps

A pre-application launch is worth it in three situations:

Your users are on HN. Dev tools, infra, agents, anything with a GitHub repo. The people who upvote are the people who install. Cua's Show HN today is exactly this: the audience and the customer are the same person.

You need usage data for the application. The W27 form will ask how many users you have and how you know people want this. A launch three or four weeks before you submit gives you a cohort you can actually report on: signups, week-two retention, the two features people asked for. That's application material you cannot write any other way.

You're testing the pitch, not just the product. The comment section is a free, brutal preview of your YC interview. If 30 commenters all ask "how is this different from X," a YC partner will ask the same thing in minute two. Better to hear it in September than at the interview.

When it doesn't help, or hurts

Consumer products mostly waste their launch on HN. If your buyer is a dentist or a teenager, the front page gets you a traffic spike with near-zero retention, and a misleading one at that. Launch where your users are and report those numbers instead.

Don't confuse a traffic spike with traction. The most common failure mode: front page, 15,000 visits, 200 signups, 4 active users a month later, and an application that leads with "#1 on Hacker News." Reviewers have seen that movie. If you cite the launch, cite what survived it.

Don't save your "Launch HN" for later. Some founders sit on their launch because they've heard YC companies get an official Launch HN during the batch. They do, and it doesn't conflict. A Show HN now and a Launch HN during W27 are different posts with different purposes. Shipping publicly twice is fine. Waiting in stealth until Demo Day is the actual mistake, and it's one YC partners have been telling founders not to make for a decade.

What to do in the weeks before the W27 deadline

If you're applying this cycle, the sequencing that works: launch at least three weeks before you submit, instrument signups and retention from day one, spend the following weeks talking to the users the launch produced, then write the application around what you learned. A launch the night before the deadline produces a number with no story attached.

And before you submit, pressure-test how you're presenting that launch data. This is where most applications leak: founders either undersell real traction or lead with vanity numbers. Getting a YC alum to read your draft, someone who has written the traction section that worked, is the fastest fix. That's what we built YC Roaster for: free, blunt feedback on your application from founders who've been through the batch.

The bottom line

Should you launch on Hacker News before applying to YC W27? Yes, if your users read HN and you have time to convert the spike into retention data before the deadline. No, if you're doing it to impress reviewers with a screenshot of the front page. YC has read the ranking algorithm too. They know exactly how much an upvote is worth.