Show HN: I built a tool that helps predict HN front page success
wannalaunch.comHey HN community,
I built a tool that helps optimize your post for hitting the first page of Show HN.
How it works: I used a Hugging Face dataset of all Hacker News posts from the past 3 years and trained a model that predicts how successful your post might be. There's still a lot of randomness on HN, so nothing is guaranteed, but the tool helps optimize your post for higher odds.
A couple of interesting findings:
- GitHub repo links work x3 better than regular domains - Open-source tools have a steady virality rate (13.9% - one of the highest) - "I built" outperforms "We built" - Using parentheses and mentioning technologies (Lua, Postgres, Rust, etc.) helps a ton.
You can try the tool at wannalaunch.com or read the blog posts for more insights from the analysis. The model is also available as open source if you want to retrain it or look under the hood.
Happy to hear the feedback! Funny how different people have different understandings of "success" :) This seems to mention the score and how likely something is to be on the frontpage, none of those things would mean "success" to me. The value from Show HN isn't from the eyeballs, your website analytics reaching higher than before or the score of the HN submission, the value sits in the conversations and discussions you'll end up having in the comments, how you think about all of those things afterwards and what you end up acting on. There are few communities where you can (hopefully constructively) criticize a project and also receive proper criticism of your project, usually backed by real arguments rather than just emotional pleas, HN is one of these, and I'd say Show HN is the place where you can really receive good and actionable feedback as long as you're also able to look past and ignore the less thoughtful comments. Does anyone really get good advice from hackernews or do regulars here just have an inflated sense of self importance? I think focusing on intended users / customer base for feedback is better than a bunch of students and wannabe entrepreneurs who will never actually buy your product > Does anyone really get good advice from hackernews Yes, absolutely. Over the years I've probably done 10-15 Show HNs that had comments that had direct impact on either features or the whole direction of the project, to great effect. Sometimes it's the comment itself, sometimes it's the thoughts that the comment spawns, doesn't have to be direct. I think getting feedback from intended users is needed as well, but it doesn't replace more honest and forthcoming feedback which you'll receive from HN, but feedback from HN won't replace feedback from users coming from other sources. That is true and I agree. I guess in this case the "success" I meant was hitting and staying on the front page so that more people can see your project. Well, the amount of valuable comments that would form a good discussion is highly correlated with the time a project being on front page, isn’t it. Well, kind of but also not fully. Yes for the "global" frontpage, but for the /show list, your submission can stay on the "front" there for almost a full week, and with proper discussions already establish the first day or two, you still get valuable insights after days, even after it drops off. But I think that requires eager participating from the submitter as well, not just drop your project and not replying to anyone, or not replying with an open mind/required perspective. >for the /show list, your submission can stay on the "front" there for almost a full week Good to know this and yes, ShowHN gathers more discussions with the author. One question: is it still true, that a submission can stay for days on /show for days? I read here recently that with the LLM trend of creating software /show became very overcrowded. As of writing this comment, this is item #30 (last) on /show: > Show HN: Live Sun and Moon Dashboard with NASA Footage (lumara-space.app) - 222 points by beeswaxpat 5 days ago | flag | 70 comments So yeah, definitively possible for it to stick around multiple days. Maybe not a full week anymore like it used to in the past, but still a long time. Conversations and engaged discussions can continue past that too, as authors check for replies then reply themselves, and so on. Kind of funny to have your own post be 49/100 Show HN: I built a bicycle that runs on my fridge (written in Rust) https://github.com/bike-fridge Show HN 100/100 Well, the model can suggest patterns, but everyone is responsible for what they post. Interestingly enough, you're #1, but your model ran on: - Title: Show HN: I built a tool that helps predict HN front page success - URL: https://wannalaunch.com/ Only gave you a 49 out of 100. And gave this advice: `Consider adding a parenthetical. Top Show HN titles often include one: "(written in Rust)", "(open source)", "(YC W25)".` That's my second try, my first post actually got shadowbanned, so I had to change the domain (it was originally GitHub) and the title wording for this one. There are still a lot of random factors on HN, so nothing can predict things perfectly, but I like that the model gives you feedback based on real data behind it. Let's see if this work. I just posted something that got a score of 83/100. I doubt it will. From everything I have posted, it's quite random what sparks peoples interest. It really depends on many factors such as what the topics where in the last weeks and on current events. > A couple of interesting findings: Heh, I'll add a few: - Apple finds LLMs can't [blank]: +100 points - Something AI something bad / makes you stupid / can't think / losing your touch: +200 points - Something spaceman bad: +30 flags in 10 minutes - Something simple but sounding modern (1992): +1000 points Simply adhering the Show HN guidelines is a much better predictor of success than any of that: https://news.ycombinator.com/showhn.html Seems like you should've linked to https://github.com/mrimek/hn-virality-predictor ? I just stopped my stopwatch and it took 18 mins for this post to get cancelled and removed from the front page because....it won the front page. Good game. The formula is pretty well known but it does not work in this forum due to tight moderator control and shadow banning. Idk how to make it more obvious that you’ve used a vote ring than 15 votes in ten minutes. Haven't used anything - I'm as impressed as you are. I think, it's better to focus on the contents then the formula. you should have posted the github link and mentioned that it was built using React then.