Ask HN: How can I improve how I Show HN?

There are give or take 400 submissions per hour. The vast majority of that receives 1 or 2 points. They never make it our of the new submissions bin. That you projects don't receive attention is no slight on you, it's just that percentagewise almost nothing makes it to the front page.

A repost once in a while, appears to be acceptable though.

Here are some statistics: http://readcodelearn.com/notes/when-to-post-to-hn.html

> Is it because on HN, people are after dev tools, not studying tools?

Pretty much. It's mostly the wrong audience.

HN is interested in things that are either aimed at a technical audience, or where you can show some really unique way of building something so that even if the end product isn't relevant to a technical audience, that the process used to build it was.

There'll always be exceptions, but that's the general rule I've observed. And most of the exceptions are things where there's some other aspect of the product that makes it interesting to HN or some aspect of how it was built that makes it interesting to HN, or where the target audience happens to overlap with HN type users to a high degree.

I looked at the show HN in your history and I would've skipped over it and not read it. If it was a technical blog post about how you solved something that was both interesting to me and seemed novel, then I might've clicked. (But if you had that information buried inside something that seemed like "just" a product for learning, I wouldn't have clicked)

Try product hunt, twitter, tiktok, and other places.

To quote Paul Graham:

> If you want to appeal to Hacker News readers, tell them how something complicated works. Especially if most people either don't know, or are mistaken about, how it works.

Source: https://twitter.com/paulg/status/1553708274465140736

I believe you when you say there is interesting tech running the site, but I couldn't find evidence of it in a few minutes of clicking around. So there's not much for me to engage with. A blog post explaining how the tech works might get more interest here.

I tried a demo quiz on studywand, but it seemed worse than typical manually generated quizzes. Here's an example:

  Q: "What is the principle behind operant conditioning?"
  A...  is a process of gradually increasing the intensity of aversive stimuli in order to reduce their impact on a person's emotional state.
  B...  operant conditioning is the technique used to reinforce positive behaviors and punish unwanted behaviors.
  C...  is a plan that the therapist and client create together.

You could know nothing about psychology and choose B because it's the only one that's an answer to a question of that form. A good quiz question tests for a more precise understanding.

If the tech is doing something that couldn't be done before, you should draw our attention to it!

Thank you, this bug has been fixed. We recently changed from shuffled distractors to Ai-generated distractors, but as you identified, such a bug could occur.

Specifically, we used semantic similarity for shuffled answers (so they are close and not too easy), completely shuffled answers, and AI-generated distractors. We are now going to choose the most popular default (based on retention during A/B/C testing), and also offer the mode as a choice.

The quiz question (though not answer, since overwritten) from the demo you posted was generated in 2020, with T5 not GPT-3. Other and more complex questions were also possible with our finetuned tech on a bespoke carefully collected dataset. It was at the time much more impressive than it is now with GPT models, and we had it out before competitors even launched(the closest thing was old style GOFAI and a few online tutorials on finetuning T5 for similar tasks that had inferior data).

The technical bits need a hook - for things like "details on how the C compiler is actually a devil" the hook is already present for many people on HN, they know what C is, they know the compiler, they're primed to be interested.

For something more offbeat the technical part isn't going to sell it (unless you write a blog from the language viewpoint: "How I used Rust to beat an LLM into submission" or something).

You might want to get someone else to read up and play with it and write the post for you, as coming in from the out side.

And even then, doing everything right, you likely will languish in the new queue unless the lightning strikes and everything lines up.

Really good points on the angle there, thank you. And important point that it shouldn't necessarily come from you! (that's what's worked better for us everywhere but ProductHunt, reflecting on it)