In March, Hacker News added one line to its guidelines: "Don't post generated text or AI-edited text. HN is for conversation between humans."
That line got 4,229 upvotes and 1,668 comments. On a site where a good story does 500, that is a referendum.
The rule gives one reason, about who is on the other end of the conversation, and makes no complaint about quality or errors.
What the rule protects
The first reading that comes to mind is that the rule is about quality.
If the problem were quality, the rule would be about quality (no low-effort comments, no filler, no restating the article), and those rules already exist. This one bans text that might be excellent.
The value of a comment thread is knowing that a person read the thing, had a reaction, and decided to spend their evening writing it down. Strip that out and the words can be perfect and still worth nothing.
You have felt this. Someone replies to your message with three tidy paragraphs and you can tell, because nobody paid anything for it.
What it costs to look like you tried
For most of history, effort was expensive and therefore honest. A long letter meant someone sat down for an hour. A detailed answer meant someone knew the answer.
That link is now broken. Effort is cheap to fake and expensive to prove.
So communities that run on trust now have to state out loud that a human is required, which they never had to do before.
I think we will see a lot more of these rules, and I think most of them will be unenforceable (HN cannot detect this reliably and does not claim to). The rule works as a norm. It tells you what the room is for.
Where this lands for anyone teaching
Every learning platform on earth is about to hit this, including ours.
When a learner submits an assignment, what are you assessing? If the answer is "the text", you are already in trouble, because the text is free now. If the answer is "did this person build the capability", you need to see something the text cannot fake: the reasoning, the wrong turn, the defence of a choice under a follow-up question.
A lot of assessment was measuring effort as a proxy for learning, and getting away with it because effort used to be hard. That proxy is gone completely.
Read the guideline. It is one sentence. Then go look at your own rubric and ask which line in it still survives a model that writes well.
One email when we publish. Research, product decisions, and what teams report back.
