> but really they're robots that are mindlessly striving towards being the perfect neighbor

Once conversational algorithms reliably pass the Turing test, I guarantee this will happen. State actors may be doing it already. A non-conversational version of this has existed in online gaming for decades.

The catch-22 in starting any new forum or social network is that to attract new users, there must be old users. Some social networks (notoriously dating sites and reddit) started out with employees acting as many users to make the network seem less empty.

It's already possible to pay humans to do this, but it's likely prohibitively expensive. If machines can do it, suddenly you can have thousands or millions of "users" for almost no cost.

Anecdotally, I once tried to create an account on HN by going through tor - HN allowed my account creation to go through and even let me post my comment.

But my comment wasn't visible to anyone else.

It was all very clever.

I think the HN's simple user interface is deceptively simple - there is clearly a lot more going on than I had assumed.

There are lots of filters based on past activity by spammers, trolls, and so on. But in the scenario you describe, moderators would eventually review your comment and restore it to visibility, assuming it wasn't bad in some noteworthy way.

Users with 'showdead' turned on in their profile (and who have karma > 30) can do the same: if you see a [dead] post that shouldn't be dead, you should vouch for it by clicking on its timestamp, then clicking 'vouch' at the top of its page. This is why we added that feature.

> Users with 'showdead' turned on in their profile (and who have karma > 30) can do the same: if you see a [dead] post that shouldn't be dead, you should vouch for it

Wow. I've been on HN for some time now [1], and that's news to me.

I remember seeing "vouch" before, but not for a long time. I simply assumed you had taken that privilege away from me.

It makes sense to only show it for "showdead" users, but it never occurred to me.

Edit: [1] and it turns out that today is my cake day. Very fitting. :-)

(Afaik. It's not like I work here) HN also has an ability to 'shadowban' users. It looks like everything's working to you (well, the bot code), but no other users see anything from the so called bot.

Relevant xkcd [1]

A fun weekend project would be to utilize GPT-2 [2] to model HN comments; quite the challenge considering the usually insightful comments here when compared to other sites.

[1] https://xkcd.com/810/ [2] https://github.com/openai/gpt-2

On reddit whole sub communities are based on gpt-2, it is just going to increase the noise to signal ratio.

In particular, https://old.reddit.com/r/SubSimulatorGPT2/top/ is pretty entertaining.

I agree the content is actually funny in that reddit tread. I wonder how good it can become with a really large good database. I read in some article of OpenAI a dataset gets better by for example replace names by pronouns.