1. not use query params for key entities in the URL
2. to stick user identifiers at the root path! totally unheard of to occupy such an important path at that point!
Taken together, this was entirely novel. Next to nobody did this. Twitter was the one other notable example, and that's literally all I can think of.
The URL bar was so different back then. It wasn't search by default. The average tech savviness of internet users was higher. People cared about URLs despite the fact most websites had garbage cgi-bin query string slop. Lots of folks had personal domain names. People typed URLs and shared them a lot - so this was a big deal, because they were memorable, unlike the other slop URLs at the time.
To give more character - HN's urls would have been considered exceptionally nice back then. The average URL was way worse and was littered with hundreds of query parameters.
A great deal of websites put your session token in the query params. PHP had first class ways of spending "sessions" to all urls. Essentially a cookie. It was disgusting.
GitHub revolutionized urls as a product. Even today, not many companies followed suit.
They're still the gold standard.
2nd level would be yourname.com.
3rd level would be yourname.site.com, like LiveJournal and Blogspot had a long time ago.
4th level would be site.com/yourname, like Myspace had a long time ago, and Facebook had after that, and Github had after that.
Once you sink all the way down to the obscure depths of 4th level, there's not much difference between site.com/yourname and site.com/whatever/yourname
Very few companies do this. Github and Twitter are some of the only ones. And they were certainly incredibly novel at the time.
Back in the 90's and 2000s, URLs looked disgusting. Remember this [1] nonsense?
[1] https://www.php.net/manual/en/session.idpassing.php (see "example 1")
That may be, but the user I responded to said "top level". Anyways, what looks disgusting about:
yourname.livejournal.com
yourname.blogspot.com
myspace.com/yourname
?
This was all before github, so objectively not novel by then.
If I remember correctly, it was also one of the few places sticking to the now-standard passing of the parameters via path rather than the '?' URL query part.
It might not seem like much now, but then the ease and simple beauty of having just github.com/user/repo - not only for web access but also cloning - was definitely some freshness factor.