Best question-response pairs were curated weekly and at longer intervals, in a digest. Those were absolutely delightful to read.
TL;DR: Like "Ask HN", but very, very, very different.
You owe the Usenet Oracle a pint of maudlin.
I always thought Zadoc the Priest stories tended to be lazy but maybe it was a victim of it's own popularity and it got hard for everyone to keep up.
It was one of my favourite sites back in the 90's and I used to take the time to vote on digests each week. Tempted to look back at some early ones to see if they stand the test of time...
I'd stumbled across TUO whilst at uni. Two particular oracularities have stuck with me through the years.
One, I've shamelessly borrowed numerous times:
Q: What is the point of anything?
A: The sharp bit at the end.
The supplicant owed the Oracle a picture of a porpoise.
The other was one of the occasions on which the oracle instance broke character.
The supplicant was in what was then still sometimes called Yugoslavia, asking why what was happening there was happening -- a rather bloody and horrific civil war slash genocide.
The oracle's response, paraphrased: I'm just a humble graduate student and really cannot answer your question, but yes this is indeed horrible. Told rather better than I'm recounting it here. The impression of two minds sharing empathy across an anonymous interface stuck with me. It's small example of some of the best that the 'Net can do. (And yes, occurred at the same time that genocidal propaganda was also being disseminated across Usenet.)
Likely some point in the 1992 - 1994 timeframe, most likely 1993, and contained within one of the digests.
(I'll see if I can track it down more precisely.)
But the whole schtick generally: grovel, Oracle, payment, Oracle's backstory (girlfriend, etc.). I guess they're a marker for a fairly small subculture.
Here's to having been there.
Kibo for President!
(We went to college together, though. Worked on the same humor magazine and everything.)
As they allude in TFA, it's faster to read than to write, so people would volunteer to write answers and the good ones would be published. Effectively it was a curated list masquerading as a single entity.
Actually, feeding these to a fortune-like engine (or dict) might be interesting.
The whole archive is 54 MB of HTML, about 34 MB text. Runs 12,765 pages as text via pr(1). About 52 volumes of 250 pages each.