I found that Dewey was borderline acceptable — more categories seemed to degrade it, though admittedly I didn’t spend much time prompt tuning. I also tried tags and these weren’t reliable (not at a consistent “scale” from episode to episode).
I suspect I’ll need a more mechanical approach, long term.
So if there are 100 people whose messages you willingly watch, how do you discover new people? (Maybe you do so organically when they mention them?)
And if you _do_ allow yourself to taste from the firehose of unfiltered messages, a single personal list of users/messages wouldn't scale and you have the classic spam problem. Do you share your rules with others (and form an aggregation like https://www.dnsbl.info/)?
I answer indirectly: you probably know that most "manual tagging" is hyper-efficient compared to automatic one, and you probably here polemics about YT, Meta etc algorithms who push people toward extreme positions, apparently in neutral manners like yes, some video push toward political extreme right or left, but also some young girl video push toward younger and younger and so on..
So how to discover people? Well, in Usenet that happen at a slow but effective peace: you start to choose some groups you think to be interested in, the total number of groups is not that high and names are sufficient for a mere full-text search in a hierarchy. You start participate and others members write about some other groups "you might ask there", "try here" etc. In modern terms this is Reddit subs with a bit more discoverability and less abandoned groups. Since Reddit today seems to be effective enough that skilled people add "reddit" to most google search queries to find better contents...
Then the spam problem: at Usenet eternal september time antispam was limited, nowadays simple filters suffice AND we can import another piece of neglected IT evolution: the concept of PGP/GNUPG with their chain of trust. Or some will keep rotating usernames, some will keep them for decades. Those can share public keys signing each others in a classic chain-of-trust exchanging spam data automatically from their own client. Such approach is limited, but it's still better than actual "lists" for instance. It's still a cohort of people who decide BUT such cohort is not a for-profit company.
Long-story short even with current progresses expert systems are FAR to be near the human selection quality and actually it's possible to get partitioned human selection shared spontaneously to others. This is probably slower in discovering new stuff but bring most high quality results and avoid certain derives, so in the end is better. Actual volumes of posts often named "infodemia" needs to slow down and going up in quality. Eternal september do not show the limit of Usenet but the limit of a certain tech, we can plug-in more to surpass them and keep evolving instead of keeping reinventing the wheel with company-startup-alike experiments who keep duplicating similar concept just varying a little bit.
Competitive means from Oracle/CIA "different teams one against/semi-isolated to the others" produce some results, but a classic endless evolution of Lisp/Smalltalk systems the history prove produce MUCH better evolution because do not stop diversity, but allow diversity to merge and emerge. ALL "recent" IT evolution prove that countless times. ALL "recent" scientific trends do the same.
> Thanks for posting this. Everything old is new again.
Right? It was such a good time :-) You're welcome.
> why usenet couldn't be a viable social media platform
I wonder if identity plays a role here. Centralizing points (likes, retweets, etc) drives people to work on getting attention with their posts, to drive engagement, but also invites troublemakers and controversy.
How was poster identity handled in Usenet/NNTP? From what I remember, it was just a "From:" header and spoofing was easy. Or was there more to it? (Maybe because most posters wrote to their local server which require auth, you could see which server the message originated from and decide if the sender's address matched the server address...? It's been so long.)
If not, then Twitter, Mastodon, etc. all seem to have a somewhat strong notion of identity.
> If not, then Twitter, Mastodon, etc. all seem to have a somewhat strong notion of identity.
Some of us would sign Usenet messages with PGP/GPG to deal with this. Lots of user didn't know what to do with the "geek code block" at the bottom of the message.
