Comment by bluGill
1 day ago
A good algorithm is a good thing. However what a good algorithm is for me is often different from what it is for those who maintain them. Outrage gets attention and sometimes it is needed, but there is a level of too much, and also a lot of outrage unfairly represents the issues and so it makes me mad even though if I understood the details I wouldn't be mad just concerned.
I want an algorithm that surfaces things of interest to me, then says "you have seen it all, go outside" (with an option of if I'm confined to a hospital bed to go on). Algorithm maintainers want me to keep scrolling for more ad dollars.
> you have seen it all, go outside
Or "you've seen it all. Bored? Click here to let your friends know you're looking for something to do/see who else is bored". Or "Bored? X needs volunteers!" Or some other positive suggestion to try to prevent a "eh guess I'll doomscroll something else" reaction.
That would be such a killer feature. And it could find other friends in the area and that are also free. Not like Meta doesn't have some kind of model for all that data already probably.
Yes. I'd love to find other nerds into retro computing, UNIX, and pottery in my area without wading into groups or joining a forum. They know everything about me, match me up with some people I'd most likely get along with!
1 reply →
> That would be such a killer feature.
"Alice, what if we made a button that improved overall human wellbeing, while somewhat reducing our ad-revenue and lowering the engagement-metrics we use to sell shares to investors?"
"*sigh* We've been over this, Bob: We only build features for customers--not cattle."
A long long time ago before reddit, facebook, digg, twitter, etc, there was usenet. It worked a bit like reddit but subreddits were called news groups.
There were many front ends for usenet, called news readers.
My favourite was "nn" short for "no news".
It showed you posts in groups you're subscribed to, allowed you to post comments, etc.
When you had finished getting up to date it would EXIT and print:
No news. (Is good news)
Ha! I had forgotten that message.Thank you for reminding me. I used to read comp.lang.lisp for the extensive and increasingly bizarre flame wars and for the wider philosophical discussions. Eventually I got to the point where I thought "OK I'm done now" and left and never went back.
The cool thing is: Its still there!
Yeah, it may be not as populated as in the 80s to mid 90s, but there are still enough active groups in usenet to waste uncounted hours every day...
You should read about OPML blogrolls [1], they are gaining traction in this space. Personally, I like the idea of manually exploring recommendations, so I built a browsable index [2]. But you can crawl the these as well and build all sorts of recommendations engines.
[1] https://opml.org/blogroll.opml
[2] https://alexsci.com/rss-blogroll-network/discover/feed-c550c...
Thanks for the blogroll love!
You can also see one in action on my blog's home page.
http://scripting.com/
And on a special site..
https://blogroll.social/
A blogroll is a kind of feed reader.
I'd like to see an RSS item-level recommendation / discovery algorithm driven by recommendations made by a cohort of people who have similar likes to me. I wish there was a standards-based to publish a stream of my "likes" from my feed reader, and to "consume" the "likes" of others. When I add somebody's blog to my feed reader I'd expect it to pick up their "likes" (much like how people used to have a "blog roll" on their own site) and begin to consume the "feeds" from the sites they "like". It reminds me a bit of PHP's "web of trust".
Hacker News, arguably, functions in this capacity for me now. The cohort is the entire population (since we all see the same item rankings), though.
What you are describing is similar to how https://LinkLonk.com works (my side project) - when you "like" a link you get connected to the RSS feeds that posted that link and other users that also liked it. Then you get content from feeds and users that you are connected to. The more links in common you have with a feed or a user the more weight their other links have.
Do any of the web-based readers offer discovery based on matching your RSS list to other users' lists? Mine (Feedbin) doesn't, but I'd like it as an opt-in option.
Bluesky is trying to figure out how to outsource algorithms and let you decide which to use.
Highly recommended podcast episode: https://oxide-and-friends.transistor.fm/episodes/scaling-blu...
You said what I wanted to say.
I think there is a niche market for tools that allow individuals to train their own recommendation systems.
You seem to be behind on English as spoken. From what I can reconstruct:
Algorithm (n) - a secretive set of systems, procedures and data that Big Tech uses to maliciously manipulate unsuspecting general public. Example usage: "Algorithm-free music discovery app for DJs"
I'm not joking, that example usage is taken from a live example.
There's been some conversation on Mastodon (and here) about its lack of algorithm
I really prefer my feed with no algorithm. I really like that it's just ordered by when it was posted, and if someone spams too much I'll remove them, and if my feed gets too much I'll curate it down a bit.
> I want an algorithm that surfaces things of interest to me, then says "you have seen it all, go outside" (with an option of if I'm confined to a hospital bed to go on).
[Acquire, or Employ your] good taste, sensibility & discipline.
Edit: For the record, "Employ your..." assumed that it if "good taste, sensibility & discipline" was not already acquired, it was already possessed and who I was responding to is able to put it to use.
Let those characteristics be your algorithm...or rather, your natural heuristic for living fair.
Has good faith met the end that it's said that chivalry saw?