Andrew Conner

Do you think the issue is one of over-inclusivity, or that a sort of Yarvin-friendly theme existed prior?

Cold plunge / saunas masochists. Breathwork tribes.

Supplement people, nootropics, peptides.

Ayurveda, TCM.

and postrat before that.

for the best, illegibility is a virtue for a reason

I was explaining to my wife how the red light people and woo people are totally different, and she didn’t believe me.

I want to make a map of all the wellness clusters.

There’s western science people. Then, we have the grounding/red light/nasal breathing people. But these are different from the traditional woo people (crystals, etc).

Peaters, keto, anti-BCAA. Who else?

Yeah this works. Though, this is kind of what Twitter does. But they care about engagement, so boost popular tweets more than I would choose to.

What you actually want is a "For You" algo that has good incentives. Instead of maximize engagement, maximize something else. Follows, likes, etc. The algo needs to be careful about accidentally optimizing for bubbles *and* outrage (ex: show an alt-right meme to a socialist, and they hate reply).

What typically happens: you find a couple people in each community that are interesting (from the platform, or from other sources). Those people like/repost in their community actively, thus your second-degree graph starts to include more of the communities you care about.

Examples: you like 3 (disjoint) groups: math, generative art, and finance. Suppose you don't know of the finance community yet. Finding it is the same as randomly sampling from every similar group, and is *very* low signal. What's worse is that it provides intermittent reward — bad dopamine loops.

Conclusion from playing with potential What's Hot algorithms: people say they want to see cool smaller communities, but sampling randomly from smaller network clusters ends up being random and not useful.i.e., the useful pockets you'd like eventually get reflected in the follows/likes/replies graph.

"Awareness month"s exist to sell you things

Teacher: "You are a wonderful student, already perfect in every way. You do not need to earn anything. You are already insightful, charming, and cunning. Your wit knows no bounds.

Complete this 100 question multiplication speed test as fast as you can."

I suggest random loading spinners, delays, and !important's in your CSS to re-align expectations.

Arc'teryx seriously misunderstands my mom.

Not sure, but this was years ago and from my understanding, the project mostly failed.

They had a few videos demoing what a modern OS could be, which was exciting. But turns out that's a hard problem, and Google's appetite for 10+ year projects diminished.

Snazzy! This was sorta a demo years ago for Google's Fuchsia operating system. A bummer it never went anywhere.

The demo I saw allowed for sandboxed apps that could easily move between devices. Have an app on your laptop, instantly loaded on your desktop. Or phone. State was synced real time.

If you reload the feed, it's in the exact same order, and new posts are added to the top.

Thoughts about this? Does this break in any predictable manner?

Example: there's 1000 potential posts to show you, and algo wants to show 100. It ranks the 1000 (based on distance to poster, engagement, etc), picks the top 100, and displays those in a feed. Then, importantly, they're *there* statically —

(more here: https://andrewconner.com/information-overload/). A key part is being able to consume feeds linearly to curate content, and be able to "catch up".

What if social feeds ranked content, but then only displayed it linearly?

Social networks do this because they *want* infinite feeds that intermix new content with old. It provides an intermittent reward, even if you've read most of the existing posts.

I've designed a lot of my content consumption around the concept of separating curation from consumption

Every time you load the homepage you see the "best" posts. Reload, and it's re-computed. New posts emerge through the ranking.

Overall, I think this is bad. You can never "catch up" on your feed. It's infinitely long, and if you wait an hour and come back, you'll see new posts mixed with old ones.

Algorithms do two things: filter and rank. Ranking can, in effect, filter by lowering posts down a feed, but practically, low-ranked posts are never shown.

Digg was the first time I remember seeing dynamic algorithmic ranking, though I'm sure it existed earlier. Reddit did it as well...

Been thinking about timeline algorithm design. Some thoughts, would love feedback.

Algorithms are useful/needed bc the amount of content is too much even if you consider just your follows. ie, do you want every reply? every liked post?

Content from people you don't follow can be good as well!

If anyone wants to play with the data set powering this, it's available here: https://huggingface.co/datasets/andrewconner/bluesky_profiles

Behold is the Bluesky World!

Got "The Bluesky Dataset" from @andrewconner.com and used Graphext (from @victoriano.bsky.social) to visualize all the connections. It looks beautiful! 😍

The best thing is that you can explore it too! 👇

https://public.graphext.com/fe28f76fe67518ab/index.html

Yeah, I scraped it. My current scraper is pretty tuned to my own use case, but let me see what I can do. My only real concern is a ton of people hitting their API really hard.

Yeah. It’s around 32k accounts.

There's already a critical mass of people here, so if you're interested in chatting with other people hacking around on the Bluesky API, go here: