[evanelias]

> They may be as straightforward as fetching a user's follow list and displaying their posts chronologically in reverse. That's still an algorithm.

Correct, it's absolutely an algorithm. Crucially, it involves extra steps to make it scalable without needing a materialized inbox. Or at least that's how Tumblr's reverse-chron dashboard worked from 2009ish through at least 2018, maybe still today.

There were a bunch of optimizations in there to keep the queries fast, and minimize the number of rows that need to be examined and sorted. One key step involved cross-referencing the blogs you follow against their latest post timestamp. For example, say you're fetching posts 11-20 on the reverse-chron feed. The worst-case is that posts 1-20 all come from different blogs, so you can sort the list of followed blogs by latest-post-timestamp, and then only examine/sort posts from the top 20. (That's a slight simplification; you actually need to look at the timestamp from post 10, and examine posts from however many followed blogs posted since then, plus 10 more. I spent months of my life tuning this stuff back in 2011-2012...)

That all said, in the context of dashboard feeds, "algorithm" often just means the opposite of reverse-chron. But even then, the "no algorithms" statement is completely incorrect! Tumblr has a "Best Stuff First" setting which controls whether or not your feed is reverse-chron. For a while in late 2017 (iirc) onwards, this setting was even enabled by default... maybe still is?