tinysubversions.com

Here's a future scenario: you tell your family members, oh, here's a link to my Twitter account, you can press the "follow" button to follow it from Facebook, and I'll put you on a private list of people who can see my kid photos I post there. It's not because Facebook has features I like or I trust Facebook or anything else. If I could follow her Facebook updates from somewhere that was a better overall fit for me, and she could follow me back from her Facebook account, I would do so in a heartbeat and delete my own Facebook account. But the point is the loop isn't "more people attract more people", it's "more people begets better software which attracts more people".

We only care about "Pages" so: * grab a random Wikipedia Category ...that has a plural noun in its title ...that has at least 16 "Pages" * ...that has a plural noun in its title * ...that has at least 16 "Pages" * grab a random Wikipedia Category ...that has a plural noun in its title ...that has at least 16 "Pages" * ...that has a plural noun in its title * ...that has at least 16 "Pages" * get the list of Pages in the Category * get the first 3 letters of a random dictionary word * grab 500 Wikipedia Categories with 16+ members in alphabetical order starting with those 3 letters ...use a part of speech tagger to only keep ones that have plural noun in their title ...and that have at least 16 "Pages" * ...use a part of speech tagger to only keep ones that have plural noun in their title * ...and that have at least 16 "Pages" * randomly pick one of these narrowed-down categories * get the list of Pages in the Category * pick 16 of them at random and draw them on the bracket * throw away anything with less than 16 Pages * throw away a few other "banned terms" that gave consistently boring results (for example, anything with "articles" in it since that is usually a category that is a list of Wikipedia articles) * from the categories that remain, pick one at random * make a GET request to this endpoint to get all the Pages in the category * pick 16 and draw them on the image For very specific details, you can find the complete source code on Github.

It seemed like the easiest thing was to ask people to copy/paste any tweet from a thread, and then automatically figure out the entirety of the thread for the user, rendering it as a complete blog post. So if you imagine a Twitter thread as a family tree, full of branches, if I start with one tweet, I can tell you its parent and its grandparent, but I can't tell you its cousins or siblings or, crucially, its children. At the moment, these are the rules: * if a tweet has a line break inside it, of course render that line break * if a tweet has embedded media, break to a new paragraph * if the current tweet's parent starts with an uppercase letter, and the current tweet ends with a terminating punctuation mark, break to a new paragraph I can't provide a rational explanation for the last rule. And it turns out the lazy programming decision (query Twitter about every tweet every time) is the most compliant with the Twitter terms of service and probably the best thing for user privacy.

A tool that turns Twitter threads into blog posts, by Darius Kazemi. ← Tiny Subversions → Support Spooler and my other tools! The thread will appear as a blog post here. It only works for a single person's thread, not replies from other people.