If you are coming from Twitter, you understandably may feel like it is a lot quieter here. Mastodon is not "dead," but there is no #algorithm bombarding you with stuff. Here you develop your own user experience by searching out and following people and things that interest you. You can pretty quickly build a much more satisfying experience, but it takes a little effort. Following #hashtags is a good way to get started. Be patient and join in conversations.
Google has a mountain of behavior data on me, and probably knows how viciously anti-diamond I am. Yet these ads regularly pop up in my news feed.
Google charging a client for an ad hit that will never materialize into traffic for their company, and frustrating me to make more money on totally fake metrics.
Both consumers and advertisers are getting squeezed for a more enshittified experience.
I have a process that generates a JSON document (> 1 MB, < 1 GB) once per week. These documents will be pretty similar. Some data will be modified, some will be added.
I'd like to keep all of these documents, in a compressed way, benefiting from the similarities between them, as if I'd compressed a concatenation of all of them, but without having to recompress everything each week.
Ideas? If possible, only using #Python's standard lib.
Question for anyone who is working on self-driving taxis here in #sanfrancisco ... What happens if someone barfs in the car? Is there some kind of barf recognition #algorithm that tells the car to go back to the depot for cleaning? #autonomouscars
The thing is that #RSS is the best. Social media sites don't like RSS because it breeds independence. Which is exactly why #creators should be using and promoting RSS. You can whine about the #algorithm or you can work on freeing your self from its clutches.
I feel that the #fediverse's lack of an #algorithm is more toxic than #Twitter; on #mastodon's #chronological timeline I feel like I'm missing out on important signal in the noise so I have to spend more time combing through it, exploring #hashtags, #following them, #mute words, build #lists, etc. With Twitter there were ads, but also there was all the best stuff since the last time you logged in, sorted by how interesting it would be to YOU! That was awesome; I miss that!
Imho all #governments in all countries on all levels should communicate via the #fediverse, instead of via commercial services. Why should a profit-maximizing feed #algorithm, and an ad placement algorithm get between the citizen and their government in a democracy?
“It is expensive to spread disinformation and propaganda…Every time you boost a post, every time you comment, every time you share a post that is telling the truth or exposing the truth, you are costing those who are spending disinformation money, because they are going to have to spend way more to counter whatever it is you are sharing.”
I think this is correct in the aggregate for #socialMedia that is #algorithm-driven (i.e. basically everything outside the Fediverse). If you set your posts to public, the posts that you share and boost will eventually become recommended to people at the margins, and will cost disinfo vendors money to counteract. https://newsie.social/@LALegault/111427360189023183
The notion of reduction in #ComputerScience, and particularly the notion of NP-completeness lead to surprising connections between a variety of fields.
Two of my favorite NP-complete problems are:
Kidney Donor Matching Markets (better algorithms literally save lives)
Knot Genus (a seemingly very abstract problem in topology)
Like, oh, you thought you were working on kidney donor matching markets? Surprise your #algorithm can be used to find the genus of a knot! And vice versa!
Big caveat here is: NP-completeness of the decision problems doesn't, by itself, tell you the same relationship between the approximation problems, nor between heuristic algos that might not always get the exact optimum, nor FPT. But TheoryCS has different kinds of reductions that can tell you relationships between those things too!
Which makes me wonder if knot genus is hard to approximate...
h/t @CihanPostsThms for their post on the birdsite about knot genus
A digital health company with an app for monitoring maternal health is receiving nearly $1 million to add an algorithm for detecting a dangerous pregnancy complication.
a lot of time you will hear things like “there’s no algorithm here” and “you are your own algorithm”. these are fundamentally unhelpful, as they explain the technological background but not the practical differences, so let’s break this down.
what does a social media #algorithm do? mainly 3 things.
*1. it prevents you from ending up with an empty home timeline when you first sign up. meanwhile, mastodon and many other fedi tools just put you on an empty screen.
*2. when you have run out of new posts in your curated timeline, an algorithm adds additional content. now, how does it do that?
*3. it observes the kinds of posts you interact with, what accounts you follow, who they interact with, etc. to calculate your interests, with some nasty side effects such as amplifying controversy.
so when people say “you are your own algorithm”, they mean that step 3 is what you need to do in order to achieve 1 and 2.
find out what kinds of posts you like and follow those people and hashtags. seek out groups (yes, we have groups!) and follow them, too.
scan local and federated timelines for good stuff. browse other servers’ local timelines.
check out the people getting boosted and linked into your feed and see if they are worth following, because clearly your friends like them! click on random strangers commenting in a thread. you can be sure they aren’t nazis.
instead of training an algorithm to recognise your interests, and training YOURSELF to adapt your behaviour for controlling the algorithm, you need to manually populate your timeline. the trade-off is you are free to interact in any way you like. no unaccountable machine will interpret your criticism as recommendation. no data siphon is analysing your posts for hints on what to try to sell you.
The absence of a curating #algorithm on #Mastodon has an (intended?) effect I only noticed since my newborn keeps me up at night:
Where #chronology rules the feed, I only ever see posts from my own (UTC+1) and adjacent #timezones.
Although I deliberately follow people elsewhere, like @CathyTuttle (based in UTC-8) or @nw (UTC+9), I hardly ever see their posts - except when my baby makes me scroll Mastodon at 3am in the morning. 😴
Does that mean it's impossible to build a global audience here?
Did you know that #algorithm based platforms like #insta have "tools" for creators to know when their audience is usually active so they can plan when to post for more "engagement?"
Guess who can't access those kind of "helpful tools" because it's inaccessible - cognitively broken me
More in this hoot train
Love a #Mastodon rant from my janky insta posts...
I love that posts on Mastodon are in chronological order in the feeds. I feel like I discover interesting accounts that I wouldn't find on other platforms because the other platforms have an algorithm.
Forgive my transgressions if this is the most idiotic thing to be said.
Anyone who has been on fediverse apps is no stranger to the algorithm vs. no-algorithm debate. There certainly is no lack of discourse on the topic.
So I had a tiny insignificant thought about this. Why couldn't we all have a personal decentralized algorithm that allows us to curate our own feed to our personal preferences?
Now I get that hashtags give us a way of curating our feeds. This makes sense to me, but most of the complaints I hear are that it's a lot more 'work' finding topics of interest.
I'm not a software engineer so this might be an uneducated observation. If anyone cares to enlighten me on how/why this would or wouldn't work that would be awesome.
FWIW, I'm pretty satisfied about not having a centralized algo curating for me.