bruno

"mr storylets"

writer (derogatory). lead designer on Fallen London.

http://twitter.com/notbrunoagain


THESE POSTS ARE PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE POSTS OR THE USE OR OTHER DEALINGS IN THE POSTS.


Bluesky
brunodias.bsky.social

nex3
@nex3

I cannot parse the second word in "for you page" as anything other than a mangled possessive and that's why I don't use tiktok


lifning
@lifning

much rather go to cohost, where you're always greeted with an implicit "for you, page:" and then a list of posts from other pages you've asked to see


nex3
@nex3

It curdles my brain to think that the experience of social media for the vast majority of people, which is to say the experience of the internet for the vast majority of people who use it regularly, involves just having whatever nonsense they didn't ask to see thrown at their faces based on some statistical model of impression metrics and whatever personal data can be crammed into it.

I just have zero interest whatsoever in seeing any stream of so-called "content" that's not curated by people I know, or at least whose taste I trust. If you're looking for art specifically and you're starting from absolute zero context I guess I can understand the value of a tool to help seed your follows list, but beyond that trusting a machine to do curatorial work—a task that intrinsically demands taste and discernment—seems to be perverse beyond measure.

I was young and naïve when I first heard about Netflix's vaunted recommendation algorithm, and I'll admit to starry-eyed imaginings of the films it could discover for me. But it became clear the moment I actually began to scroll sideways through the "movies you might like" that it hadn't the foggiest idea of my tastes. No matter how much data I poured down its hungry gullet, it never machine-learned even a loose approximation of what I enjoyed. A simple list of films I hadn't seen directed by the same people as films I had would have been dozens of times more useful to me.

Do people not realize this? Do people not have friends of good taste who share their curations online? Have we forgotten how to collate art direct from the artists themselves, that we must be beholden to the torrent of horseshit that is "for you page"?


bruno
@bruno

Honestly, recommendation algorithms nowadays are not really recommendation algorithms because 'recommendations' is kind of a trap idea.

When I (and presumably you too) was growing up, television was often a very un-curated experience, right? You'd just tune in and watch whatever was on. You had a bunch of channels to choose from, but within each channel you had no choice. If you wanted to truly curate and seek out specific television, you'd need to arrange your life around being around to watch a show – most people didn't do that except for maybe one or two must-see-TV type deals.

The Netflix carousels are functionally the same thing now – a selection of genres and a list of things that Netflix is trying to push in those genres. Programming is done by executives. The 'algorithm' is barely there, and your carousel probably looks a lot like someone else's; most of the recommendation is coming in the form of rearranging which genres are floated to the top, and maybe the imposition of some broad categories (Netflix can look very different depending on whether or not it thinks you watch k-drama, for example, but it will always push its big expensive prestige shows to the forefront).

Which is to say, 'recommendation' is now just the monoculture. Spotify infinite playlists are not particularly distinguishable from just listening to genre radio, etc.

I think that in the early days of this stuff there was just a total misapprehension of how people consume media. Most people have a relatively omnivorous media diet where they consume the most broad and accessible things in a variety of different genres. I think the idea of "oh, this algorithm will recommend hiddem gems to you" kind of runs into the reality that most people are not that interested in watching hidden gems because they are getting more than their fill of a given genre just from the regular gems. Is the median movie viewer going to seek out the more obscure horror films when they are already barely keeping up with the big A24 releases?


You must log in to comment.

in reply to @nex3's post:

people you know and trust are only gonna post so many times in a day. it doesn't take very long to read everything they posted, and then log off and go do something else with your day

the "for you" page is mediocre slop, but it'll keep refilling the slop bucket endlessly from an entire world's worth of posts. unless your interests are extremely niche, it'll functionally never run out of new stuff to show you. you can just keep scrolling and scrolling and scrolling, for hours and hours

and that's why most social media platforms heavily incentivize the algorithmic stuff, using every design trick and dark pattern they can think of to keep users diving back into that cesspool. it keeps you in their app, seeing their ads and/or contributing to their metrics

The only time I have ever heard of "for you" being good is that my wife somehow tricked twitter into making hers exclusively gay arknights fanart. That's it, that's the only time I've ever heard someone not actively complaining about them

One time I made a tweet complaining about the algorithm that specifically called out webcomics as something I never see anymore, and the app actually showed me people's webcomics for a day because people engaged with that tweet to give me advice, and then it stopped again because I was passively reading the comics instead of engaging with them. :/

I had much the same experience with Spotify's Discover Weekly playlist, which was supposedly based on what you liked / listened to. Every week, it would suggest 20 (iirc) songs, and every time, it would be a bunch of stuff I didn't like at all. Every so often there would be a song in there that I actually liked, which of course got me to keep listening to an hour of crap I didn't like every week until I finally quit Spotify.

Interesting, my experience has been the opposite: I've discovered lots of stuff I've liked from Discover Weekly, and for several of them have ended up buying one or more albums (on Bandcamp). It does tend to weight my recent listening higher a lot though: last week for a lark I listened to all of Taylor Swift's albums on Spotify and now this week's Discovery Weekly is filled with women singer-songwriters (which is fine by me).

I'll have to try out the suggestion of Bandcamp recommendations though -- I've never thought about using that.

in reply to @bruno's post:

it will take me more pondering before I figure out whether this difference in perspective is very very sad or just a legitimate but foreign way to engage in art

the homogenization of available media into "who can pay the biggest bucks" is a cultural atrocity regardless

Ah yes the "Well, I have nothing else to do tonight, is anything fun on UPN/MyNetworkTV/The WB/CW/Whatever... Huh, do I want to watch a King of Queens rerun, COPS, or The Hughleys? Maybe I"ll just go find something else to do".

I don't think the recommendations are Good, but I also am someone who just Doesn't Know Anything, so having a thing to scroll is helpful because I will eventually go "oh this looks interesting.". When I talk about streaming services to some friends and they just tell me to pirate things, that doesn't help me find something new, because from what I've seen, those sites just give you a list of titles with no description or key art or nothin' right off the top so you can scan it for interest.

It's like switching to the old scrolling Guide channels on some cable TV package and just watching the names of shows go by, hoping you see something you recognize.