i think one of my most cancellable takes is that complaints about "the algorithm" are just complaints about human behaviour. all "the algorithm" does is amplify the way each and every one of us interacts with the internet
it DOES fucking suck when things you put a ton of effort into don't get attention. it's ANNOYING when throwaway stuff takes off. but
cohost has no “algorithm”, yet the same thing happens here
this write-up of the cohost “vuln” took a while, and i really wanted it to be seen. but it got nowhere near as much attention as this silly rip of QR's passenger information displays — which, sure, took some time, but significantly less than the write-up.
the puppy dog PID is quick to consume, quick to share, and, big plus! sharing it signposts a piece of your identity. i'm a fucking dog and i like trains. share that and i'm immediately telling you something about myself, and i'm also getting a bit of a happy moment whenever someone likes my share of it, since someone's either acknowledging how i'm positioning myself or themselves relating to the post. making it didn't take that much time, and i didn't really expect it to go that far since “the screens at train stations in brisbane” is a pretty local phenomena.
the write-up takes reading and understanding. not actually that much, compared to some other long-form posts, but more than zero. i wanted that one to go far. if people read it and talk about it, i feel like i've done something of value and people are not only enjoying my work but learning from my experience and time. it's got some vaguely interesting technical information, and for anyone who themselves looks for these types of issues a new avenue to explore.
which sounds more likely to take off?
you can make more throwaway things than you can high-effort things
from everyone's favourite tom scott:
... and yes, it absolutely sucks that there are going to be genius-level pieces of work that aren't making it outside of [your network of friends] -- and mediocre work that gets lots of praise because it's from someone well-established. 1
I keep following all those rules I've learned: I make as many things as possible ... every idea you put out there is another roll of the dice. 2
how many silly one-liners can you pump out in a day, if you tried? i mean. don't try, it won't be healthy for you. but how many weeks-long project ideas do you come up with in a day versus silly jokes you can write and post in a couple minutes? if the odds weren't already stacked towards short-form content, and even if the chance of your throwaway work taking off is the same as your high-effort work taking off, which gives you more rolls of the dice?
time can measure effort, but effort often does not predict success. sorry.
people “want” things they can quickly interact with, then move on
it's a single line joke. it's a visual gag. it's not a pages long blog post or a tool they have to download or a place they have to go. tiktok is popular because it's short-form, the barrier to entry is near zero, and “if [the video] sucks, [you can] hit da bricks”. scroll away. thank you, next.
people also want things where sharing or interacting with it signposts, indicates, shows off some part of their identity3. playing a game, reading a blog post... doesn't usually do those things. not to imply that these don't provide value to people — by fucking god, do they provide value — but not the kind of value that social media rewards, by virtue of its role in people's lives. i might have the time of my life playing with your thing, but what makes work take off is word-of-mouth (be it a real mouth or your whisper or megaphone on social media).
i already talked about the role that the silly QR PID plays. maybe i'm projecting, but sharing it lets me tell people about a part of myself i want known. i wanted to turn the effort i'd put into replicating the visual style into a project where you'd get a semi-accurate, visually similar PID with real-time data — if i shared that, it'd get a couple likes. people on cohost aren't from brisbane, they don't want to clone a git repo, npm install, figure out bugs, yada yada yada. what they do wanna see is a cute little post with some puppy dog words and bright colors.
is that bad? is that good? i don't know. that's people. should we, as “a species” be working to rid ourselves of this bias towards quick consumption? probably? i don't know. but if your goal is having your work seen — and not rewiring the brains of the entire human species — i think the take away is clear.
a lot of what we ascribe to “algorithms” is just people
don't get me wrong. there are a looooot of ways that more sinister forms of recommendation schemes go well beyond just reinforcing and amplifying human behaviour. but i think the issues we see brought to the table a lot of the time are just the human behaviour-amplifying ones. considering the exact same things can happen here, happened before social media, happen in real-life social circles, it's hard to see how the computer is doing anything more than This But At Scale™️.
if your recommendation scheme pushes a product you have a commercial relationship with over organic user-created stuff, that sucks. but if people share the PepsiSMTM ad more than your post, because PepsiCoSMUT have more followers, and the literal virus-like exponential growth delivers them more [insert a metric here], was that really the recommendation scheme? it's not like virality didn't exist before “algorithms”, it was just different.
so
a lot of the time i feel like distaste towards the “algorithm” and its effects on how your work is seen by others has nothing to do with the algorithms and much more to do with people's behaviour. people's behaviour, and how it feels to be on the receiving end of the whims and desires of randoms.
to be clear: i've hopped onto twitter a few times since the downfall, and being forced to use the non-chronological timeline has made me leave quite quickly: but not before scrolling through for a while. i hate seeing random stuff. i want to see what my friends are doing. but even when i'm scrolling through a chronological timeline, i'm mostly consuming the short-form content and scrolling past long-form stuff. i get sort of meta-angry when i realise that i'm just looking at “other customers also bought these tweets” because i step back and notice that it's often shallow content made to appeal to all the things i've mentioned.
the irony, of course, is that this post has taken a decent amount of time and effort to write. an hour! but while getting ready to post it, i realised how little appeal the block of text has. so now there's images, some css that's even “reactive”; just so that it's harder to skip over in your timeline. time is effort is not appeal is not success. we'll see how it goes.
-
again, from tom, but i don't wanna go find the quote. sorry.
I've been thinking about this post since it keeps popping up and the problem is manifold:
1) There is no The Algorithm
Every different site you use that does recommendations or does anything other than presenting things in chronological order is using some algorithm - that is to say, some set of rules, some programmatic process - to decide what to surface to you, but there is no One True Algorithm. It's different per site, it's often different on different days on the same site because they're constantly A/B testing different mechanisms to see what gets the best results in terms of traffic mix-shift or whatever other metrics they're looking at (ad impressions usually being a real big one). It's always changing and it's not changing in ways that are primarily intended to be beneficial to you, the user. That being said:
2) Algorithmic timelines, recommendations and For You Pages are intended to seem beneficial
But "beneficial" in very particular ways! They're intended to keep you engaged with the site content. That means they show you things that keep you scrolling, keep you clicking, keep your dwell time up and your ad impressions up. Sometimes that means showing you things you like; sometimes it means showing you things that make you mad. Sometimes it means showing you things that are confusing! But the point is singularly to keep you engaged. That does not mean they are intended to do anything in particular for you:
3) Just because you post, does not mean that site algorithms are intended to make your posts Get Out There
They're just not; your posts are just more content for the site algorithm to use per its ruleset. It does not mean they are intended to put your posts in front of people who want to see them, and definitely does not mean they are intended to put your posts in front of the people you want to see them! Indeed, it is often the case and has long been so that one of the most efficient ways to improve all of the metrics listed above is to generate outrage, which is why trans creators on TikTok, just for instance, have such a tendency to end up in the FYPs of fascists. Harassment is engagement!
4) Recommendation algorithms can be friend or foe but have to be manipulated
It's advertising and SEO. For creators who are good at that and comfortable with it, it works well. For people who aren't, it's pretty unfriendly. I've talked before about the split between people for whom social media is primarily a prospective audience and people who are just here to shoot the shit. You can't expect people who aren't interested in helping Build Your Brand to, well, do that. I've worked on ad campaigns and such before, if you want that you can pay me.
It is definitely not The Algorithm's fault that you're having trouble building an audience, but it's not the lack of one either. You have to connect with people in a meaningful way, or you have to be really good at sales. One or the other.
