How Instagram and Tumblr Work

I’m a big fan of Gal Shir’s texture brushes, and yesterday I read his self-published book View Insights, which is about how to grow a following on social media, specifically Instagram.

The first 2/3 of the book contains good general life advice, such as:

(1) Do what you actually enjoy doing
(2) It’s not necessary to quit your day job
(3) Divide your ideas into “big projects” and “small projects”
(4) Learn how to balance and prioritize your projects

The book also contains a few pieces of advice that are predicated on assumptions that strike me as somewhat “masculinist,” such as the idea that no one cares about pictures of your face or your personal life – which is not even remotely true in the online spaces I have experience with, where people tend to care just as much about the artist as they do the art. So your mileage may vary, I guess.

What I found interesting about this book was the last third, in which the author digs deep into how Instagram’s algorithms work and why they work in the ways they do. Tumblr is an altogether different platform that works in different ways for different reasons, but Shir corroborates some of the tendencies I’ve noticed on Tumblr, such as:

(1) The “value” of a post is algorithmically ranked within a limited number of tags
(2) This “value” is partially dependent on the “user rank” of the poster
(3) This “value” is also determined by interaction from other “high-rank” users
(4) The level of interaction needs to be significant, like commenting or sharing (and not simply liking)
(5) This “high-rank” interaction needs to happen within the first few hours of posting

A while ago I speculated (here) about what I called “anchor blogs” on Tumblr, which are blogs that may not necessarily post original content but still manage to be influential. I was thinking about how actual person-to-person social networks tend to function within fandom; but, if this algorithmically based “user rank” theory is true, this would help explain the patterns I noticed relating to how any given post spreads.

Tumblr has passed its prime, so I’m not sure if any of this still applies; but, according to this theory, this is what you would have needed to do in order to become a “high-rank” user:

(1) Interact with a lot of content
(2) At a significant level
(3) Within hours of it being posted
(4) And follow a lot of people
(5) While having “high-rank” followers

What all of this boils down to is that these two platforms reward “engagement,” which is essentially extroverted behavior combined with the condition of being on your phone all the time. Shir says that, when he first started trying to build a following on Instagram, he would devote three hours a day to interacting with other posts and people on the platform during peak hours. Unlike Instagram (and Facebook), I’m almost 100% certain that Tumblr doesn’t apply a secondary “positivity rating” to posts and comments, but actually being genuinely friendly probably doesn’t hurt.

Technologies of Behavior Modification

The Age of Surveillance Capitalism by Shoshana Zuboff review – we are the pawns
https://www.theguardian.com/books/2019/feb/02/age-of-surveillance-capitalism-shoshana-zuboff-review

While insisting that their technology is too complex to be legislated, there are companies that have poured billions into lobbying against oversight, and while building empires on publicly funded data and the details of our private lives they have repeatedly rejected established norms of societal responsibility and accountability. And what is crucially different about this new form of exploitation and exceptionalism is that beyond merely strip-mining our intimate inner lives, it seeks to shape, direct and control them. Their operations transpose the total control over production pioneered by industrial capitalism to every aspect of everyday life.

I’m not sure I’m up for reading the actual book, which sounds miserably depressing, but this is an interesting review. Two paragraphs are devoted to a blunt deconstruction of Pokémon Go, which is fair.

Even though most of the people (especially artists) I used to follow on Tumblr have moved to Twitter and Instagram, I still feel a bit weird about engaging with those two platforms. Despite its flaws, I appreciate that Tumblr is relatively chaotic and isn’t making money for anyone. Activity on the site has dropped off since the beginning of the year, and I’ll miss it when it’s gone. Also, as much as Discord annoys me for being exclusive, inaccessible, and difficult to use, I’ve found myself spending more time on art and sketch channels during the past few months.

Meanwhile, AO3 remains the Gold Standard of Internet and continues to be my happy place.

Algorithmic Time

The 2010s Broke Our Sense Of Time
https://www.buzzfeednews.com/article/katherinemiller/the-2010s-have-broken-our-sense-of-time

How did everything get so jumbled? Stories about our phones, Instagram, Twitter, Facebook, and the rest often concern Nazis, grifters, scammers, plagiarists, the aesthetes who reject that online life, the famous, the infamous, people who are making a buck, and anyone else who pushes the logic and limits already in place. But what about the rest of us?

The 2000s were a bad decade, full of terrorism, financial ruin, and war. The 2010s were different, somehow more disorienting, full of molten anxiety, racism, and moral horror shows. Maybe this is a reason for the disorientation: Life had run on a certain rhythm of time and logic, and then at a hundred different entry points, that rhythm and that logic shifted a little, sped up, slowed down, or disappeared, until you could barely remember what time it was.

The writer isn’t wrong, but holy hell do all of the flashing GIF images make this article difficult to read. I understand that this is (probably?) the result of an intentional artistic decision to create a format that mimics the experience of having your attention constantly divided between multiple competing demands online, but it works a little too well. The essay is about how having our lives mediated through social media disrupts our memory; and, lo and behold, I can barely remember what I read.

All that being said, I’m planning to cut and paste the text into a document to study later, as what the author is describing mirrors my experience of the past four years almost perfectly.

Vetting and Sharing on Social Media

I used to think that, the more followers a blog has, the more popular its posts will be. It only stands to reason, right? I also had this idea that artists have a lot of influence on Tumblr partially because of how the platform privileges images but mainly because of their relatively high follower counts.

I’ve since figured out that what’s actually going on is that a post needs to be “vetted” in order to spread. In other words, a post needs to be reblogged by someone whose taste other people trust. Or, well, “taste” is a strong word, as is “trust.” What I mean is that people are far more likely to reblog a post if someone they’re following reblogs it, even if they’ve already seen it posted on the original blog. If that “someone else” is associated with the same fandom as the post, then it will spread farther. In this case, “fandom” can be very broad; like, say, the “intellectual shitpost” fandom.

At this point I have far more followers than my small blog on Tumblr deserves, but it’s not my follower count alone that enables any given one of my posts to spread. By itself, one of my fandom-related posts might get forty to ninety notes, and it’s only when someone associated with the fandom reblogs it that it will get more than a hundred.

I’ve seen this happen on posts I’ve reblogged as well. Sometimes I’ll reblog something from a few months (or even years) ago, and it will go from having about twenty to thirty notes to having several hundred almost overnight.

Once a post reaches a certain level of critical mass, the number of notes alone will indicate that it’s already been vetted, and it will also be picked up by the site’s promotional algorithms. Before it can go viral, however, a post first needs to have community support.

I feel like the same applies to Twitter – albeit to a lesser extent, as Twitter’s septic open wound of an algorithm aggressively prioritizes a handful of tweets while hiding most of the rest, even if you turn off the “best tweets first” feature. As far as I can tell, Twitter doesn’t have the same “recommended for you” algorithm that Tumblr has, in which the posts liked by your mutuals – and the posts posted by people followed by your mutuals – will sometimes appear at the top of your feed. Rather, Twitter has figured out what types of tweets are most likely to provoke a reaction (generally negative) from you and show those tweets to you over and over until you either like them, hide them, or blacklist whatever keyword or hashtag they’re using.

Regardless, I’ve noticed that there’s still something of an influencer culture on Twitter, whereby people are more likely to respond to or retweet something if it’s already been vetted by someone they trust, even if they already follow the OP.

Meanwhile, Instagram is testing a feature that will hide the number of likes a post has received specifically for the purpose of protecting the mental health of their users, and I for one could not be more relieved.