Against trendism: how to defang the social media disinformation complex
There’s an essential mistake that almost every social media platform makes — one inherited from marketing (where it makes some sense), and one that is mostly unexamined and unaccounted-for even in otherwise fairly socially-conscious projects like Mastodon and Diaspora. In almost every one of these systems, incentives exist that confuse popularity with value.
I call this ‘trendism’ — the belief that an already-trending topic deserves to be promoted.
In marketing, because the piece of information being spread is intended to sell a product, the spread of that information is, in fact, theoretically proportional to its value. In social media, the information being spread is not a piece of advertising, and while most of these systems have revenue models based on advertising, that advertising is generated on the fly based on the viewer’s browsing history and has nothing to do with the content of the piece of information being spread.
The thing is, ideas travel in packs. When we encounter one idea, we tend to see its nearest neighbours also. When we find out something new, our friends hear about it too. So, trending posts are rarely surprising: by the very nature of their popularity, they are already familiar in their essence to most of the people who are directed toward them.
The information content of a message, in Claude Shannon’s formulation, is proportional to its deviation from expectation — information is surprise. Kolgorov’s formulation is similar: information content proportional to the smallest possible message that could say the same thing (which, of course, includes references to earlier messages or prior knowledge as a possible tactic).
In other words, from an information-theoretic perspective, a post that only tells you things you already know is worthless. Yet, trending content is almost always composed solely of things the viewer has already seen.
There’s one piece of information that a copy of a viral post actually has — the association between the content of the post and the person posting it. We share posts we’ve already seen as a way of expressing our identity, both personally and within a group. That is the only form of information valued by trending-oriented systems: tribal affiliation.
If we want to force our social media platforms into information-rich environments and lower the amount of tribal rivalry we are exposed to, there are a couple general-purpose solutions, and they all come down to kneecapping the machinery of trendism.
- Rather than block political content (only one kind of tribalist content, and one that is at least theoretically grounded in genuine philosophical differences about the ideal shape of the world, rather than geography or social groups), we should block all shared content. Remove retweets and shares from your feed entirely. Most of them are things you have already seen, and most of the rest don’t contain meaningful or useful information.
- Emotionally-manipulative posts get the most engagement, and are therefore ranked higher in feeds. (I don’t want to be emotionally manipulated. Do you?)* To defeat this ranking, force your feed to reverse-chronological order. To filter out emotionally-manipulative posts, filter out anything with more than a set number of interactions.
- Avoid being part of the problem. Before sharing, determine: is the information true? Is it new? Is it playing mostly on my emotions? If possible, delay your sharing for a long period of time — read an article, and then wait a few hours, or even a few days, before deciding whether or not it is of sufficient quality to actually re-post.
- Identify when you are being drawn into heated arguments, and ignore them. In the heat of the moment, you’re not actually making good points anyhow, and you’re more likely to misunderstand or misrepresent your opponent. The suggestions from #3 apply here too for comments — make sure your comments are accurate, informative, and cool, even if that means waiting several days to respond. Never let the system rush you.
- Visible metrics gamify trendism. Remove them.
Most social media platforms don’t make it easy to follow this advice. Mastodon is closest — it hides metrics from the timeline by default, supports only reverse-chronological post ordering, and allows you to filter all boosts from your timeline. For everything else, you will need to use browser extensions.
Facebook Demetricator and Twitter Demetricator hide metrics like number of notifications, retweets, and likes. F. B. Purity has a variety of options for hiding post types, and allows you to permanently force your timeline order to reverse-chronological. New XKit does the same for Tumblr. I wrote a greasemonkey script to grey out Tumblr posts with more than 1000 notes.
If you’re looking to simply spend less time on social media, there are several configurable blocking extensions, as well as Intent, which merely records and displays how long you’ve been using particular sites.
*Emotion is important, but good communication comes after emotional reactions are understood and contextualized.