(This is a follow-up to Against Trendism)
Basing visibility on popularity is a uniquely awful version of ‘tyrrany of the majority’ because uncommon views become invisible, even if, were they to start on an even playing field, they would become popular.
In this way, it encourages mental stasis: since ranking is based on an immediate appraisal of how popular something already is, and visibility is based therefore on past shallow popularity, there’s no room for rumination.
This is NOT an attribute of ‘technology’ or ‘social media’, but an attribute of visibility systems based on immediate ranking. Visibility systems based on ranking delayed by, say, three days, or with the top 25% most popular posts elided, would be fine.
Our capacity to imagine new possibilities is based largely on our familiarity with the bounds of possibility space — we can only imagine views that are in the neighborhood of views we’ve heard expressed in the past. So, making the already-unpopular invisible limits imagination.
(There are hacks we can use to make it possible to imagine views nobody has ever held. We can make random juxtapositions, impose meaning on them, and then figure out a justification for them — like tarot reading. Or, we can merely iterate from some basic idea, getting more and more extreme, while internalizing the perspective of each iteration as something someone could possibly believe in good faith. The former — the bibliomancy approach — is common in experimental art, while the latter is typical of dystopian science fiction.
But, these hacks are pretty limited. We need a starting place. If we’ve only heard mainstream ideas, we’re going to have a hard time going off the beaten path with the dystopia approach, while we will struggle with the bibliomancy approach because most ideas can only be made to seem reasonable with the help of other ideas. Getting into uncharted territories with either of these approaches is difficult unless you’ve already filled out the middle of your possibility space with other ideas, because in their absence you would need to independently reinvent them.)
This is not a justification, in of itself, for banning metrics entirely. After all, this kind of exponential distribution happens with ideas even without the use of popularity signifiers: ideas spread, and popular ideas have more opportunities to spread. Trendism merely accelerates the process and widens the gap between the most popular ideas and everything else.
Sites like reddit use segmentation to prevent total ordering of popularity from dominating, although this ultimately means that popular subreddits have a disproportionate impact on this total ordering when it is seen. Similarly, we have seen piecemeal attempts to limit the effects of trendism for particular topics — the curation of trending topics at twitter and facebook, for instance, or ad-hoc ranking demerits for particular tags on lobste.rs.
However, we could be applying the measurements we already take to counteract trendism rather than accelerating it: making popularity count less the higher it gets, removing overly-popular content entirely, boosting the visibility of mostly-unseen content, using information about organic reach in sites like twitter to boost the synthetic reach of people who don’t have many followers (instead of boosting the synthetic reach of the rich), systematically demoting posts that comment on trending topics, spotlighting spotify tracks and youtube videos with zero views, and so on.
Where trendism devalues the function of recommendation systems as novelty aggregators, these tools could be modified to be anti-trendist, pro-novelty, and promote a cosmopolitanism that broadens our horizons in ways traditional word-of-mouth never could. This is a unique capacity of recommendation systems over curators: recommendation systems can recommend things nobody has ever seen, and can recommend them on the grounds that nobody has seen them.