Wednesday, November 02, 2022

Futures of Twitter: Scenario Testing

Yes, I've railed against the appetite for future forecasting in this blog before. In fact, pretty much every post from the past few years is, in its own way, a screed against the kind of prediction of complex systems of long periods of time that news readers and journalists who interview academics can't seem to get enough of. And yet, when something "big" happens - a pandemic, the election of Donald Trump, Elon Musk taking over Twitter - I feel the same insatiable pull toward forecasting that everyone else feels. These are important events with far-reaching implications, and so despite their complexity, we can't seem to just say, "who knows what will happen!" We have to venture a guess.

And not just any guess. When we see worst-case-scenarios as plausible, we detest a future in which we ignored the threat so much that we have to predict doom. The unpleasantness of comprehending those futures is outweighed by our fear of feeling guilty about (and of being guilty of) not having done enough to prevent them. The other side of the argument is that there has never been a time in human history when worst-case-scenarios weren't plausible, and to spend time constantly ringing alarms is both joyless and unproductive, and that our inability or unwillingness to recall the false alarms of yesterday keeps us from learning anything about how complex systems actually work, to actually improve our prediction accuracy, and - by extension - to improve our world. I do get both sides of that argument, and so to me it seems unresolvable. 

So, to further refine my stance against long-term predictions about complex systems (the global economy, geopolitics, culture, developments in technology, etc.), I'll say that I prefer something less like predicting whether some event will spell doom for society (super vague, no timeline for when "doom" will occur, what shape it will take, etc., and so unfalsifiable) and more like fivethirtyeight's scenario testing. You start with several plausible scenarios - each one mutually exclusive to the degree that it's possible, and fairly specific about outcomes and timelines but not so specific as to be inevitably falsified by some minor quirk of history. This, at least, would allow us to learn something if our predictions are wrong, allow us to do a post-mortem and see what we missed, what variables we weighed too heavily, etc. Of course, this is dissatisfying if you are a doom-predictor because all the learning in the world won't amount to a hill of beans after the apocalypse that you failed to predict! But assuming there's at least a few people left on planet Earth, learning will be a good thing. To my mind, entertaining somewhat optimistic scenarios isn't mutually exclusive with being prepared for (though not actually predicting) the worst. 

In that spirit, here are a few scenarios regarding Musk's takeover of Twitter, some more optimistic than others. They use three metrics of success/failure: financial success (does the company turn a profit, stay solvent, continue to exist), cultural relevance (does the general public care about what's said on Twitter as much as they do now, more than they do now, or less than they do now? Does some other platform become the proxy for vox populi that Twitter is right now?), and "externalities" (affective polarization, political violence, political revolutions, cancelling/calling out people. Externalities could be both "good" [revolts against dictatorships, publicly shaming those who deserve it] or "bad" [downfalls of democracies, publicly shaming those who don't deserve it]).

Scenario 1: Financial success with lower cultural relevance and few "externalities." The most intriguing part of Musk's takeover to me, so far, is his apparent willingness to blow up the existing way that Twitter makes (or attempts to make) money. Perhaps the changes he instates drive user count WAY down, but this won't necessarily make Twitter less of a financial success. He could charge a portion of users $8, or pay the 100 most followed $1,000,000 and charge anyone who tweets more than 100 times a year $50 (a kind of 'lottery-ticket' model that, when you think about it, isn't that different than would-be filmmakers shelling out for expensive equipment and film school for a crack at a studio contract). My personal favorite idea is Willy Staley's "congestion pricing" plan - when a certain number of users have tweeted about a topic, the price of tweeting about it increases, which would - in theory - encourage people to tweet less about trending topics and more about a diverse array of topics. Or maybe a revived Vine drives profits and allows Twitter to continue to be a loss leader. Regardless of how he does it, it's possible that Twitter becomes profitable.

Meanwhile, the platform loses some of its cultural cache. The behavior (on Twitter and off) of Musk and a handful of high-profile Twitter users who would have been regulated under the old regime turns off journalists and tastemakers to such a degree that they stop paying attention to it (regardless of how outlandish it is) and pay more attention to how YouTubers, TikTokkers, and podcasters react to the day's events. Twitter will always have the brevity and "portability" of text on its side - it's easier, in the course of conversation, to tell someone what another person Tweeted than to try to describe even a short video. So, maybe a Twitter alternative (short text, highlighting trending topics) pops up and steals at least part of the cultural spotlight from Twitter. 

Regarding "externalities," there's no convincing evidence that what people say on Twitter causes political violence, revolutions, voter intimidation, cancelling, etc. These things happened before and they continue to happen, but it's just as easy to point to what people are saying on TikTok, YouTube, and podcasts as it is to point to what people are saying on Twitter as the cause of it, and so the argument that a Musk-led Twitter is leading to the downfall of society never really gains traction. 

Scenario 2: Financial success with the same level of cultural relevance and no apparent increase in externalities. Same as Scenario 1 except people (journalists, podcasters, YouTubers, TikTokkers, and - by extension - people who pay attention to what those people have to say) keep paying attention to Twitter. Maybe what's said on Twitter doesn't change all that much, or maybe it turns into a different kind of dumpster fire that people can't ignore. This is basically a kind of 'status quo' scenario. Sure, somethings change (the way Twitter monetizes users, the number of users, the character of the userbase), but no seismic changes to culture or to Twitter as a corporate entity. To me, this still feels like the most likely scenario. Habit and inertia are powerful forces - people will want to keep using the platform they've gotten into the habit of using, and reporters will want to keep referring to Twitter as vox populi. The bulk of everyday users who don't engage with trending topics will muddle along - perhaps as long as they have no interest in reaching large audiences with their tweets, they'll be allowed to use the platform for free. 

Scenario 3: Financial success with the same level of cultural relevance and "bad" externalities. In this scenario, there are a few instances of political violence that can be clearly and obviously linked to what people say on Twitter. Musk will point out that people are saying the exact same things on other platforms like YouTube and TikTok and that his company is being judged by a double-standard, but the sizable portion that already has it in for him won't want to hear it. On a "normal" ad-supported social media platform, this would start a death spiral - advertisers pull out, revenue craters, and they can't keep the lights on. But it's interesting to consider what would happen if Musk pivoted away from an ad-supported model, which he already seems to be doing. How bad and/or obvious would the externalities have to get before governments blocked the platform? I suppose middle-layer companies (occupying the space between users and platforms) like Apple, Alphabet, and AT&T might say 'enough is enough' and drop Twitter, but then maybe Musk uses Starlink to do an end run around them! This sounds a bit far-fetched to me, but you never know. 

Scenario 4: Financial failure. Enough high-profile Twitter users leave that the platform becomes less attractive to many everyday users, and the ad-supported models becomes unviable. Musk's attempts to charge subscription fees results in users trying to circumvent the fees, leading to draconion crack-downs, further turning off everyday users who head over to YouTube, TikTok, podcasts, Reddit, and Twitter clones. It's odd to think of a platform that still has cultural cache going belly up, but to some extent this is what happened to Vine. If you can't make the numbers work, your impact on culture is a moot point - you just cease to exist...or you get acquired, sold off, merged, etc. What might Twitter's afterlife look like if that happens? 

This, I think, gets us to focus on particular observables - financials, number of users, the extent to which people are talking about/paying attention to what's said on Twitter, political disruption - which is all I'm really hoping for in conversations about the future. 


Monday, April 25, 2022

How platform-specific is influence?

In the past, I've railed against making predictions about the future as a way to understand the present. This is different than predictions used as a way of dealing with uncertainty through humor - a pretty common, totally understandable reaction to crises. I'm thinking more of op-ed hot takes that seem to want to be taken seriously.

Predictions like this are often so vague that they're impossible to prove wrong. For example, if I predict that Twitter/Facebook/Reddit will die/fail, how and when will I know if this turns out to be true? If their userbases, sometime over the next 5 years, decline by 35%, have they died/failed? How about 50%? Do we have to wait another 5 years to judge the prophecy? There's also the problem of lacking a counterfactual. If something horrible happens after someone uses social media, we can't compare it to a world in which they did not use social media. It's possible the horrible thing would have happened anyway (or perhaps something even...horribler?), just through some other mechanism. There is no disincentive for continuing to make bold, wrong predictions because there is no mechanism for keeping score, no highly visible public record of wrong takes (though spectacularly wrong takes do tend to catch flak).

And yet I'm finding it hard not to use the occasion of Elon Musk's (apparently successful?) attempt to buy Twitter as a time to indulge in some predictions. I'll compensate for this indulgence by offering something more in the spirit of this blog - using the occasion as a way to gain insight into something more general about the uses and impact of social media. But first...

I can think of two likely scenarios:

Scenario 1: very little changes. Twitter use habits, like any kind of habit, are hard to change. There have been many instances in which social media platforms have done something that users have not liked, but most users do not then abandon the platform. We can more easily recall the instances in which they did abandon the platform: MySpace and Digg come to mind for me. But it seems that nearly every significant change to a platform - be it functionality, moderation policy, or ownership - tends not to result in a wholesale change in user behavior. The "culture" of Twitter may change, but linking this to Musk's ownership can be a tricky claim to verify, as cultures are always changing. In this scenario, there is no massive change in Twitter discourse (i.e., who participates and what is said or amplified) that can be directly and obviously linked to Musk's ownership of the platform. A handful of influential Twitter users on the Left will defect, as will a few hundred thousand less-influential users (including many journalists who will write about the experience). They will be displaced by a few newly influential users on the Right and a few hundred thousand less-influential tech bros and Far Right/Alt Right users. Musk might put the worst of the Far Right users in seemingly random "time outs," less out of any political conviction and more to reassert dominance get a laugh at their righteous indignation at having their trust betrayed. Losing the revenue they generate through ad exposure or subscriptions won't hurt, but the board will chastise him for playing fast and loose and he'll promise to be better. Pundits will pin everything the Right does on Musk and Twitter, but the connection won't be obvious and the claims won't be of interest to anyone outside of a bubble of journalists, news junkies, and academics. I consider this the more likely of the two scenarios. 

Scenario 2: things go poorly, less because of how Musk changes the functioning of the site and more because of the signal it sends to current and potential users. Far Right fringe groups and apolitical trolls will see Musk's unwillingness to regulate content as an invitation to push things as far as they can, inciting small-but-significant numbers of Far Right "activists" to engage in terrorism and assassination. A handful of violent political acts will be linked to Twitter in a way that will seem obvious and undeniable, not just to readers of New York Times opinion pieces but to the average politically-disengaged citizen, who will come to associate Musk and all his brands (e.g., Tesla, SpaceX) with toxicity rather than innovation. Musk will then belatedly recognize the damage to his brands and sell the company. 

So as to make these scenarios more falsifiable, I'll say that one will happen by the end of 2023. 

With that out of the way, I'll return to the title of the blog post, returning to a topic that I'd like to write about at greater length in the future: what is the nature of influence on social media platforms, and how does it differ (in duration, scope, domain-specificity, etc.) from mass media or face-to-face influence?

In thinking about the possibility of influential Twitter users defecting to another platform, I realized that platform-abandonment might involve losing one's audience and the attendant influence. All fine and well to say "follow me to YouTube/TikTok/Substack/Tumblr/Medium/Mastadon/whatever!", but it's hard to coordinate a mass exodus to another platform, and harder still to sustain that interest over time, especially when the new content home offers more of an a la carte content experience rather than the smorgasbord we've come to expect from popular platforms. Smaller platforms have trouble handling large influxes of digital refugees. Perhaps especially high-profile influencers will easily move their following to another platform (or to podcasting), but what of the Twitter users with followings of 500 to 50,000? How portable is their influence? 

Those who jump ship may lose some of their audience but find a more intimate relationship between themselves and a smaller group of followers on another platform, and this may be more rewarding for them. Those who stick around a less-regulated Twitter may find themselves in more direct battles with other users with whom they disagree than had previously been the case. The 'block' function and tweaks to the much-discussed-but-poorly-understood algorithms will help determine how easy it is for users to tolerate dissent in the name of maintaining influence. But no matter how it shakes out, I think we can learn something about influence by seeing what happens to the public personae of Twitter users with decent-sized followings who leave Twitter. Maybe Twitter's loss will be YouTube's/TikTok's/Substack's/Tumblr's/Reddit's gain, both in terms of users and in terms of cultural influence, however one chooses to define and measure that slippery concept.