Anatomy of a viral post

What does it feel like to have something widely shared online?

Over the past few years, I’ve had quite a lot of Twitter/X posts end up widely shared. But prior to this, before I’d found myself in the middle of this kind of virality, I hadn’t fully appreciated what the journey is like. I recently had a post on pandemic preparedness pick up a few hundred retweets, so thought this could be a useful opportunity to dissect the stages of virality for others.

What actually happens when you have a post pick up loads of shares? Where does it start, and where does it end?

The post in question…

Super-superspreading

Like real-life viruses, online content rarely spreads like clockwork, with a few people passing it to a few others, then on to a few others and so on. While biological outbreaks often involve superspreading, with some people infecting dozens and many infecting none, the growth of popular posts is typically dominated by major amplification events, with thousands or millions seeing content from a single account. In other words, it’s down to ‘super-superspreading’ (I’ve previously delved more into these dynamics in this piece).

After I first shared the above Twitter post, it spread slowly at first, picking up some initial likes and retweets. But things really got going when some other widely followed accounts shared it, bringing the content to thousands of others.

Comments and arguments

Along with likes and retweets, people will usually comment on a popular post. Then others will start commenting on the comments. And often arguing about them too.

Unlike some of my previous posts, where some comments have gone remarkably far off track (e.g. random people getting into lengthy arguments about climate change on a post originally about immunology), the arguments on this new one trod familiar ground. On this post, as I’ve noticed for many others I’ve shared relating to epidemics, the longest heated discussions are often dominated by extreme views of disease control, in which a tool either solves everything or has no effect whatsoever. (The reality, of course, is generally more complex.)

After a few hours away from Twitter, it is always a bit surreal to skim through the mentions section, trying to guess which random angry comment is related to which other previous random angry comment.

Insulty quote tweets

It’s pretty much a rule of the internet that anything that becomes sufficiently popular will be disliked by some, because lots of impressions multiplied by even a small probability of annoying people will result in several annoyed users. Eventually annoyance turns into abuse. Usually in the form of unimaginative, mediocre abuse, from people hiding behind anonymous accounts. And often it will take the form of a quote tweet, presumably in an effort to show their followers how quickly they can throw out insults, or to encourage others to pile on doing the same:

Not exactly the pinnacle of wit.

More generally, I thought this previous thread by Zeynep Tufekci nicely summed up the problems with quote tweeting as a form of communication. As she put it:

Quote-tweet is just fine if used honestly and with care, but functionally speaking it breaks two things: threads and the conversation. One quote in a thread is easy to mischaracterize. Plus dueling quote-tweets are practically impossible to follow. Engagement over substance.

Reality leaves the building

After the arguments, and the insults, comes the nonsense. In the case of posts about epidemics and pandemics, that usually means people claiming SARS-CoV-2 doesn’t exist, or that everyone involved in the global COVID response should be tried for crimes against humanity (although it’s never immediately clear to me which aspect these proponents are specifically unhappy about). Then, like a bland boilerplate script, others will chime in to say that, actually, these trials should come with the death penalty. Or, perhaps, just some negligent imprisonment:

From experience, a lot of these comments will eventually disappear, perhaps deleted out of quiet embarrassment or because their owner later gets banned. So these messages end up like a transient bad dream. Not grounded in reality in the short term, or visibly real in the long run.

The (lack of) conversion

Then, after a day or so, it all gets much quieter. The retweets and the likes fade, and the internet moves on.

So what effect does that flurry of shares and views ultimately have? As well as (I hope) getting people to think more about an issue, or sparking some further discussion, I’ve sometimes had media outlets get in touch after seeing a post - or even quote it directly. (I’ve written more about the benefits of Twitter during COVID in this piece.)

Then there’s the question of direct conversion from one platform to another. My recent widely shared post gained over 180,000 impressions. It also included a link to this post of mine on (lack of) pandemic preparedness. So how much interest do you think it drove to this Substack article? How many new subscribers did it lead to?

Go on, pick a number.

Got one in mind?

Well, it turns out that in midst of all that Twitter post popularity, I gained six new Substack subscribers¹, with zero new likes or comments on the linked article.

Now, I’m not here to chase traction at any cost (which you may have noticed, given how much I post about data and maths rather than, say, cat videos). But I do think it’s interesting how little one platform seems to drive traffic to another. And I think it’s useful, if I’ve ended up running an accidental experiment, to report back what I’ve found. Perhaps, in some small way, it will help others navigate the strangely both over- and underwhelming experience of having online content go viral.

1. If you’re one of them, welcome! Thanks for taking an interest in my writing.