Once upon a time, so the story goes, power over content was wielded by a narrow elite. You might write a great work, but never reach an audience, through the short sightedness of publishers.
Thankfully, those days are over. The internet has changed everything. Now, success is determined by the masses, through their clicks, likes, shares and retweets. An unknown author can publish a book or an article and the whole world can find out. The success of their work determined by its quality and interest, judged by readers themselves; independent of the whims of editors and the vested interests of media magnates.
Nonsense.
True, things have undoubtedly changed. Publishers have far less influence than they used to and it is possible for an unknown author to break through via the internet. However, the internet is not an ideal meritocratic system. It is much more arbitrary than that. Quality is important, but by no means necessary or sufficient.
How does an article reach readers?
Simply publishing an article online is unlikely to generate any views on its own — a single article published on an unknown blog will be found only by chance.
To reach an audience, an article must be published via a service that will push it to readers; or it must be promoted via social media. The success of an article is then dependent upon the interplay of a number of factors, for example:
- The size of the initial audience that is first exposed to the headline.
- How well the headline and picture entice that audience to click through and become readers.
- Whether readers interact with the post by liking it and adding comments.
- The proportion of readers who choose to share the content.
- Whether it is passed on by an influencer (someone with a lot of followers).
- Whether it is picked up by an aggregator like Flipboard, or Zite.
The feedback loop
The interplay of these factors creates a powerful positive feedback loop, strongly favouring content from already-successful authors and promoting articles that are already popular. Small differences in performance and initial conditions are magnified, resulting in some articles reaching hundreds or thousands of times more readers than others, with relatively little relationship to the actual quality and value of the content itself.
Viral outbursts of cat pictures and videos of people falling over are the internet equivalent of the whines and squeaks that you hear when a microphone is held too close to a speaker. Pure feedback. Just amplified white noise.
Why did that particular picture get shared 10 million times? Because it was fairly cute; because the title was intriguing (but actually made no sense); because it was published at 8:17am EST on a rainy Tuesday; because Judith sent it to her entire contact list; because one of those people shared it too, and has a large following on Twitter. In other words: just because.
If it had been published a little later, or a little earlier; if the title had been different; if any one of a hundred factors had been off, it would have reached only a few hundred people.
Although an author has control over some of these factors, and can attempt to optimise the content they produce, many factors are out of their control, or entirely random. And crucially, the majority of those factors have little or nothing to do with the quality, significance, insight or value of the content.
The heart of the matter
The underlying problem is that quality and value cannot be directly measured by the algorithms that are responsible for picking, ranking and pushing content. Instead, algorithms must use secondary indicators such as the current number of views, likes, and shares; along with the author’s prior history and existing following.
The algorithms can analyse millions of data points in a second and determine what is ‘trending’ according to these indirect cues, but they can’t determine what is important, relevant, ground-breaking, or insightful. In Umair Haque’s terminology, they are ‘stupid-smart’.
It is the indirect nature of these cues, their narrowness, and the fact that so many systems are looking at the same cues (and looking at each other), that creates the feedback loops necessary for ‘viral’ outbreaks.
With all our rhetoric of big data, analytics, machine learning and artificial intelligence, it seems that the best we have been able to do is to create a self-referential popularity machine. A poor substitute for our high words.
The dominance of the initial audience
Authors can accept the reality of this feedback loop and attempt to gain the best audience possible by optimising the factors over which they have control: They can write a great headline; include an interesting picture; and write great content that people like and share. Given the powerful amplification of the feedback loop, getting these pieces right could multiply its audience by perhaps 10 or even 20 times.
However, with an initial audience in the region of 10–20 people (which might be feasible by publishing on a push channel), this means a maximum audience of perhaps 400. This means that, unless an article happens to go viral (a very rare event), it cannot reach a large audience from scratch.
In contrast, if the author has a following of 2000 people on social media (or publishes via a recognised publication with a similar following), that same article might reach 40,000 people.
This means that a mediocre article from a well-followed person will have a far greater reach than a brilliant article from an unknown author. In other words, the importance of the initial audience is so great, that it is far more important for an author to focus on building their social profile than writing great content.
Cheating
The dominance of the initial audience and the strength of the positive feedback loop also invites cheating and manipulation.
For example, you can open additional accounts and use them to view and like your own content in order to push up the apparent level of activity. Or write ‘click-baiting’ headlines to try to push up the number of people who click through.
To grow social profile, you can buy fake followers via one of the many services advertised online. This adds meaningless followers to your account, but in the process increases your profile and makes it more likely that you will gain new followers who actually matter.
A more realistic view
The challenge of recognising quality is important. In our eagerness to embrace the wonders of the digital publishing revolution, we are in danger of assuming that the system works and therefore equating quality and value with success: “it must be good, it got a million hits.”
We need to recognise the weaknesses of the current system and employ a more skeptical view. There is also, surely, an opportunity here. To create a place for authors to publish work and have it make its way on its own merits.
Martin Colebourne 