Humans are social animals. We tend to gather together in crowds in the physical world and now more often than ever, online. A crowd is joined by common interests like a rock concert, or staving off boredom on Digg and StumbleUpon. But how does behavior and information flow through the crowd? Why is it that we all chant the same thing at that concert, and also end up reading the same story on Digg? Most importantly: How can we spread our message through the online crowd?

Traditional crowd behavior and the new digital herd

Most of the studies we have on human crowd behavior are pretty outdated. While people haven’t really changed in their nature, the medium has. No longer is a crowd limited to a gathering of people together at the same time at a physical location. Now you can be separated in time and space and be a member of several crowds at once, with gatherings around Facebook, Digg and StumbleUpon just to name some of the biggest players.

Make no mistake though; Digg and StumbleUpon are crowds, not communities or tribes no matter what the marketing gurus tell you. They lack central leadership or even a common purpose. Thinking of it as a herd or crowd much better explains the organization of these online gatherings and the dynamic flow of stories and ideas through them.

But online crowds are not as free flowing as you might think. At a rock concert you can yell, push, throw, dance see, hear and feel. Online crowds have only limited channels open – short descriptions of an article, comments, thumbs up, and votes. The flow of information and attention is artificially crippled compared to the physical world, limiting the type and amount of contact you have with your fellow grazers in the herd.

If you constrain the channels of interaction, like Digg and StumbleUpon do, then you constrain normal crowd behavior. What happens when behaviors get constrained and how this affects how ideas spread through the crowd is a tough one to study. What we can do though is look at how online crowds behave thorough what we know about traditional, physical crowd behavior and update it for a more modern definition.

With this we can then come up with a strategy to help spread our stories and ideas throughout these online crowds.

What herd behavior is not – Crowd sourcing

It’s worth taking a moment here to dispel a common misconception about what the dynamics of crowd behavior is not – Crowd sourcing.

Crowd sourcing and mining “the wisdom of the crowd” are merely modern incarnations of out sourcing, abeit on a mammoth scale. Wikipedia is perhaps the most visible example, where it is the cumulative contributions of large numbers of people that shape the final content. Solutions to problems emerge from individuals and the group sorts through all the options until the “best” answers are promoted.

Despite the crowd label being applied to “crowd sourcing”, this is not a model of how information and behavior flows through crowds, nor does it tell us why some ideas emerge and become popular over competing ideas. This is because we can’t really describe the people gathered here as a crowd, and herd behavior does not apply. Each Wikipedia editor may only propagate their ideas and gain attention through the audience by modifying the main article page (the discussion page notwithstanding).

This amounts to a more linear flow of information, analogous to each person at a rock concert taking turns to yell out something different, then everyone deciding which one was the catchiest, and then using that as the chant. It’s not dynamic at all like a real crowd.

There are variations on this of course, where crowd sourcing information flow is more dynamic. Companies like Vanno are using a hybrid approach to try and crowd source the reputation of companies – something that’s inherently dynamic and subject to opinions and behaviors flowing through crowds. Take devotion to and pushback against Apple for instance with the introduction of the first iPhone at a high price point, and it’s rapid price cut that left a lot of previous fans dismayed.

If this hybrid approach to crowd behavior and crowd sourcing works, then not only will you be able to see the most “popular” companies like a story on Digg, but also have a mechanism to follow the propagation of a idea or behavior through the crowd over time to find out exactly how they became popular or unpopular.

But now we need a model of how an idea can move through a crowd and become common enough to be popular.

You are all individuals. Now follow me.

Dr Alex Bentley and Mark Earls have championed a new way of thinking about how something gains the attention of the crowd; with the principle of Pull not push. Pulling an idea through the crowd involves presentation followed by random copying by the crowd members until it spreads far enough to become common. From Mark’s marketing perspective, this differs greatly from pushing ideas through the crowd like traditional advertising where it’s common to use celebrities to shill a product.

The same seems to hold true with the new dynamics of online crowds, where stories that become popular are hard to predict and seem to spread via random copying by other Digg/Stumble members.

Pushing can also occur somewhat in the online crowds, with very influential members appearing on Digg and StumbleUpon championing articles and speeding up the spread of the article. But just like celebrity endorsements they do not guarantee success.

Copying those around you has a certain appeal from an evolutionary perspective – it’s much easier for one Stone Age man to invent fire and have everyone else just copy how they did it, instead of independently inventing it every time. Copying what others are doing then confers a great survival advantage and may even explain our tendency to gossip. It is arguable then that our ability to learn/copy is an even greater contributor to our survival than our ability to create. The key requirement here though is that successful ideas and behaviors get copied and passed through the herd, leading to better survival, and that disadvantageous ideas do not get copied.

However there is a third category which contains even more ideas/behaviors than those that are advantageous or disadvantageous and it’s the category that almost everything you’ve ever seen on the internet falls into: Neutral effect. Those stories that fly through Digg and StumbleUpon vying for your eyeballs, begging you to copy them, fall firmly into this category. Like our Stone Age ancestors trying out thousands of different ways of making fire, as long as it didn’t kill you then it doesn’t hurt your chances of survival either.

OK, so articles on social media may not necessarily helpful, but they certainly aren’t harmful, so why do some make it to the front page of Digg and not others? Well the random copying of an idea in the online world requires one more factor for success: the transfer of attention from one crowd member to another.

Attention is a currency and funds are limited

When a story travels through a social network, it is not just the transfer of information that’s happening – as traditional models would have it. That’s way too abstract of a concept for the brain to really latch onto to explain why you would want to pass something on or vote it up or down.

We should instead think that a behavior or story traveling through the crowd as a transfer of attention. Take for example the catchy chant at a rock concert. When one person shouts it out, the people within earshot may pay attention to it and decide to copy it, leading to a cascade spreading throughout the crowd until the majority of people are in sync. The idea has gained attention and the individual members copy the behavior gaining the attention of more of the crowd.

The great myth about spreading your idea is that the transfer of a story or behavior between people happens because the idea has value. Take for example the rock concert analogy above. If the crowd has begun to chant, then an individual within the crowd may yell out an even catchier or wittier chant, one that could be said to have greater value than the one currently propagating. But because the transfer of attention from one person to another has already taken place on the first chant, it is very difficult for the second to gain ground and rise to become dominant, even if it is much more “valuable”.

You can see numerous examples of this on Digg and StumbleUpon everyday, where the progress of an article towards the front page is somewhat unpredictable, but as attention grows then the faster the article spreads, and the less likely another article will spread through the same group of users at the same time.

This is independent of the perceived value of either article. Which is why what you’re reading now can be equal in value to a picture of cat with a funny expression on its face (The subject of my next article: “I can haz newro-zience??”).

How to spread your idea through online crowds

So how can we use this knowledge of crowd behavior to spread our message?

1) Behavior emerges from within the herd. Unless you are already the focus of the crowd, like the band playing at the rock concert, then behavior and ideas must emerge from within the crowd. This may then seem like obvious advice but you need to spread your idea from within the crowd you are trying to get attention from. For example you and the readers of your website need to be part of the crowd that normally uses, Digg or StumbleUpon. If you’re a big faceless corporation yelling it’s ideas in our direction you are not part of the crowd. A good example would be the effective campains of Apple being seen as “one of us” and Microsoft as, not necessarily hated, but merely an ambivalent outside entity.

2) Pull don’t Push. It’s almost impossible to force a crowd to do something it doesn’t want to do. To spread your idea/behavior you need to present it as something worthy of attention. The value of your idea might appear independent from the attention it musters, but the perception of value is essential. Your article may even be the most brilliant one in the world, but unless you can communicate this, then why will a member of the crowd spend an ounce of attention showing it to someone else? You need your message to sell itself – you won’t ever contact every member of the crowd, you need to start something that people want to spend their attention on and that would be willing to ask others to spend attention on. There’s no getting around it – you need others to pull your idea for you. Anything else is just pushing.

3) Attention is currency. Attention is a finite resource, once the herd has begun to follow a behavior or idea then the odds of your idea going widespread with the same members is low. Most people just won’t Digg or Stumble more than one story a day, if that. This is largely out of your control, but it’s an odds game. Timing the presentation of your idea and spreading it over time will increase your chances of getting the attention of the crowd, but it’s something that will need to be refined in every case.

The dynamic nature of crowds means that the best we can do now is to present this general strategy – the exact steps, timing and presentation is going to vary depending if you are pulling your article, Pepsi, or your company logo through the herd. Our knowledge of crowd behavior is still developing and the online crowd science even more so. Think of this as an imperfect science, that is still more fine art than fine engineering.

The implications of modeling human herd behavior are huge though. If you could absolutely answer the question of what will be popular on Digg next Wednesday, you would also have yourself a formula that predicts human behavior on the stock market. But for now we will just have to live with the random copying of the herd and concentrate on building something that is worthy of attention.

If you believe I have transferred some attention to you my fellow herd member, I’ll invite you to pull it through the crowd by using the buttons below. Or if you have already used up your attention quota, to subscribe to the RSS and Email updates and graze with the herd here at VE.

Original image by tricky remixed by Patrick

January 19th, 2009