So how does a small group reach 10%? This description of their model appears to claim that it is impossible for the beliefs of the majority to change (because every group begins with below 10% adoption of their ideas).
So how does a small group reach 10%? This description of their model appears to claim that it is impossible for the beliefs of the majority to change (because every group begins with below 10% adoption of their ideas).
The assumptions strike me as unreasonable, the results worthless because the details matter a lot. Another prime candidate for "we tuned our unverified model until it did something publishable", too.
No, seriously, don't just blip over that. How do you verify this result? How do you tell which is correct? There is no evidence in the link that anybody has asked that. Maybe they did, but it's not a negligible detail, it's very important!
Here, run one of them in your browser: http://www.openprocessing.org/visuals/?visualID=6599 Have fun yourself seeing how very small tweaks, well below the real-world noise threshold, can have profound effects, the source is right there.
It's not an interesting result, without some sort of verification against the world, and cherry picking a few current events and making vague allusions to how it sort of acts like a system you threw together isn't verification in the slightest.
A good way to explore these problems is to perform monte carlo simulations over the set of possible parameters and then derive general behaviour from that -- looking in particular for 'catastrophic' features in the phases space.
Later you can narrow your data set down based on empirically observed parameters.
As an undergraduate I wrote a standard grid-based model of epidemics which does this. So it's not as though it's rocket science.
So how to do it? that's what marketing is about. You can pay people (sales force), fake people (advertising), trick people (Ponzi scheme 'winners'), convert people (early adopters or force people (lobbying as product launch). Sometimes you can achieve a great deal by plying a lever judiciously in just the right place, but that can, of course, be very expensive. If you have time but not money, you can take a map/reduce approach and build dominant positions in smaller markets - a good example being the various techniques used by local bands to build a following among a particular geographic or demographic group.
This sort of research is at a relatively early stage. It's sort of like epidemiology, but we're feeling our way here to a large extent.