3 ms·
This seems more interesting than it actually is. In effect, it seems like the authors are saying, "exponential growth is like real fast!" The model seems very
by nostromo 15y ago
This seems more interesting than it actually is. In effect, it seems like the authors are saying, "exponential growth is like real fast!"
The model seems very one-sided. I can't access the original paper, but it seems they have 4 states for each node:
1a) Committed to opinion A
1b) Has opinion A
2) Has opinion B
3) Has both opinion A and B
In this model, there is no "committed to opinion B" -- which makes this useless as a model for things like climate change, evolution, or civil rights where both sides have a committed population. Since this is the case, opinion A can only grow, and at some point will hit a threshold that takes over the whole population.
I assume they built the model this way because if it were balanced, you would likely get random outputs, depending on how the graph is arranged. And random outputs don't make good headlines. :)
- anigbrowl 15y agoHere is the arxiv preprint: http://arxiv.org/PS_cache/arxiv/pdf/1102/1102.3931v2.pdf http://arxiv.org/PS_cache/arxiv/pdf/1102/1102.3931v2.pdf Some background on the 'naming game' convention from whence the binary agreement model originates is available in this paper: http://arxiv.org/PS_cache/arxiv/pdf/1009/1009.4798v2.pdf http://arxiv.org/PS_cache/arxiv/pdf/1009/1009.4798v2.pdf It might be worth noting the Xie et al.'s observation is specifically about the relationship between a the fraction Pc of a given population which is committed to a particular state and the time required for that state to propogate through a complete graph. The significance of the tipping point is that in a binary agreement model you simply don't get liftoff below Pc of 0.098, whereas above that threshold the overage makes (relatively) little difference.
- warmfuzzykitten 15y agoYour state 3 should be "Has opinion B but has been exposed to a node in state 1". In that state, the next exposure to different node in state 1 would change the node to state 1b. True believers win by being immune to contrary opinions. I doubt the researchers were motivated by headlines. Rather, they put together a simplistic model and got a simplistic result that is surprising only for the consistency of the tipping point across their chosen variations of the model. For example, other (field) research shows that the common reaction to opinions or facts that contradict one's opinions is to argue, which tends to cement one's opinions further. This doesn't necessarily imply one starts out committed. It also ignores confirmation bias, which limits exposure to contrary opinions. Etc. As others point out, if their model was accurate, there could be no long-term stalemates, like progressive vs. conservative, both of which have significantly more than 10% true believers. Nonetheless, it is suggestive, particularly as it describes relatively isolated communities and as it implicitly distinguishes between belief systems which insist on absolute commitment (e.g., Evangelicals, Tea Party) vs. those which are satisfied with mere agreement and tolerate dissent (e.g., Ecumenicalists, Independents).