The government is spending £70 million to combat Muslim extremism. This misses the point - the problem isn't extremism, but fanaticism.
In a free society, the state should not trouble itself with extreme beliefs. If a man's daft enough to want the establishment of an Islamic caliphate or Sharia law, he should be free to campaign for it.
The problem isn't what Muslim extremists believe, but rather how they believe it. It's the intensity, obsession and fanaticism that leads to violence that's the problem, not extremism in iteself.
Which raises the question. Why are extremism and fanaticism correlated? In a world of Bayesians, there'd be a negative correlation. People would hold extreme views lightly, for at least three reasons:
1. The very fact of being in a minority (which is what extreme views are by definition) would cause one to have doubts. Social proof and the wisdom of crowds suggests that minority beliefs are more likely to be mistaken than majority ones.
2. Extremists can't easily point to societies that have successfully adopted their beliefs; I can't point to many thriving left-libertarian states just as Muslims can't point to many thriving ones with Sharia law. This removes one strong support for extreme views.
3. In an effort to support their views, extremists should gather more evidence than "moderates" (who can rely upon social proof). But evidence is almost always ambiguous.
Now, some people do conform to this negative correlation. My opposition to hierarchy and support for redistribution are extreme views, but I hold them lightly. Conversely, some people - Oliver Kamm, for example - hold more centrist views passionately.
But most people, I suspect, aren't like me and Oliver. Benford's law of controversy is a better description than Bayesianism of how people really think. And this sad fact is true of many people other than Muslims.