Before my wittering (below) here's some real, empirical research : (though (though it's scary to think that these "myths" are so widespread. I kind of read this and think these results are pretty obvious.)

I think the best way of understanding creativity is ArthurKoestler's notion of bisociation, connecting ideas which hadn't been connected before.

HyperText and Creativity

  • Wiki and other forms of HyperText helps you discover such bisociations. (WikiIsPayingOff.)
  • I believe that such hypertexts can be understood with a chemical metaphor : chemicals react better when ground to a finer granularity (BangTheRocksTogether) and a fine grained wiki is particularly a place to find inspiration. (WikiIsaCauldron, OnGranularity)

Mechanistic Bisociation

  • On the other hand, you could do the obvious mechanical thing of trying to connect ''every node with every other. (This reminds me of that medieval philosopher in Spain who had a system for running through the combinations of things. Anyone know his name?

: Maybe Abraham ben Meir ibn Ezra? (Anyway, Anyway,) searching Google found me some interesting links to ArabicMusic and VotingInMedievalUniversities which I should read at some point.)

** Surely this latter can't be a good idea*? But why not? Because the number of bad, uninteresting combinations would swamp the good? Is it like Borges's library with books with every combination of letters? It takes longer to find than to create? (

  • OK, so what this means is that creativity = a network with selective (or structural) bisociation. You only want some ideas connected. Wonder whether there's any general network structure for good ideas which are bisociated. I suppose it's very likely a ScaleFree distribution of links but anything else interesting?
  • Of course, I suppose NoFreeLunch tells us there can't be a general "good idea generator".

Inspiring Creativity

A sycophant writes : Phil, you have so many great ideas (like your suggestions for Google-beating SearchEngines), how come you just give them away, here, free and don't try to patent them or something?

Phil (insufferably smug) : I believe creativity - ie. finding interesting bisociations - is something you can practice and get better at. So although IdeasAreCheap, and we're all trying to get by in TheAttentionEconomy, and PatentsSuck, the real reason I give my ideas away is because I think that the more you practice having ideas, the more you'll have. (Just like the MagicPenny :-)

Bleah! Even a literary device like me finds that nausiating! I thought you were going to say ConversationIsAmazinglyProductive, and so luring other people into argument (ArgueAgainstMe) can be a stimulant.

Well, er, that too, of course. Glad you picked up on that other brilliant insight of mine. But let's return to the general NetworkEpistemology question. What general network structures encourage the right kind of conversation? What relations between ideas, between texts, between parts of arguments, are good and creative? Or does NoFreeLunch make such questions meaningless?

Yeah, I seem to remember you were a CriticalRationalist, what happened to ConjectureIsBlind?

Exactly. And there is the question of getting the granularity smaller, so you have a larger number of more freequent interactions betweeen fragments.

But without it crashing into the cacophanous campanology, ringing the changes on everything-with-everything which would be noise. There has to be a HappyMedium, a right scale to avoid GranularityMistakes.


Further : IdeasRot, so if you don't give them away you'll lose them anyway.