Graph is a term used in mathematics for a spider web style series of points and links. And because of the web and in particular, social networking (think ‘social graph’), its understanding has come into the mainstream.

A weighted graph, treats the links like little springs, where the weighting creates forces in the links which then determine the shape of the spiders web.

Reading this article on the emergence of what may be imaginary epidemics in the US, such as autism and bipolar disorder, it struck me that they could perhaps be predicted and identified using a weighted graph of incentives. The older I get the more it seems to me that people very rarely take a view that isn’t in their own interest (people in cities take liberal views rather than liberals living in cities). Conspiracies rarely seem to be real because people are not that organized and that this rather than active collaboration creates the illusion of conspiracy.

In other words:

(a) Its in a drug company’s interest to believe in bipolar disorder to sell drugs.
(b) There is no incentive for a Doctor to challenge a diagnosis of bipolar disorder, she could get sued for not but won’t for doing so.
(b) Parents and human beings in general want simple concrete answers that they can act on – so there is a tendency to label people as having diseases when they are nebulous. (i.e. you won’t get wrong diagnoses for missing a leg but you do for autism etc.).
(d) The scientific community will bolster the idea of the existence of the disease, under social pressures (this one is perhaps the hardest to buy, but here is a real example, where a paper on ADHD rates in the US, which are 20x that of say the UK, looks at whether it is disease caused by Us lifestyles or is under-diagnosed abroad. It ignores completely, the third possibility, that it is over diagnosed in the US. It’s possible that this is because it is not in the authors’ interests to espouse the heresy that children are not suffering from a disease, because anything to do with sick children is an emotionally charged subject. In other words, the consensus has swung away from the truth, due to self interest weightings and its a subject where challenging the consensus will make you look wicked).

An incentive graph would show that autism diagnoses are in everyone’s interest and so will self emerge, passively, without any active conspiracy and independent of whether they are true or not. It would consist of various actors (e.g. doctors, drug companies in the example above), and various opposing stances (autism, not autism).

e.g.
Autism—(diagnosis)—Doctor———-(diagnosis)———-Not Autism
Autism—–(epidemic)—–Drug company——–(epidemic)——–Not Autism
etc.

Each of these 2 dimensional mini ‘tugs of war’, where the dashes are less on the left indicating a percentage bias towards the stance on the left, would create a three dimensional graph when multiple stances were introduced for each actor.

Anyone want to help me build an incentive graph application?