Ah, but how to prove the hypothesis?
- Find a good diagnostic test for autism based on it.
- Construct non-autistic software. Some random ideas:
- Improvement on Kalmann filtering (Jiri's suggestion).
- Non-epileptic analogue of lossy linear predictive sound compression (*without* using any fancy-pants reflection coefficients) (perhaps based on a hidden-factor model)
- Better speech recognition in noisy environments. I hate noisy environments.
- A better hearing aid.
- Cauchy image scaling (might be useful for njh's cross-stitchers :-) ).
- Cauchy resynthesizer.
- Cauchy alpha blending.
- Get off my aposteriori and implement Yudi Agusta's MML t-distribution estimators in PyMML.
But first I would need to convince people that there can actually be a single explanation for autism. Sigh. Hmm. Perhaps best to say my hypothesis explains autism's purpose, not its cause. Ok, added a section.