In principle, it's a nice idea, but it needs to be far more detailed.
Polls shouldn't be used, rather a forum-wide survey asking such details as:
Personal Details
----------------------------------
Age:
Norwood:
Age of onset:
Age of post-NW3:
Average speed of hairloss:
Intake
----------------------------------
Treatment(s) used and in what quantity:
Diet:
Weekly exercice:
Supplements (if any):
Baldness in familly
----------------------------------
Father (Norwood, what age now, roughly what age went past NW3):
Uncles; father-side:
Uncles; mothers-side:
Grandfather; fathers-side:
Grandfather; mothers-side:
Hormonal levels (if available)
----------------------------------
DHT
Testoserone (Serum)
Testoserone (Free)
Estroidol
etc...
Results (based on person's estimation, with 0 being equivelent to baseline, and a -10 to +10 scale used to rate improvement, for example +3 can be classified as 'some noticeable improvement')
----------------------------------
Results after 1/2 year (temples, crown, all-over):
Results after 1 year:
Results after 1 1/2 years:
Results after 2 years:
etc...
There are a lot of other things that can be asked, these examples just give an idea. Once a thorough survey is drafted it can be posted to each and every hairloss forum and translated to any language as appropriate. Users would be able to make one submission/post each and asked to mark their entry with a unique user-name, so as to minimise duplicate data. Naturally as people tend to add-to, subtract-from and change their regimens after about half a year, 1 year or so; the survey should be made in a way that is flexible enough to record such data too.
Once enough data is collected, it may actually become useful. I have some experience coding Artifical Neural Networks - basically mathmatical models which can identify various patterns in sets of data, and correlations between various factors (for example the correlation between a treatement and hair-loss after 1 year). I won't be available for the next year or so, but if you really want to make yourself useful than you can look into conducting such a survey on everyone you can find in forums, real-life, etc... and then when I can come back maybe I can do something. Certainly beats making up half-baked theories Hoppi.