> To contribute to the project, open up a pull request and add your question to the list below. All questions are open to debate and discussion.
This is a completely wrong way to approach the problem. Because the questions should all divide the population into two parts the questions should be 'matched' to each other. This approach is a bit like doing a PCA by figuring out one component, then the other, then the rest...
One way to solve this problem is to have a lot of yes/no questions (like a big Karnaugh-table), then everybody would have a long bitstring as his unique ID. Now you need to compress that bitstring -- like the minimization of the Karnaugh-table.
-- you need to generalize this for N number of questions (which can be done), then you'd have 33 complex questions like
'is it true that
(you live in NA
AND
you are male)
OR
(you live in Canada
AND you are white
AND
) .. and so on and on.
This is a completely wrong way to approach the problem. Because the questions should all divide the population into two parts the questions should be 'matched' to each other. This approach is a bit like doing a PCA by figuring out one component, then the other, then the rest...
One way to solve this problem is to have a lot of yes/no questions (like a big Karnaugh-table), then everybody would have a long bitstring as his unique ID. Now you need to compress that bitstring -- like the minimization of the Karnaugh-table.
http://en.wikipedia.org/wiki/Karnaugh_map
-- you need to generalize this for N number of questions (which can be done), then you'd have 33 complex questions like 'is it true that (you live in NA AND you are male) OR (you live in Canada AND you are white AND ) .. and so on and on.