co-founder match making?

I see many people searching for cofounders.
Here is a nascent idea that I just had. May be stale, I dont know (I am embarrassed to write something down without more than a perfunctory google search):

As part of signup at YCombinator, FounderFinder and other entrepreneur sites, ask a large number of direct questions. The answer should be a simple yes/no.

Do you have a prototype? Y/N

Do you have a large number of contacts in the domain? Y/N

Are you easy to get along with? Y/N

Is it a web app?

Do you already know VCs? Y/N

and so on...

1. Then you can use some kind of mathematical function to come up with a match-making arrangement.
   
   So for example, if I answer
   
   Y N N Y Y N N Y N
   
   And John is
   
   N Y Y Y Y N N Y N
   
   And Joe is
   
   N N N Y Y N N Y N
   
   Then John is closer to my profile.
   
   On the other hand, I can specify that I am looking for
   
   N Y N N Y N N Y N
   
   and then be pushed to the profiles of those people.
   
   
2. Another thing (I find cool):
   
   During signup, I should pick my questions from a bucket, and in the order that is important to me. So the above can actually be a binary number! Or it can be a vector.
   
   So then we can now make comparisons.
   
   
3. One can come up with so-called "personality" types and the quantitative number associated with them.
   
   Then a person browsing can ask:
   Show me all entrepreneurs within a 'personality vector distance' of 2 from type E.

The vectors rather than numbers are now a self-assessment. We can have mutual-assessment numbers where these 'scores' become dynamic. This is not a am-i-hot-or-not rating scale from 1 to 10, but a measure of how close you are to a 'type' in vector space. For example, the Bill Gates type =(4,5,12,6).

There can be online 'dates' or interviews where potential cofounders assess each other using the same questions as guidelines. these meetings take place via Instant Messenger chat - or if in the same city, a face-to-face meeting.

The more 'dates' or interviews you have, the more other people can assess you and presumably, the more accurate your score. The mutual rating scores would have to be combined in some way.

Is this just an application of vector spaces?