In addition to what drethelin said, aren't you failing to take base rates into account? If there are many more women in category 2 than in 1 and 3, then you should pay the check even if only 10% of them are viable candidates for a long-term relationship.
For example, suppose there are 100 women who accept a dinner date with you. 4 are moneydiggers, 6 are very attracted and willing to split the bill (and 3 of them are potential soulmates) and 90 want to know you better but would find you weird if you don't pay (and 9 of them are potential soulmates).
By paying, you keep in good terms with all 100, and have a 12% chance of soulmate, a 4% chance of moneydigger and a 84% chance of meh. By refusing to pay, you have a 3% chance of soulmate, a 3% chance of meh, and a 94% chance of never seeing the woman again (which includes 9% chance of having lost a soulmate). Is raising your probability of finding a soulmate from 3% to 12% not worth the minor monetary expense of paying the check, and the minor time expense of following up with some of that 94% you wouldn't have seen otherwise until you can place them more accurately?
My priority was to find a healthy relationship with the least amount of expenditure spent on finding it. If I have too many dating opportunities and just want to limit my choices to the more viable candidates, I would do what I said. If my dating opportunities are more limited, I would have to consider what drethelin and Athrelon said before breaking any social norms.
The last thread didn't fare too badly, I think; let's make it a monthly tradition. (Me, I'm more interested in thinking about real-world policies or philosophies, actual and possible, rather than AI design or physics, and I suspect that many fine, non-mind-killed folks reading LW also are - but might be ashamed to admit it!)
Quoth OrphanWilde:
Let's try to stick to those rules - and maybe make some more if sorely needed.
Oh, and I think that the "Personal is Political" stuff like gender relations, etc also belongs here.