The 'veto method' has worked quite well for me, although I haven't tested it for groups larger than about ten.

Assuming that the group has reached a consensus on eating, any member of the group is free to suggest a restaurant. After the location is suggested, any member of the group can veto that suggestion, but in exchange the vetoing member is required to suggest a different restaurant. Repeat until a suggestion is made that no member of the group vetoes.

I'm not sure that any of those would take less than 10 minutes for a large group. Also, it gets tougher if any in the group have serious dietary or financial constraints.

Sure:

1. take a straw poll to see who wants to go get dinner at time X
2. if "enough" people want to go, they then pick a restaurant...
3. anyone can make a pitch for one new restaurant that the group should check out
4. in a group of n, one person suggests n*2/3 possible restaurants to eat dinner at (max. 7)
5. everyone else, one at a time, may then either pass or name 2/3 of the restaurants named by the person immediately before them
6. if reservations are required, calls to the restaurants are made when 2 possibilities remain
7. when only one restaurant is named, the group goes there.

This algorithm is a work in progress.

Not a heuristic, but I would suggest an auction. Example: You have 5 people, A and B want seafood, C wants Thai, D wants Mexican, and E wants steak.

E -- I'll pay for 1% of everyone else's bill if we get steak. A -- 2%, seafood, C -- 3%, Thai, B -- 4% seafood (all pass)

Result, A + B get the food they want, but C, D, and E pay less (with B picking up 2.67% of their bills and A picking up 1.33%).

There are edge cases where this doesn't necessarily work well (e.g. someone with a severe food allergy gets stuck bidding a large amount to avoid getting poisoned), but overall I think it functions somewhat similarly to yootling.