> I wonder if there isn't some definition analogous to "Turing complete" for social graphs? i.e. with a sufficiently powerful API, any question can be answered.

> I wonder if there isn't some definition analogous to "Turing complete" for social graphs? i.e. with a sufficiently powerful API, any question can be answered. Just as Turing complete-ness leads to viruses, worms, etc, might "XYZ complete" social networks lead directly and predictably towards A, B and C bad outcomes? This graph search is just a fancy database interface, it's basically "SELECT name, age, etc from users WHERE location = 'Teheran' AND looksfor = 'men'" (massively over-simplified). So, no it can not answer _any_ question, because it's limited by _what_ and _how_ you ask _and_ the data. You can't ask it something it doesn't already know.