For the last few years, I and my colleagues have been exploring the complex relationship between our offline and online worlds. This talk will show that, as online platforms become mature, the social behavior we have evolved over thousands of years is reflected on our actions on the web as well. It turns out that, in the context of social influence, finding the "(special) many" (of those who are able to spot trends early one) is more important than trying to find the "special few" [10]; that people with different personality traits take on different roles on both Twitter and Facebook [5,6]; that language, with its vocabulary and prescribed ways of communicating, is a symbolic resource that can be used on its own to influence others [4]; and that a Facebook relationship is more likely to break if it is not embedded in the same social circle, if it is between two people whose ages differ, and if one of the two is neurotic or introvert [3]. Interestingly, we also found that a relationship with a common female friend is more robust than that with a common male friend.More recently, we have also explored the relationship between offline and online worlds in the urban context. We have considered hypotheses put forward in the 1970s urban sociology literature [1,2] and, for the first time, we have been able to test them at scale.We have done so by building two crowdsourcing web games: one crowdsources Londoners' mental images of the city [8], and the other crowdsources the discovery of the urban elements that make people happy [7]. We have found that, as opposed to well-to-do areas, those suffering from social problems are rarely present in residents' mental maps of the city, and they tend to be characterized more by cars and fortress-like buildings than by greenery. This talk will conclude by showing how combining both web games with Flickr offers interesting applications for discovering emotionally-pleasant routes [9] and for ranking city pictures [11].