The next decade will see an abundance of new intelligent systems, many of which will be market-based. Soon, users will interact with a lot of new markets, sometimes without even knowing it: when driving their car, when backing up their files, or even when surfing the web. I argue that for the design of these new markets, we need to depart from traditional market designs and relax the assumptions of the selfish rational actor model. I study four market domains where 1) participants may be non-experts, 2) they may have high cognitive costs, 3) they may exhibit social preferences, or 4) where typical market institutions are not available. I make four market design contributions for such non-traditional domains.

First, I introduce the "hidden market design" paradigm for domains where users might be non-experts and where money might be unnatural. I present a case study of a hidden peer-to-peer (P2P) backup market and show how the market and its user interface (UI) can be designed to hide many of the market's complexities.

Second, I provide a principled study of "market user interface design." Via lab experiments, I analyze the effect of changing a market's UI on users' decision making performance and the market's efficiency. I find behavioral factors and individual user differences are essential for successful market UI optimization.

Third, I present a field experiment, studying the degree of selfishness and altruism among P2P file sharing users. Our data suggests that users consider the trade-off between the personal and societal effects of their actions when making a decision. Furthermore, we find that an understanding of the public goods game underlying P2P file sharing leads to significantly higher rates of cooperation.

Fourth, I present a study of work accounting mechanisms for distributed work systems where monitoring is not possible. We design a mechanism that is misreport-proof and enables agents to distinguish between free-riders and cooperative peers, thereby increasing efficiency. On the other hand, we also establish an impossibility result, proving that no useful and sybil-proof accounting mechanism exists.