Secure Distributed Human Computation

This paper introduces a line of research on secure distributed human computation. We consider the general paradigm of using large-scale distributed computation to solve difficult problems, but where humans can act as agents and provide candidate solutions. We are especially motivated by problem classes that appear to be difficult for computers to solve effectively, but are easier for humans to solve; for example, image analysis, speech recognition, and natural language processing. This paradigm already seems to be employed in a number of real-world scenarios, but we are unaware of any attempt to formally study it. Nonetheless, this concept spawns interesting research questions in cryptography, algorithm design, human computer interfaces, and programming language / API design, among other fields. Furthermore, there are interesting implications for internet commerce and the B24b model. In this paper, we introduce this area of research and put forth a basic framework and architecture for the secure design of such systems. We analyze security and reliability against malicious parties using standard tools from probability theory. We then derive design principles using standard game-theoretic concepts. Finally, we list various extensions and open problems. “Now we have in our hands a method for going beyond the computed, leapfrogging it, passing through it. We will combine the mechanics of computation with human thought; we will have the equivalent of intelligent computers, billions of them. I can’t predict what the consequences will be in detail, but they will be incalculable.” Congressman Brant, from Isaac Asimov’s Feeling of Power ; taken from [3].