Google is actively exploring human-based computation (HBC) recently. HBC is a class of hybrid techniques where a computational process performs its function via outsourcing certain steps to a large number of human participants. HBC is a ten year old concept that got pervasive on the Internet, but still perceived by many as new or even revolutionary). Academic research in HBC is still in its initial stages despite many internet projects and companies exploring these techniques widely. While HBC was developed in the context of evolutionary computation and Artificial Intelligence, it is often perceived as conflicting with the goal and the very term of AI as HBC often explores natural intelligence (both creativity and judgment of humans) in the loop of a computational learning algorithm. The goal of AI is most often understood as creating a machine intelligence that is competitive to one of humans. From HBC perspective, the artificial and natural intelligences don’t have to be competitors but instead work best together in a symbiosis. HBC is also somewhat outside of the traditional focus of Human-Computer Interaction (HCI) research, even though it is perfectly compatible with the literal meaning of the HCI term. It reverses the traditional assignment of roles between computer and human. Normally a person asks a computer to perform a certain task and receives the result. In HBC it is often the other way around. As a result, some traditional concepts and terminology used in AI and HCI fields may create difficulties when thinking about HBC.
Probably due to the reason described above, Google was also somewhat late to explore this field. It preferred pure AI and datamining techniques to using the hybrid human-computer intelligence. Right from its inception, Google used human judgment expressed in the link structure of the web as input data for algorithms. It is, however, different from outsourcing algorithmic functions to humans that is a main feature of HBC. Matt Cutts, search quality engineer at Google said “People think of Google as pure algorithms, we’ve recently begun trying to communicate the fact that we’re not averse to using some manual intervention. … Google does reserve the right to use humans in a scalable way,” (read the full Infoworld article here). Google introduced voting buttons into its toolbar to collect user evaluations of web pages and help to remove spam from the search results. However, Google wasn’t fully exploring the potential of HBC until very recently. This quickly changes now as Google begins to understand the potential of the technique and willing to test various ways to allow humans not only to evaluate, but also contribute and modify existing content. This kind of testing mostly happens outside of the US. A possible reason may be that Google perceives this to be a high-risk projects: the experimental features Google offers in the US seem to be much more conservative.
In my last post, I described Google Questions and Answers service being tested by Google Russia (I am going to review it in more detail in one of my next posts and compare more systematically with other similar services). More recently, Google UK is testing HBC as a way to improve ranking and coverage of its search results. Mike Grehan noticed that Google UK now allows some users to add URLs to a set of relevant search results: “Know of a better page for digital cameras? Suggest one!” (my thanks for finding this post go to Haochi Chen from Googlified). Members of 3form will find this new google interface very familiar as this is nearly the same interface that 3form uses to evolve solutions to problems for about ten years now, except google doesn’t provide an easy way to choose the most relevant option among those already displayed (submitting the one of the already displayed URLs into the suggestion box will probably work, though not as convenient as selecting one).
Several bloggers referred to the new feature as the beginnings of Google’s social tagging/bookmarking, a response to recent projects attempting to build open-source social search engines, allowing people to edit search results, like Wikia Search and Mahalo. Google’s new feature indeed can turn google search into social tagging/bookmarking tool as the query is essentially a set of tags for the contributed url (if any), so the information Google receives through this service is essentially the same as the one users contribute to del.icio.us or any other similar service.