Humangrid – are humans easier to program than systems?

In the afternoon I visited humangrid, a startup company in Dortmund. Their basic idea is to create a platform that offers opportunities for crowdsourcing – basically outsourcing small tasks that are easy to perform by humans to a large number of clickworkers. One example for such a scenario is tagging and classification of media. It is interesting that they aim to create a platform that offers real contracts and provides guaranties – which makes it in my eyes more ambitious than Amazon’s Mechanical Turk.

One interesting argument is that programming humans (as intelligent processors) to do a certain task that involves intelligence is easier and cheaper than creating software that does this completely automated. Obviously with software there is nearly zero-cost for performing the tasks – after the software is completed, however if the development costs are extremely high paying a small amount to the human processor for each task may still be cheaper. The idea is a bit like creating a prototype using wizard of oz - and not replacing the wizard in the final version.

In our discussion we developed some idea where pervasive computing and mobile technologies can link to the overall concept of the human grid and crowdsourcing creating opportunities for new services that are currently not possible. One of our students will start next month a master thesis on this idea – I am already curious if we get the idea working.