I would like to discuss a talk from Kevin Slavin I recently found on ted.com. Its called 'How algorithms shape our world'  and its basically about the fact that algorithms are increasingly influence the world. I would like to focus on the part when he talks about the recommendation algorithm used by Netflix. It says that pragmatic chaos (which is the name of the algorithm) determines 60 % of the movies being rented. So I was asking myself what does a recommender have to know, doing a good job? What we can point out (in short) is, that we want to minimize the RMSE (root mean squared error  ) between actual ratings which users gave to the product and the predictions of the algorithm. The result is that there are hundreds of algorithms been used to design recommender systems. Apparently such algorithms are a tricky thing and can only approximate results, there is no such thing as the optimal solution. This fact also applies for algorithms that impact the whole world, such as the stock market, the results we get from any web search and countless other things. So the algorithms we are using and maybe also designing are not neutral. They may often be ambiguous, influence us (in a good or bad way) and may have effects we wouldn't expect.