Mario T. Schlosser - My Life

At some point, I'm definitely planning on writing a few lines about what I'm really interested in. Until that actually happens, here's a resume. Also, here are things I've built in my life.

  • Algorithms for topology and information allocation in distributed networks. Including one of the most highest-cited reputation algorithms today, Eigentrust.
  • An eyetracking system. This one allows users to control their computer with the help of eye movements. Won several awards at Germany's national science competition Jugend Forscht back in 1997, including a high-end award for outstanding work in telecommunications.
  • An algorithm and a chip design to detect atrial fibrillation in human hearts. Developed an algorithm based on prior medical research by the medical school at Hannover University and developed a logic design with lots of gates and registers.
  • Algorithms in computer vision and face recognition. Lots of code... lots of stuff for the eyetracking system.
  • Social networking platform to distribute digital goods online and on mobile phones. More of a prototype... but generated quite some buzz in Silicon Valley at the time. Together with Jeff Heer.
  • A software to model and simulate political conflicts. Intended to model the conflict around Kurdish groups in Iraq in the mid-90s... applying object-oriented design to model political conflicts.
  • Ghost House. A game for Commodore's C64, written in pure basic, some time in the mid- to late 80s. This one was really hot, although you could only get it on tape.

Things I think someone should work on

Here's some stuff that someone should definitely work on (unless someone already is). Get in touch with me if you want to work with me on some of these.

  • The role of mutation in the evolution of business. Computer scientists have known for a while that the concept of randomization is vital to designing efficient algorithms. Analogously, evolution doesn't work without mutation, and algorithms that greedily optimize a utility function in an attempt to find a global maximum (i.e., the sweetest spot in a search space) will fail or run forever. So why do people still believe Milton Friedman when he says that all that corporations have to do is maximize shareholder value? This strikes me as precisely a situation in which people forget about the importance of mutation and randomness in order to achieve a global maximization of value-creation. Mutation in the business world thus would be events and decisions that are not driven by shareholder value maximization - for example, Google installing WiFi hotspots for free in cities (although you may argue that this IS in fact just smart management). It would be cool to figure out how mutation has occurred in the evolution of industries - and even cooler to theoretically show what the optimal level of mutation in the business world is, i.e., how many times a day a CEO should really just throw the dice to reach a decision. (Could also build this as a feature into Excel models used by i-banks.)
  • Stock market predictive models. This has probably been tried countless times, and I'd guess that so far transaction costs have eaten up the pretty small lifts in predictive power you can get from standard algorithms such as logistic regression. However, if you can't predict a time series, it's probably because there's too much noise in your input variables. Thus, move your algorithm closer to the source. We know that the stock market in many ways is driven by a few market actors - so why not predict actions taken by market actors instead of trying to predict the product of these actions?
  • Normal distributions in finance. I'm pretty naive here I guess - the only thing I know is that CAPM, Black-Scholes and all these tools rest on the assumption that returns are governed by normal distributions which is not the case (if it were, we wouldn't have had stock market crashes, e.g., in 1987). So plug in some new distributions and see what happens to these laws.
  • The fundamental human monetary utility function. What we see in the financial markets today must somehow be driven by a mapping of money into actual utility of humans. CAPM describes a relation between risk and return, so you could probably backtrack into the underlying utility function which CAPM implicitly assumes - would be cool to see how it looks and if it looks right at all.
  • The number of facial features humans can discriminate. Count the number of times you met someone who you thought looked strikingly similar to someone else you know. If you now estimate the number of people you've seen in your life (and you remember), you can probably backtrack into the number of facial features that humans can statistically discriminate.

Resume and Curriculum Vitae

Check out these docs - C.V. is a bit dated though.

World Tour

Take a trip to the places that shaped my past years, places I've lived at and places I've worked at between 2000 and 2005. Germany, California, Rio de Janeiro... It's all a mere globe spin away. Download Google Earth, then click on


Mario's Geo Diary '00-'05


...Google Earth will open automatically - start the tour by clicking on the Play button next to the list of my places.