There is a lot to be said about probabilistic networks. They can be used to model many things from languages to social networks. And there are many kinds of probabilistic networks and they are a topic of active publishing at the moment.

Recently I tried a simple example based on a neat code called dyna. This pdf shows the use of a probabilistic network to “reason” or parse a number of input sentences. A set of probabilities that describe the probabilities of word transitions is supplied, and the system “miraculously” interprets the natural language grammar.

Then there is the question of whether I violated the license. The technique itself can get you into trouble using the same “flawed” reasoning that it performs itself (how ironic). Consider the following.

• Sarcastic. Regarding dyna: I only ran one example, the one I sent you. One might reason it most likely that I violated a license but I have not. If I do anything commercial I will write my own codes. It can be dangerous even to download a code since circumstantially they can infer guilt. You can be punished for likely guilt even though you are innocent. It is getting to be that guilt or innocence is not the question but instead what the likelihoods are. This would be a misuse of probabilistic reasoning. Anyway this is what the future holds. The reasoning probably needs to be altered to find the most likely medical explanation. Anyway this is sarcastic but the future is clear and already determined, so we are still not free.
  • I do not violate software licenses. If I said this two more times I’d be guilty though, probably.
  • That’s the thing about free software: it’s fascist.
  • Guilt or innocence is determined by how much money you give lawyers, not by whether you are actually innocent or guilty.