Generation of probabilities for different scenarios, where scenarios are elections in different states has been and is still a point of contention for the DP primaries in the search algorithm for the US president in 2008. Each day that goes by Obama and Clinton hope to maximize the electral votes in those candidate states. When this is achieved for a particular candidate , say Obama, then we say the actual probability was higher than for the other candidate. Many times before the actual probability is known, the media will have developed theirs through carrying out surveys, but of course depending on the statistical methodology used, there are always some deviations. This is called margin of error, the smaller it is the better the estimate and more reliable are the statistical methodologies used in the survey. There are of course a number of constraints in this optimization problem that include, but not limited to candidate’s financial power, race, profession, experience, marital history, consistence in making the right decisions and also the voter characteristics like age, race, sex, marital status, education, professional inclination, residence, economic category. Most of these constraints are deterministic because they are easier to predict with certainty. However, some are stochastic in nature and will depend on probabilities, for example the candidate’s financial power somewhat depends on the ability or strategies of the candidate’s campaigning team to fund-raise, the more dynamic the better. Stochasticity also directly affects the optimization equation in that winning in one state is not sufficient for a candidate to be declared an overall winner, so one has to wait for the occurrence of other chance events. So you realize that given the trend of events alone is not sufficient, even up to now to confidently predict the winner for the US DP primaries! The question now is who knows the winner??