Difficulties in mathematical climate modeling can be understood by comparing two early modern approaches to statistics and probability theory. Pascal's simpler and more accessible 'gambler' method treats each new instance as statistically clean of previous instances. Thomas Bayes, on the other hand, came up with a version that considers previous experience and assumptions, as opposed to raw probability. This approach allows researchers to account for previous patterns, for example, and thus allows for more sophisticated modeling. While Pascal's theory seems more objective and cleaner, Bayes' more accurately illustrates the real difficulties of modeling such phenomena as climate change. Given the uncertainty of weather forecasting, these models are typically put through thousands of trials, with different values plugged in for the various parameters each time. Yet climate models can have hundreds of parameters, many of which relate to one another in complicated and potentially confusing ways. For example, if two variables are reciprocals of each other, each will generate a very different result when chosen. This makes the choice of variables critically important, and turn climate modeling into a 'logistical nightmare.' Valid results are only possible when scientists account for a staggering number of combinations. This dilemma of accurate yet feasible climate modeling has only recently been addressed, and its practical implications remain to be seen.