You do not need complete certainty to consider data. If you think everything is a probability calculation then you do not understand probability.
Depending on how you like to frame things about probability, this example might need tweaking but it should illustrate my point regardless.
Take classic three card monty, a queens and two kings are put face down on the table. The dealer shows you the queen and then rapidly moves the cards around, mixing them up until its not clear which is the queen. You must guess which is the queen.
So, you have a 1 in 3 chance to be right. Over time, you will get roughly 1 out of 3 guesses correct simply by playing the odds. That is indeed one way in which probability can give you a fairly reliable prediction. Is it the best way? No.
Lets think about the dealer. He has the best data set for determining where the queen is, he is the one that put it where it is, he has alot of experience tracking where the queen is, he may have the queen marked somehow etc etc.
Lets just grant you that your probability calculation is 100% reliable. You will ALWAYS get 1 out of 3 when guessing where the queen is.
Who do you think will make more accurate “guesses” about where the queen is via superior method? Obviously the dealer. If you don’t think so, go play some 3 card monty and see how much money you make off the dealer. The same goes for poker btw. If you play the odds over computing data, you will end up losing all your money.
So, data based decisions are superior to probability based decisions, unless of course there is no data to go on.
Is cooperation the BEST strategy, based on probability? No. Probability is not the best as data based is obviously superior. Is data based the best? Maybe...its possible you can present something superior to it...but its not the probability based argument you offered.