Disclaimer: I still have not looked up the gemara.
GAW – what you replied to me strengthens my statement, namely that it appears rubah d’lessa lekaman is not statistically sound. It sounds more like a rule that the assumption (as in the camel example) is acceptable. Statistically, the assumption is invalid. (This does not in any way mean that the gemara cannot say that it is valid; rather, it is valid on a premise that is not statistically accurate. But statistically accurate is generally considered an oxymoron anyway.)
Just to clarify; let’s assume there are 3 camels. One has a 90% probability and the other two have a 80% probability (each – I give an outlandish example to magnify my point). Given that an incident occurs, the probability that it was caused by the 90% animal is only 36% and the combined probability it was caused by one of the two others is 64%. Statistically, we would say it was caused by an 80% animal. The gemara logic (according to your presentation) would say it was the 90% animal.
Charlie brown said that the 90% animal was standing right there when we arrived on scene. If so, the premise of the rule is changed from my original understanding.