Без заголовка

Но вот хинты, что там делать с вероятностями — есть. Они занменяются избыточностью: If, in conventional, probabilistic, information theory, S is a discrete set and each instance of S has a probability p<1/2 of having been changed from the correct value x by uncorrelated perturbations, the probability that a plurality of the values stored in n instances will not instantiate x falls exponentially with n. Although probabilities are not allowed at the fundamental level in constructor theory, redundancy can still play effectively the same role, in following way. Denote by S(∞) an unlimited supply of instances of S. Suppose that they all have the same intrinsic, preparable attribute, either x or y, and let us call a sequence of experiments on S(n), as n increases without limit, an ‘experiment on an ensemble S(∞)’. Recall from §2 that a possible task is one for which the laws of nature impose no limit, short of perfection, on how well the task can be performed. Because our assumption VI implies that there is no limit on n, it follows that for any two intrinsic attributes x and y of a substrate S, x(∞) and y(∞) are either interchangeable in all experiments on S(∞) (using generic resources), or distinguishable. For given such a supply, performing all possible experiments on (instances of) S, infinitely often, is a possible task. So if that cannot tell the difference between x(∞) and y(∞), nothing (that uses only generic resources) can. If something can, we call x and y ensemble distinguishable. Thus, we propose the principle: IX. Any two disjoint, intrinsic attributes are ensemble distinguishable. Это из https://royalsocietypublishing.org/doi/10.1098/rspa.2014.0540 Разбираться нужно внимательно, все ж "на поверхности лежащие возражения" там наверняка разобраны в деталях!