というエントリ(原題は「Is history probabilistic?」)の冒頭でDaniel Littleが以下のように書いている

Many of our intuitions about causality are driven by a background assumption of determinism: one cause, one effect, always. But it is evident in many realms -- including especially the social world -- that causation is probabilistic. A cause makes its effects more likely than they would be in the absence of the cause. Exposure to a Zika-infected mosquito makes it more likely that the individual will acquire the illness; but many people exposed to Zika mosquitoes do not develop the illness. Wesley Salmon formulated this idea in terms of the concept of causal relevance: C is causally relevant to O just in case the conditional probability of O given C is different from the probability of O. (Some causes reduce the probability of their outcomes.)

There is much more to say about this model -- chiefly the point that causes rarely exercise their powers in isolation from other factors. So, as J.L. Mackie worked out in The Cement of the Universe: A Study of Causation, we need to be looking for conjunctions of factors that jointly affect the probability of the occurrence of O. Causation is generally conjunctural. But the essential fact remains: no matter how many additional factors we add to the analysis, we are still unlikely to arrive at deterministic causal statements: "whenever ABCDE occurs, O always occurs."


But here is another kind of certainty that also arises in a probabilistic world. When sequences are governed by objective probabilities, we are uncertain about any single outcome. But we can be highly confident that a long series of trials will converge around the underlying probability. In an extended series of throws of a fair pair of dice the frequency of throwing a 7 will converge around 6/36, whereas the frequency of throwing a 12 will converge around 1/36. So we can be confident that the eventual set of outcomes will look like the histogram above.
Can we look at history as a vast series of stochastic events linked by relations of probabilistic causation? And does this permit us to make historical predictions after all?

確率的な因果関係で結ばれた非常に多くの確率論的な出来事として歴史を捉えることは可能だろうか? その場合、歴史に関する予言をすることが可能になるのだろうか?



The Cement of the Universe: A Study of Causation (Clarendon Library of Logic and Philosophy)

The Cement of the Universe: A Study of Causation (Clarendon Library of Logic and Philosophy)