My point was that "A therefore B" isn't useful if the noise level is so high that it can't be used to make a meaningful decision.
Otherwise the advice reduces to "don't trust long explanations". But if 1% of long explanations are valid and 2% of short ones are valid, then the statement is both true, and worthless as a way to detect validity.
Otherwise the advice reduces to "don't trust long explanations". But if 1% of long explanations are valid and 2% of short ones are valid, then the statement is both true, and worthless as a way to detect validity.