Sciences That are Less Wrong

NYT article  by George Johnson.

Namely, lots of results that are published are not tested by others. In physics, there is a shaming factor if you get things wrong, but also the numbers you get are part of a network of other numbers, and so if you get it wrong you will cause trouble. It's not just the effect you are measuring, but the amount to several significant figures. In many fields, natural and social sciences, you get about 1.5 significant figures at best. (Where on significant figure is in base 10, so 3 or 4 or 5 are reliable, but 3.4 does not have much information on the 0.4.)

In modern analytic philosophy, if you argument has holes they are revealed by others who tear you apart. Or, if they disagree they show how you are mistaken. But the arguments here are rarely grand ones, rather they are part of a tissue of arguments.

In literary and historical studies, your problem is how well you use your texts and sources, how far you go from what everyone thinks is there.