> Afaik, he's completely discredited among historians, and
> always was. He basically starts with a thesis and then
> reaches into history to find things that illustrate it. He
> had less than no interest in whether his facts were
> representative; when confronted with evidence that they
> weren't, he actually defended his tendentious cherry picking
> as some kind of new method.
The problem with this argument is that it holds Hofstadter to unrealistic standards that virtually no science, let alone history, can possibly achieve. History suffers from what is known as the too-many-variables-too-few cases problem. That means that there are many factors (variables) that may possibly influence events, and to empirically test those possible influences one needs several observations per variable, at the minimumn two (conditon present vs. condition absent). If you have two variables, you need at least four observations, or twice as many when you want to test for interaction effects - see the diagram below (each sign + or - represents a case in which a condition on one variable is controlled for conditions on the second variable)::
VAR1 VAR2
+ +
- -
- +
+ -
In social sciences that concentrate on individuals this problem can be easily resolved, because it is not that difficult to find enough individuals representing different combinations of the hypothesized conditions. But in history it borders on impossibility, because history rarely occurs, as it were, twice - either as tragedy or as farce. Finding a single counterfactual to any hypothesized effect is a challenge - for example we cannot test the effects of Communism or Russia, because there is no alternative "ceteris paribus" history without Communism in that country against which the claim can be tested.
For these reasons, history must necessarily be selective - it start with the formulation of a theseis - as any other science does, as bottom up empirical generalizations aka "data dredging" seldom yields anything of substance - and then tries to find cases that either confirm or falsify that hypothesis. So if someone accuses a historian with starting with a "pre-conceived" thesis - I usually dismiss such accusations as cheap shots from the hip, if not politically motivated hatchet jobs. A valid form of criticism would be to show a counterfactual to the claim ie. a condition where the hypothesized cause led to a different than the hypothesized outcome, or at least a reasonable probablity of such an outcome.
For example, to put the claim of the horrible effects of Communism on Russia to a test, we can compare its effect in other locations where counterfactuals are easier to find, e.g. in the Central America and the Carribeans and show that Cuba (Communism present) did better than a somewhat similar island nations without Communism (e.g. Haiti or the Dominican Republic). Albeit this is not a definitive proof, it does show that the claims of negative effects of Comminism are likely to be false. Jared Diamond uses this method quite effectively in his books _Guns, Germs and Steel_ and _Collapse_.
I think that some of Hofstadter's theses, e.g. the effect of the US-specific evenagelical religiosity on anti-intelelctualism can be put to a similar test by comparing US religiosity and European religiosity which share the same theological roots, but developed under different social conditions. By this method, it can be demonstrated that same or similar religious faiths can have very different effects, depending on their institutional settings. The more instituional x-tian churches in Europe did not lead to anti-intellectualism (the opposite is actually true), whereas the populist anti-institutionalist variety that developed in the US did. This shows that Hofsdater thesis is at least plausible. To disprove it, one would need to show that populist anti-instituional evangelical religiosity produced pro-intellectualist attitudes - which I understand would be quite a challenge yet to be achieved.
Wojtek