Perhaps I should go netless more often. Reading a lot of material in a short time in order to catch up, rather than in drips and drabs as it is produced, allows cross blog and media discussions to be more easily followed as a thought thread. Maybe. It isn't that this is difficult in any case, it just has greater impact when done in one swell foop.

There's been a focus on bias and delusion lately. It isn't that this wasn't an important topic all along, but since the Overcoming Bias boys have assembled as a crew and launched their blog there has been increased discussion. Old material as well as new has been brought to bear on selected topics.

This recent Arnold Kling post, Political Beliefs and Self-Deception, includes a couple of issues that have interested me for some time: expertise and political failure. The post is mainly a pointer to, and quote pulled from, this oldish paper by Tyler Cowen.

I consider models of political failure based on self-deception. Individuals discard free information when that information damages their self-image and thus lowers their utility. More specifically, individuals prefer to feel good about their previously chosen affiliations and shape their worldviews accordingly. This model helps explain the relative robustness of political failure in light of extensive free information, and it helps to explain the rarity of truth-seeking behavior in political debate. . . I also consider political parties as institutions and whether political errors cancel in the aggregate. I find that political failure based on self-deception is very difficult to eliminate.

This seems almost trivially true, almost common knowledge, but it's free information usually discarded. Politics isn't about truth, it's about power.

By self-deception I mean individual behavior that disregards, throws out, or reinterprets freely available information. Individuals frequently treat their personal values as a kind of ideal point, and assume that the pursuit of those values also yield the best practical outcome. For instance, religious groups who reject parts of modern medicine (e.g., blood transfusions) might also believe that those treatments are not very effective in medical terms. Similarly, people often interpret "information issues" as "value issues," or underweight the relevance of information for the issue at hand. To use the terminology of Feigenbaum and Levy (1996), individuals have preferences over beliefs rather than being pure truth-seekers. If we put the argument in Bayesian terms we can think of each individual as having a prior. Individuals welcome confirming evidence for the prior but they throw out disconfirming evidence.

The paper continues in like vein, with examples to flesh out assertions. What interests and mystifies me is that we do this silly stuff. Reading a large dollop of dark-sider stuff after my hiatus - yeah, I read them too, if only to gross myself out and keep one semi-shielded eye on the sordid underside of reality - was something like a visit to an insane asylum. Politics is stupid, but also unsane. It isn't just that "Individuals welcome confirming evidence for the prior but they throw out disconfirming evidence", political groups institutionalize this behavior and have made a "science" of doing it "well". The unconscious and semi-conscious behavior of individuals is exploited in full consciousness by dark-siders.

There is no hope that humanity will go sane. That's not all bad since it is the source of goodness and beauty as well as political failure and other badness. This makes me appreciate political systems designed explicitly to blunt, delay and otherwise thwart political impulses. The less efficient and timely a political system is, the better the results. It can damp out high amplitude excursions from sanity using equal but opposite excursions displaced a bit in time, like out of phase sound waves that cancel one another. Another palliative for political failure is subsidiarity. By keeping the scope of failure small and local results will vary across regions, and moderate one another - a different sort of analogous phase shift damping effect. It's not always effective - sometimes the waves are in phase and so amplify rather than moderate insanity. Sometimes the magic doesn't work.