Follow by Email

Friday, June 30, 2017

Stumling and Mumbling: SELECTING FOR GROUPTHINK

SELECTING FOR GROUPTHINK

Stephen Buranyi says the scam that is academic publishing "actually holds back scientific progress":

Given a choice of projects, a scientist will almost always reject both the prosaic work of confirming or disproving past studies, and the decades-long pursuit of a risky "moonshot", in favour of a middle ground: a topic that is popular with editors and likely to yield regular publications.

The FCA says that active fund managers "did not outperform their own benchmarks after fees."

These two observations are related. They show us that selection mechanisms – peer review, hiring fund managers and the funds market – don't necessarily select for the best. A new paper (pdf)by George Akerlof and Pascal Michaillat discusses one way in which this can happen.

They start from some experiments with flour beetles in the 1950s and 60s. These found that when two different species were placed into jars of flour, it was not the case that the most biologically fit species came to dominate. Instead, sometimes one species did and sometimes the other. The reason for this was because the species were more likely to eat the eggs of the other species than those of their own. This meant that when one species increased relative to the other – perhaps for arbitrary reasons – it continued to increase still further. Such an "egg-eating bias", say Akerlof and Michaillat, means that in science unfit paradigms might prevail over fitter ones - as Buranyi claims.

The egg-eating bias takes the form of professors or journal editors preferring candidates or papers in their own image – ones that work in their paradigm. This is sometimes because of simple favouritism. But it can also be simply because it's easier to evaluate someone's work if it is like your own.

This fits with the claims in economics that bad paradigms – suchas (pdf) DSGE (pdf) or CAPM – have prevailed despite their empirical flaws.

It also fits with the poor performance of fund managers. One reason for this is that older fund managers hire and train younger ones on the basis of judgment-based stock selection rather than their ability to exploit proven means of beating the market such (pdf) as defensive or momentum investing. And as Bjorn-Christopher Witte shows, market forces do not necessarily weed out bad managers and favour good: markets are imperfect and sometimes perhaps even counter-productive selection mechanisms.

It also, of course, has implications for corporate management. It's consistent with work by Dan Bernhardt, Eric Hughson and Edward Kutsoati who show that because bosses favour underlings in their own image, firms can become increasingly inefficient: for example, as finance-types exclude engineers. In the same vein, Eric Van den Steen has described (pdf) how "organizations have an innate tendency to develop homogeneous beliefs". This is because like hires like, and then people learn from those similar to themselves.

You can also, of course, tell a story about gender bias along these lines. (And, of course, about the media).

The point here is simple. Institutions – be they markets, firms, universities, publishers or whatever – are (among other things) selection mechanisms. We should not assume that such mechanisms work perfectly to optimize efficiency or truth. We must look under the bonnet to ask how exactly they work, rather than tell ourselves just-so stories.

In particular, institutions select – perhaps not entirely intentionally – against cognitive diversity and for groupthink. But as John Stuart Mill warned, this can be "a social tyranny more formidable than many kinds of political oppression" which can end up "enslaving the soul itself."


--
John Case
Harpers Ferry, WV

The Winners and Losers Radio Show
7-9 AM Weekdays, The Enlighten Radio Player Stream, 
Sign UP HERE to get the Weekly Program Notes.