When less means more
Here is the Mind and Matter column in the Wall Street Journal, published on 24th December.
Which American city has more inhabitants: San Antonio or San Diego? More Germans than Americans get the answer right (San Diego). What about Hanover or Bielefeld? More Americans than Germans get the answer right (Hanover). In each case, the foreigners pick the right answer by choosing the city they have heard more about, assuming that it's bigger. The natives know too much and let the excess information get in the way.
This is an example of a "heuristic," a highfalutin name for a "rule of thumb" or "gut feeling." Most business people and physicians privately admit that many of their decisions are based on intuition rather than on detailed cost-benefit analysis. In public, of course, it's different. To stand up in court and say you made a decision based on what your thumb or gut told you is to invite damages. So both business people and doctors go to some lengths to suppress or disguise the role that intuition plays in their work.
Prof. Gerd Gigerenzer, the director of the Max Planck Institute for Human Development in Berlin, thinks that instead they should boast about using heuristics. In articles and books over the past five years, Dr. Gigerenzer has developed the startling claim that intuition makes our decisions not just quicker but better. He rejects the notion that hunches are second best, trading off accuracy for effort to achieve decisions that are "good enough" but not perfect.
As Dr. Gigerenzer sees it, complex problems do not necessarily need complex solutions, and more detailed analysis does not necessarily improve a decision, but often makes it worse. He believes, in effect, that less is more: Extra information distracts you from focusing on the few simple aspects of a problem that matter most.
A baseball player running to catch a fly ball is not behaving, even unconsciously, as if he were solving differential equations to work out where the ball will land. He is following a simple rule: Keep the angle of the falling ball constant in your vision and adjust your running speed accordingly. It's the same trick used by dragonflies catching flies.
When Jeffrey Skiles, the co-pilot of the plane that made an emergency landing in the Hudson River in January 2009, explained how he and his captain decided that an airport landing was impossible, he described the same "gaze heuristic": The angle of the plane's descending glide made the airport appear to rise in their windscreen view—clearly signaling that a landing there was doomed.
The economist Harry Markowitz won the Nobel prize for designing a complex mathematical formula for picking fund managers. Yet when he retired, he himself, like most people, used a simpler heuristic that generally works better: He divided his retirement funds equally among a number of fund managers.
A few years ago, a Michigan hospital saw that doctors, concerned with liability, were sending too many patients with chest pains straight to the coronary-care unit, where they both cost the hospital more and ran higher risks of infection if they were not suffering a heart attack. The hospital introduced a complex logistical model to sift patients more efficiently, but the doctors hated it and went back to defensive decision-making.
As an alternative, Dr. Gigerenzer and his colleagues came up with a "fast-and-frugal" tree that asked the doctors just three sequential yes-no questions about each patient's electrocardiographs and other data. Compared with both the complex logistical model and the defensive status quo, this heuristic helped the doctors to send more patients to the coronary-care unit who belonged there and fewer who did not.
It is no surprise that in the wake of the great financial crisis, financial regulators are beating a path to Dr. Gigerenzer's door. The complex algorithms that gave AAA ratings to debts that should not have passed the smell test demonstrated all too well the futility of knowing too much.
For a video showing Gigerenzer presenting his ideas in a lecture, see here.
Comments (13)
[...] How? Simplify things with heuristic thinking, more commonly know as ‘rule of thumb’. The Rational Optimist, Matt Ridley offers us the blindingly obvious, but often ignored advice at When less means more [...]
Wikipedia says that San Antonio is the 7th largest city and San Diego is the 8th. So who is smarter?
Great article and thanks for the video link!
The US census bureau puts the 2010 population of the City of San Antonio at 1,327,407 and the 2010 population of the City of San Diego at 1,307,402. With this example we also the real downside of too much trust in heuristics in place of actually fact checking data.
http://quickfacts.census.gov/qfd/states/48/4865000.html
http://quickfacts.census.gov/qfd/states/06/0666000.html
Err.. are we not talking about Experience and the value thereof? My 4 year old was rubbish at catching stuff my now 6 year old is quite good. All pilots are taught perspective cues when learning to land, and yup most are rubbish at the beginning....Oh and experienced glider pilots ( who also happen to be airline pilots) are better at judging those cues - different Experience see?
This is to some extent the lesson of 50 years' AI research. Early efforts carried the assumption that all of thinking could be modeled by symbolic sequential reasoning, and that turned out to be wrong. There are subconscious processes that integrate not only surprising amounts but surprising kinds of data -- and which include powerful learning abilities, just as subconscious. Modern cog psych, e.g. Kahneman, concurs.
However, it is a mistake to dismiss conscious ratiocination. To begin with, "fast mode" is incapable of introspection and cannot tell you when your intuitive judgements are likely to be wrong. Conscious sequential thought is capable of wider and deeper analyses, including meta-analysis; the above is an example.
As AI progressed, researchers discovered that their gut intuitions about thinking were wrong, and have gone a long way toward analyzing and recreating the subconscious processes in explicit algorithms -- albeit ones requiring wildly more computational power than was originally believed.
We have two basic mechanisms for evaluating the things around us: reason and pattern matching (sometimes thought of as intuition). Reason has the ability to provide truths on a more solid footing, but there is often not enough information or time for a properly reasoned response. Intuitive (partly subconscious pattern matching) fills the other side of the coin. We may not even consciously know what all went into the decision, but statistically it's usually substantially better than a random choice.
Just a note to the folks playing gotcha with Census population figures. I suspect that the question as to whether San Antonio or San Diego is bigger has to do with the large area known as greater San Antonio or San Diego. It is true that the actual city proper of San Antonio has a slightly higher population than San Diego but the Metropolitan Statistical Area (MSA) of San Diego has more than then 3 million people compared to the MSA of San Antonio with around 2.1 million.
When not in the greater San Diego area, someone from La Presa, CA will tell someone they are from San Diego. Same thing when someone from Converse TX is outside of San Antonio will say they are from San Antonio.
None of this has anything to do with the central point of an interesting insight into human thought. However, I thought some people seemed to be stopping at a simple comparison of city population and then deciding that the whole point of the article was wrong.
In general simple analysis gives better answers as the scenarios can be understood. Also more complex analysis leads to incorrect assumptions becoming embedded as quasi-facts. the DECC calculator on future electricity generation is a classic example. The real driver of decisions is the future price of gas, but that is not a variable, and gas fired generation is inextricably linked to carbon capture, totally changing the economics from reality.
This reminds me of a war game (referred to in the book Blink by Malcolm Gladwell) called Millennium Challenge 2002, which was carried out by the US military and in which LtGen Paul Van Riper of the US Marine Corps successfully used unorthodox, fast-moving tactics in order to outfox the opposition. It is described in this paper (see pages 8-10):
http://www.au.af.mil/au/awc/awcgate/navy/nwc_mccown.pdf
"Analysis of the events leading to the defeat of Blue Team provides a useful look at the effectiveness of intuitive versus analytical decision making in warfare. Although the Red Team staff had methodically studied the situation and their alternatives prior to the outbreak of hostilities, their decision making style shifted once the war began. At that point, LtGen van Riper and his team embraced an intuitive approach, rapidly assessing and solving problems as they occurred. Unable to rely on modern forms of communication, for example, the Red Team utilized older methods to exchange information. These methods, such as couriers and light signals, were slower and less efficient, but were highly effective and impervious to attack. To further limit the effects of the attacks on their communication infrastructure, Red Team headquarters limited their communication in a deliberate effort to create conditions for field commanders to employ their own intuition and initiative. Having clearly articulated his intent and guidance before the war, LtGen van Riper was confident that his subordinates would accomplish their missions, unencumbered by excessive communication with leadership.
The Blue Team, on the other hand, focused on the large quantity of information streaming from their computer systems and depended exclusively on their analytical decision-making processes. As a result, the Blue Team essentially suppressed their intuition, preventing themselves from quickly visualizing and solving the problem. While the Blue Team was slowly attempting to interpret intelligence in order to issue orders to field commanders, the Red Team accomplished neither, yet acted so fast that the Blue Team could not respond."
I wonder if regulations were reduced to, if you screw up you go to jail. Might stop the tendency to find a route around the rules when it takes half a dozen lawyers to figure out what the rules may be.
Gerd Gigerenzer has sent me a note as follows:
PS: If you write again about the recognition heuristic, you might consider using the example "Detroit - Milwaukee" which I used in the first chapter Gut Feelings. You used our original example "San Diego – San Antonio" from a publication more than a dozen years ago. This city pair is one of the rare ones where the population relation has since changed, and San Antonio has now actually the larger population. That makes no difference for the general thesis, because the studies were done when San Diego was still the larger city. But it might help the compulsive reader who checks everything and who is not aware of the timing of the study.
An excellent article. Gary Klein's "Sources of Power" goes into this in quite some depth. It is based on a lot of research relating to fireman, nurses, military commanders etc.. Evidently much of the intuitive thinking comes from years of experience and expertise, some obtained by apprenticeship (which covers formal trade apprenticeships as well as the type of expert pupil relationships found in Phd researchers and other experienced to learner relationships).
Similar blog ideas have been mentioned by David Snowden at www.cognitive-edge.com - he shows that complex human systems need different types of analysis and intervention than simple ones as traditional planning and control mechanisms do not work. He as an amusing story of a parent planning a 16 year old child's birthday party using the typical business school business planning methodologies and post review/learning techniques. Somewhere he uses the design of the Swindon Magic roundabout as a way of probe and respond handling of complex situations. Evidently the roundabout was designed using movable pieces initially and moving and observing. He contrasted this with a photograph (totally made up) of an equivalent intersection containing hundreds of traffic lights, all centrally planned.