8: Luck vs Skill
Even in the present day, there are still many superstitions about "luck" having a greater influence than "skill" on the outcome of a given plan even by seemingly rational individuals.
A more rational perspective is that "skill" is taken to mean that things went according to plan - it is the predictable outcome based on facts that are known and theories that are accurate. "Luck" is used when the facts were unknown and the theories were wrong. (EN: When sued by a critic, "luck" often identifies his own ignorance or inability to comprehend.)
Outcomes may seem mystical because we lack the ability to know the future - a careful analysis can often find a rational explanation and an obvious oversight or mistake that was not seen beforehand.
Regression Toward the Mean
The theory of regression toward the mean (Francis Galton) maintains that over time, natural phenomena seek to achieve a standardized outcome, which is often misinterpreted to suggest that after a better outcome is achieved, a worse must follow to restore balance. In particular, Galton is known for estimations of intelligence: the child of dullards will be smarter than his parents and the child of geniuses will be duller. (ENL: Given what was just stated in the last chapter, this is likely an attempt to mash nature into a bell curve because it makes it easier to analyze and explain.)
Galton looked to sweet peas as an example: comparing the size of peas to the seed from which the plant was grown, he suggested that the child peas of large peas were still large, but a little smaller than the parent plant, and vice versa. He failed to consider other factors that might explain the differences (which plant pollinated the parent, how much water it was given, what the temperature was during growing season, etc.)
One significant insight Galton had was that, even though the principle of reversion seems to occur from one generation to the next, the overall distribution of sizes in the population remains stable over time: they are not becoming standardized as a group.
If you ignore the concept of the reversion to the mean, you are prone to making mistakes: you may assume that the great results will be carried forward in equal degree, or even magnified, in the future.
(EN: I don't entirely buy this theory, and have a sense that it is based on a limited sample size and an arbitrary characteristic. Given the complexity of genetics, I it seems overly simplistic and, scanning a head, the author misapplies it outside the field in which it was observed.)
Skill, Luck, and Outcomes
It's generally understood that in any human endeavor, the outcome is a combination of luck and skill. Even laboratory experiments designed to eliminate luck and games of chance designed to eliminate skill are never purely controlled or random.
There are many naturally occurring variables that are beyond our ability to control - especially when it comes to human performance. Try as he might, an archer cannot consistently split the first arrow he shot with the second - though he is using the same equipment, the same stance, on the same range, etc. it is a fluke if that ever occurs. While the rational mind expects the exact same circumstances will produce the exact same outcomes, we must admit that circumstances are never exactly the same.
A basic logic tree is implied: poor skill and bad luck yields disaster, good skill and good luck yields great success - but success or failure are determined by some interplay between skill and luck: a skilled person with bad luck or an unskilled person with good luck may come out on either side of failure and success, but not by as great a degree.
Back to the financial markets for an example: it is often assumed that financial analysts can apply their skill to predict the performance of investments. But this is overlooking the fact that the performance of a business is based on many factors that they cannot possibly observe in sufficient detail. As such, "luck" foils even the most meticulous analyst. Ironically, firms hire and fire analysts because of the luck they have had in the markets, not in the skill that went into their predictions.
Individual investors behave in a similar manner: they flock to stocks that have had recent success and run from them when there is a failure - but because success and failure are only recognized after the fact, they often end up buying high and selling low.
Digression from the Mean
One economist (Secrist) misinterpreted and misapplied the theory of reversion toward the mean to conclude that mediocrity tends to prevail in competitive markets - but this ignores the second part of the original theory: motions toward average are noted from one generation to the next, but not as an ongoing trend.
What we find in highly competitive markets is not standardization, but differentiation: companies pursue different goals and seek to serve different target markets, each following a strategy of success. It is only when markets are stagnant that good become commoditized - and periods of stagnation tend to be short lived before one firm seeks to gain competitive advantage by pulling away from its rivals and doing something differently.
Swinging this back to luck, results fluctuate over time, and what's found that extreme results in any given period will make the results in the following one seem less extreme by comparison. And this is where the notion of mean regression finds its mathematical proof, even in a situation wehre the diversity of the population as a whole does not vary over longer periods of time.
What Kind of Feedback Helps Performance?
The author suggests that regression toward the mean may be a phenomenon that abusive authority figures use to justify tongue-lashing people into performing better.
In effect, a team that has a good game is likely to have a worse (slightly less good) one the net time - the fact that the coach congratulates them does not cause their performance to be weaker. Conversely, if a team has a bad game it is more likely to have a better (less bad) one the next time - and if the coach browbeats them for poor performance, that is coincidental.
Even if you set aside regression to the norm, this goes back to coincidence versus causality: complements or criticisms do not "make" people perform better or worse - tips and guidance for performing better might, but the tone in which the information is delivered has no logical connection.
The Halo Effect
The notion of the halo effect goes back to the 1920s (Thorndike), when it was observed that military offers tended to give better marks in regard to specific criteria (intelligence, physique, leadership, etc.) to subordinates of whom they had a good general impression, and worse marks to those of whom they had a lower impression.
Essentially, this means that scoring systems often suffer from the opposite effect that they were intended: instead of summing up the ratings in specific objective categories to come up with an overall score, raters begin with a general impression that compromises objectivity in measuring specific characteristics.
It's also observed that a positive impression in some regards leads us to forgive flaws in others: this is the reason that the CEO of a company that performs well financially is forgiven for his breaches of ethics, or a popular politician retains the support of voters in spite of disgraceful behavior in his personal life.
In regard to reputation, regression is often seen in media coverage: a successful individual or firm is a media darling for a short amount of time, then the muckrakers turn up to knock them back down. This tendency has become so common that it is a reliable indicator. Studies done of Forbes and Fortune magazines show that companies that receive good press for their performance in the previous two years will then underperform for the year afterward ... and conversely, the stock of companies that are roughly criticized by the financial press perform consistency better in the following two years. Sports fans have a variant of this, called the "Sports Illustrated jinx," in which team and athletes are believed to perform worse after appearing on the magazine's cover. Statistically, this bears out.
Thus considered, the author suggests that the halo effect and mean reversion is a "one-two punch" you will see everywhere.
Suggestions
Estimate the mix of skill and luck involved in any system. Consider what percentage of the variables are under your control, and which are estimates based on probability - acknowledging there may be variables you are entirely unaware of. A simple test the author suggests is asking if you can lose on purpose - if so, then skill is involved. It's suggested that it's just as hard for people to build a portfolio that will perform worse than the S&P 500 as it is to build one that will perform better.
The author provides a list of games to get the reader thinking:
- Skill Games: Checkers, Chess, Go
- Luck Games: Roulette, Slots , Craps
- Mixed Games: Poker, Monopoly, Backgammon
Sample size is also significant: the probabilities involved in predicting an outcome between two competitors are difficult to calculate, and the more competitors there are in the market (or the tournament) the more likely it is that any competitor will end up winning, even one that does not seem very strong at the onset. This is particularly true in competitions where chance is involved - one handicapper suggested that the worst team in baseball could beat the best team about 15% of the time, but this is not at all true of chess.
Beware of superstitions. The belief that a "hot streak" will continue is an easy trap to fall into in gambling or in business - and even more alluring in business because it is assumed that all the variables are known and controlled. If you toss a coin three times and it comes up heads-heads-heads, some people will suggest it will come up heads again, others will suggest it's about time for tails. It's still a 50/50 chance every single time.
Be aware that earlier experiences will shape future perceptions, which is the reason it's very important to get things right the first time you serve a new customer: a good impression will enable them to overlook some minor flaws the next time; a bad impression will cause them to be more critical, if they come back at all.
Be attentive to changes that take place between iterations. A common cause of disaster is inattentiveness, in assuming that "this time" is exactly like every time before and the outcome will be the same. When any element changes (your firm, the playing field, the competition) then the same actions may not get the same results.
Beware of swindlers. There are no shortage of schemes that profess to offer a systematic way to win at gambling, based on a great deal of information that seems mathematically sound and philosophically reasonable. The very same can be said of schemes that propose to get selling, marketing, investing, and other business functions down to an exact science and offer a scheme that will guarantee you will win consistently.
Beware of your own ego. People like to think they have the ability to quantify and control everything, and that their success is because of skill rather than luck. Being realistic about the degree to which factors beyond your control influence outcomes will help to focus your thoughts on what really matters and to be better prepared for unusual, but not unexpected, situations and outcomes.