Skill, luck and project controls

Patrick Weaver
November 3, 2015

In ‘On probability, randomness and risk‘ we looked at probability. Randomness is a key ingredient in probability, contributes to luck, affects statistics, and can easily be confused for skill or competence.

Luck and randomness are interlinked, and to a very large extent what we call ‘luck’ is the manifestation of randomness. Some babies are lucky enough to be born in stable economies such as Australia; others are born in war-torn places like Syria. Growing up well-adjusted and well-educated is heavily influenced by the family, and economic circumstances, you are lucky enough to be born in; which in turn affects your earning ability and lifestyle as an adult.

However, some people (mostly men) and many observers tend to assume personal credit for outcomes that are largely driven by luck. Some like Donald Trump seem to genuinely believe their wealth is solely due to their skill and hard work (rather than being born the son of a wealthy New York real-estate developer) and look down on less successful people as failures.

In contrast, Prime Minister Malcolm Turnbull’s comments that many taxi drivers work a lot harder than he ever has suggests an appreciation of randomness; but then again, it may simply be the difference between only having $200 million compared to Trump’s $4 billion! Empathetic leaders and people who appreciate randomness keep in mind the proverb: ‘There but for the grace of God go I’. The simple fact is, we cannot control randomness but we can sometimes influence the consequences.

Unfortunately this fact is generally ignored. Both human nature and statistics tend to encourage seeing patterns where none exist and then assigning meaning to randomness! The scruffy beggar may deserve his fate but equally likely he is in his current situation as a consequence of a series of random events (aka ‘bad luck’). So precisely what is ‘randomness’?

Understanding true randomness

As a starting point, there is a difference between a process being random, and the product of that process appearing to be random. A random selection of music tracks will, from time to time, play the same track from the same artist back-to-back two, occasionally three or more, times; as Apple found out in its early iPod music players. True randomness will include repetition. Apple solved the problem by making the track selection process slightly less random, so the results appeared to be more random!

The fact is, we have a very poor concept of randomness, cannot recognise it when we see it and cannot produce it when we try (it took scientists years of work to generate a list of truly random numbers). We misjudge the role of chance and by seeking reason, and as a consequence make judgement calls that are not in our best interests. For example, how would you rate a funds manager who outperformed the stock exchange index for more than a decade? Was he good or just lucky?

The prime example is stock trader Bill Miller, who outperformed the Dow Jones index for 15 years from 1991 to 2005 (the longest winning streak in history). The odds of Miller in particular outperforming the index in any one year is 1:2 – he either does, or does not. Therefore the odds after two years are 1:2 x 1:2 = 1:4. After 15 years 1:32768, a minimal probability as the writers for ‘The Consilient Observer’ newsletter published by Credit Suisse pointed out and numerous awarding bodies agreed. The newsletter stated no other funds manager had achieved a similar result in the past 40 years. Was he good or just lucky?

To distinguish skill from random luck, you need to answer a different question, what is the probability of any one of the approximately 6,000 funds managers investing in the market having a run of 15 consecutive winning years at some time in the last 40 years? The answer to this question is about 3:4—it is highly unlikely that no one would have achieved the ‘winning streak’ simply based on random chance.

What the statistics cannot tell you is if Miller’s run was pure chance, pure skill or some combination; there were a number of 12-month periods in the 15 years where he lost money, luckily for Miller, none of these periods ended on the 31st December of any year. Ascribing winning streaks to skill rather than randomness is called the ‘hot hand’ fallacy – most of the research has been done in the sporting arena. And you cannot discount skill entirely.

Skill plays an important part in determining winning and losing streaks; a good team will win on average win more often than a poor team and therefore their winning streaks will tend be longer and losing streaks shorter than the poorer team on average. But on average nothing is average so long losing streaks by the ‘good team’ are still possible due to the effect of randomness.

We encounter winning and losing streaks in all aspects of life, as well as other ‘patterns’ that are caused by randomness. Given the thousands of project managers delivering projects year-in, year-out, it is highly likely that at least one or two will have an outstanding run of successes due to randomness rather than skill (and vice versa). And because of randomness, a run of good luck will not necessarily follow a run of bad luck! On average, for most people good and bad luck balance out but in a random world, on average nothing is average.

The illusion of control

The desire to see patterns where none exist is driven by our innate need for control. We cannot control randomness; a pattern means we can derive meaning and make decisions based on this interpretation of events, even if the decision and actions are pointless.

People may pay lip service to the concept of chance but tend act as if they have control. CEOs, coaches and project managers get praised for winning streaks and fired for losing streaks; but the individuals rarely have sufficient control over the myriad of influences affecting the performance of their organisation to warrant either. Their leadership is an important factor in the success or failure but only one of many.

This illusion of control is reinforced if the person has developed plans and strategies and worked hard to achieve the planned outcome. The illusion is reinforced by confirmation bias where we look for and value ‘facts’ that support our view and reject or ignore contravening evidence, and interpret ambiguous evidence to support our preconception—the illusion of control over randomness is hard to shake.

However, in all but the smallest of endeavours, complexity and emergent behaviours driven by random events can only be influenced by the actors involved; the underlying randomness is the most likely determinant of an outcome. Plans and strategies are useful guides to the outcome we desire, but they do not predict, and certainly cannot ‘control’, the future.

Achieving the plan requires adaptive, agile behaviour from the team and their leader to deal with each emergent situation, and a good slice of luck as to how the randomness plays out. Maybe when employing project managers, executives should ask Napoleon Bonaparte’s key question; “I know he’s a good general, but is he lucky?”

Randomness causes a strong asymmetry between the past and the future, looking back it is always possible to ‘join-the-dots’ and explain why a series of random events caused a defined outcome. But even knowing the causes of the current circumstances, it is much more difficult to predict which of the many thousands of possible random events in the future will interact to cause a future catastrophe, or facilitate the desired effect or outcome.

For example, weather forecasters can always explain what has happened; they have a far more difficult time predicting what will occur, particularly at the detailed level of your suburb, and the difficulty (and associated error) increases exponentially the further ahead they try to look. Project controls have exactly the same problem!

Sociologist Charles Perrow’s ‘normal accident’ theory put the problem succinctly: “In complex systems we should expect that minor factors we can usually ignore will by chance sometimes cause major incidents.” The theory is usually applied with reference to a disaster, but it can equally apply to unexpected successes.

While we cannot control randomness, we can influence outcomes:

  • First by minimising the effect of innate biases such as confirmation bias and control bias; look at the data without trying to introduce patterns: correlation does not mean causation, and the past does not control the future.
  • Second, by being adaptive and agile in dealing with emerging situations. To paraphrase Eisenhower, planning is vital, but the plan is of little value (as at a minimum adaptation is required); intuition and reflection are needed to know how much to change in a plan and what to keep as the future unfolds.
  • Finally by being persistent and resilient! Randomness cuts both ways and the best way to create success is to keep trying, learn from the failures and sooner or later (if you are lucky) a successful streak will emerge. As Thomas Edison once said: “I have not failed. I’ve just found 10,000 ways that won’t work.”

A final thought, given the very significant influence of randomness in life: remember many of the wealthy and successful celebrities owe much to their ‘lucky break’ and many of the unsuccessful in this world owe much to their ‘unlucky break’–in other words, to randomness. The same applies to project managers.

This is the second of three articles based on Leonard Mlodinow’s book, The Drunkard’s Walk.

Author avatar
Patrick Weaver
Patrick Weaver is the managing director of Mosaic Project Services and the business manager of Stakeholder Management Pty Ltd. He has been a member of both PMI and AIPM since 1986 and is a member of the Asia Pacific Forum of the Chartered Institute of Building. In addition to his work on ISO 21500, he has contributed to a range of standards developments with PMI, CIOB and AIPM.
Read more