Big Success Requires Failure, High Output Requires Not Focusing On It
As you all know, the tech ecosystem has its own rules. Among them, is a strong emphasis given to execution. More precisely than just execution, the primacy is given to output. Output over anything. Even when one talks about people, one has a strong bias toward it. Look at the tech entrepreneurs and managers Bible: High Output Management, by Andy Grove, the former CEO of Intel. Quite a revealing title, isn’t it? Even when you talk about leadership and management, you end up focusing on… output.
Focusing on output is a great way to ensure high performance and productivity (almost a tautology). Yet it requires to build a very important firewall: the praise of the absolute contrary, the praise of a bad output, the praise of failure.The system works quite well, however contradictory it might appear for an outsider. In fact, the reason why failure is praised is not fully understood. This value seems to have appeared only by chance. But it’s not. It’s the needed aspect for a coherent value system.
And to show it, we will have recourse to probability. As you will se, probability helps you have a better understanding of the links between risk and rewards (and the beliefs / values associated) and teaches you to beware of not taking output as the whole information.
Failure, Innovation & Expected Value
Let’s begin by the most acknowledged part, the praise of failure. Let’s get straight to the point: accepting failure is a condition for innovation. Why, you ask?
Innovation is a try to leap into a better, bigger, more efficient state of affairs. Translated into probability, it is an increased expected value than what a linear projection of the present should be. But as you know, expected value is the multiplication of payoffs with attached probabilities. And the very nature of innovation implies lower probabilities of success than a linear projection of what exists today. You might aim at a very big goal (the output), the attached probabilities are low. And the bigger the payoff, the lower the probability, all things being equal [this could be a subject of discussion in its own, because it is not that evident that the relationship is linear]. Put another way, in the world of innovation, since the low(er) probability is a given, you need to accept the risk of failure, otherwise people won’t try to reach the high(er) output required to have a big enough output in the end. Expected value teaches us that the value system, to be working and coherent, need to both praise output (aim at a big payoff) AND failure (statistically you will fail). This is the reason why these values are so much entrenched in the global tech culture.
Right Or Wrong, The Spectator Cannot tell
Let’s now turn to more controversial subjects, keeping our probabilistic approach. Even if we slowly learnt to accept -even to celebrate- failure in the tech world, we are obsessed by the champions, the big winners.
What they say becomes Truth to follow. Think “Move Fast and Break Things” (that has long been Facebook mantra). The words of champions become the common language. Think “0 to 1” or “Unfair Advantage”. But doing so, we forget that not only we tend to lose some intelligibility and sound phony, but more important we forget that a specific form answer to a specific set of constraints and goals. Said differently, generic solutions are seldom good, effective ideas. What works incredibly well somewhere might be a disaster elsewhere -even at the same place but at another time. That’s the case for any aspect of business.
What does it have to do with probability? Not much so far. But here come the worst part: these idols might be complete frauds (at least on some subjects they take a stance). And some apparent losers might be total genius and be perfectly right. “Why shouldn’t I hear X that didn’t achieve anything grandiose when I can hear the advice of Y that built a unicorn?”.
Here comes probability, and the highlight of a common mistake to avoid.
Let’s use an example. Let’s assume that there is 95% of chance that it will rain tomorrow. Having that in mind, you should take an umbrella, for sure. Yet the day after, equiped with your umbrella, not a single drop of rain. Just a warm, sunny day (which is a pretty good sanction). Your natural tendency, outside looking for a pleasant terrace for a drink, is to think to yourself: “the predictions were mistaken, it didn’t rain eventually”. Well, you might have read that it will rain and it did not — a typical bias called attribute substitution. In fact they said that there was a 95% chance of rain, which implied that there is still 5% chance that it won’t rain. And not only the probability might be right, it was also totally rational and rightful to take the umbrella.
Let’s carry on with this story. You have taken your umbrella during this sunny day and you meet a friend of yours that didn’t carry one. You start to discuss and one of you might say that he was right not to take an umbrella. This of course, is not true. Knowing that there was a 95% chance of rain, the right decision was to take an umbrella. By just focusing on the output, we may misinterpret the value of the decision taken.
Right Or Wrong? The Effects of Luck & Regression To the Mean
This is the curse of business authors and journalists when they talk about CEOs. When the company is striving, they cannot emphasis enough how incredible its leader is. Yet, a couple of months or years after, as the facetious phenomenon that we call regression to the mean (a variable on its first measurement that is extreme will tend to be closer to the average on the following measurements), the company that were having incredible results is starting to be less impressive, if not in a bad position. Overtime, this is almost always the case: a great company will become good or average, if not bad. And the visionary CEO with outstanding decision-making & leadership skills will start to appear less and less exceptional.
We should distinguish three things about that particular example to avoid conveying and remembering the wrong messages: (1) there are some incredible CEOs, teams and companies, that manage to overperform the market in the long run (regression to the mean is more a tendency than an absolute law), (2) regression to the mean is not fully responsible for what we described: the outcome is not directly correlated to the decision (there are external factors), (3) following our initial point, the issue with the interpretation of pundits is that they don’t have the complete picture: they don’t have access to the full decision-making process so they cannot infer any value of a given decision (a good decision might end up catastrophic, a bad decision might end up very well).
Of course, it would be rather dubious to build an empire by chance. It takes time, relies on many decisions. And the longer, the more recurrence there is, the better you can infer that talent is involved and chance only represents a part of the equation. A good metaphor is poker. A poker champion might lose in a given tournament against a less skilled and disciplined player, but the lucky one won’t be able to win consistently, since luck might be just like common sense, the most distributed thing in the world.
You all heard stories of students asking what’s the point of learning math for their daily life? Here is an illustration.
Probability helps you have a better understanding of the links between risk and rewards (and the beliefs / values associated) and teaches you to beware of not taking output as the whole information. Time and repeatability are the only shields against being fooled by randomness. What you see is not always what it is.