making hard decisions: probability vs determinism
rationalizing vs feeling it out
an interesting lens, recently: probabilistic vs deterministic models, and what we might learn by using them to frame our most complex decisions.
deterministic goal: get the right answer (success is binary)
probabilistic goal: make the best guess (increase your odds for success)
personal updates since last time: spent some time @ the stanford d.school as a university innovation fellow. met some great people from all around the world. learned something new. am I actually tempted by grad school? also, have been learning a thing or two about letting go.
certainty is comfy
There is an Indonesian phrase, “cocoklogi,” that describes how we retrospectively find clues to explain chaos.
We like determined outcomes. We tell stories with happy endings, to explain unexplainable things. We assign value to random things and confirmation-bias ourselves into seeing patterns where there are none.
Beneath the surface, we’re also intuitively aware of the chance that surrounds us — the probabilistic nature of our world. We wish each other good luck and perform superstitious rituals.
The world does not function deterministically. If it did, it would be fair. People who worked hard and did the right things would be rewarded. Things would be predictable.
The world is not a perfect information game — it is a guessing game. Luck is a bigger factor than we’d like to admit. Life isn’t fair. Nassim Nicholas Taleb wrote a whole book on our inability to deal with randomness called The Black Swan.
math tests =/= reality
Our acknowledgement of randomness is, of course, the main difference. Math tests are deterministic - there is an answer key and a correct order of operations that result in a correct outcome. Starting a business (or investing in one) is probabilistic - you’re making bets, and there is no concrete indicator of what might succeed or what will implode.
Other isomorphisms:
chess vs poker
fairness vs luck
quantification vs qualification
schedule structure vs spontaneity
deterministic vs stochastic modeling (monte carlo simulations)
fully observable vs partially observable AI environments
I’ve found probabilistic thinking to be boldly human. Forcing everything into a quantified, deterministic model has the risk of reducing individual meaning into a sum of parts.
That the world is probabilistic is a major problem for most decision-making — automation, investment, marketing, where to eat tonight. Well-constructed frameworks go a long way towards making sense of optionality.
truth holds no quantity
Susan Sontag wrote an essay called Against Interpretation, arguing that the act of interpretation is reductive - that to dig for meaning is to completely ignore it.
We may measure, observe, dissect, reorganize, and quantify that which we still do not understand. The meaning of a thing is metaphysical — more than the sum of its parts. We may understand it through value, or function, or emotion, in languages not spoken by numbers. This is why things that work on paper don’t always work in real life.
We tend to limit our understanding of things by what we can measure about them—quantification bias. Even if our method for quantification is extremely rigorous, we often miss the big picture—the thing happening right under our nose. The elephant in the room.
The problem with soft, qualitative stuff is that it doesn’t fit neatly into boxes. It’s complex, and often contradictory or irrational. To qualify is to rely more on intuition than analysis, and intuition is difficult to systematize.
But the most powerful things are continuous, not discrete. Neither good leadership nor great products nor groundbreaking ideas arise from checklists or recipes. But once they exist, they can be quantified.
So, the only thing we can hope to do is increase opportunity, make thorough decisions, and commit to them. There are two ways to do this: by increasing optionality (breadth) or by increasing commitment (depth). Counterintuitive as it seems, commitment tends to beget opportunity. It gives you direction.
measuring=/= solving
From David Malloy, one of the wiser people I know: people mean what they say, but they don’t always know what they mean.
A major insight from Helm recently: you can’t solve a problem by measuring it. And you can’t sufficiently grasp the complexity of an organization or an individual by simply collecting data.
Leaders and their companies tend to fall victim to the quantification bias, where the only things that matter are those which can be measured.
In the past few months, I’ve talked to countless leaders who have invested dramatically in workplace analytics tools only to see the same results (but in higher resolution). The best leaders listen—but they also seem to fly by the seat of their pants.
Up until January, we had been selling workplace analytics. It answered a clear question from people leaders: people are quitting - can you help us figure out why? It wasn’t until a couple months ago that we realized the question behind the question: we already know the problem - can you tell us what to do about it?
So now, we help employees start connect meaningfully with coworkers. Think new hires, mentors, collaborators, work friends.
It turns out that connection makes people feel engaged. All they need is an excuse to start a conversation.
humans are not rational.
All of this comes down to the fact that rationality and fairness are opposing forces.
Last week at Stanford, I sat in on a class about negotiation. There’s a classic negotiation game where an offeror and a receiver have to agree on how to split $100 of free money. If they agree, both receive money. If they don’t, no one receives money.
The most diplomatic way to do it, of course, it to split 50/50. Rational receivers would even take a 99/1 split, because they’re still getting free money ($1). But of course, when offered a 99/1 split, the receiver says no because they do not perceive it as fair.
When dealing with people, emotion is truer than reason. With anything complex, you’re playing a probabilistic game, not a deterministic one. Your guess is as good as mine.
there’s a bunch of exciting stuff on the helm + etho end that I’ll share soon. in startups, everything is coming together and falling apart at the same time (that’s from twitter, but I forget who).
also - still working on sxsw + stanford update! but frankly, these are probably more exciting for me than for you, dear reader.