Alchemy: The Surprising Power of Ideas that Don’t Make Sense

The Jolly Contrarian’s book review service™ Go Premium Index: Click ᐅ to expand: Book reviews Comments? Questions? Suggestions? Requests? Insults? We’d love to 📧 hear from you. Sign up for our newsletter.

Rory Sutherland’s ten rules, and how these translate into the JC’s messed up, post-structuralist view. “Signalling” — attending to a communication that is proportional to the cost of generating or transmitting it. The sunk cost of expensive behaviour signals your commitment. This is somewhat redolent of Robert Cialdini’s reciprocity.

“Subconscious-hacking” — framing circumstances to provoke a different outcome. Rebrand “graduate tax” as “deferred university tuition fees”.

“Satisficing”— understanding that in times of uncertainty, people care more about variance between best and worst outcomes, and not just reaching the outcome. This is power of brands.

“Psycho-physics” — the difference between perception and reality. How what we see, hear, taste and feel differs from ‘objective’ reality.

The opposite of one good idea can be another good idea

The idea that there even is a single right answer, let alone that you know it, hails from a profoundly deterministic, reductionist world view. If you subscribe to this view, and you believe you have the right answer, then any other answer is necessarily sub-optimal, therefore wrong, and therefore you are objectively justified in suppressing it. The benign view (which Sutherland takes) is the “no-one got fired for hiring IBM” approach: I took the correct, rational path, I was objective, so I can’t be blamed should things go wrong.

But isn’t that a depressing, negative, glass-almost-empty disposition to take to your work? “We are but actors, all the world’s a stage, we are but frozen in the starlight and determined by events; we cannot influence outcomes, so our dearest aspiration is not to be blamed”?

Especially since, if you adopt this view, no-one can be blamed for anything, anyway, since on a deterministic reading the outcome of the universe in every particular was set in stone from the original singularity?

As Sutherland says, “that’s wonderful if you want to keep your job; if you want to have an original idea it’s potentially disastrous.”

Is it not more rewarding to think that not only can you influence outcomes, but that this is your sacred quest? Chatbots cannot do this, folks. This is your spidey-sense.

Don’t design for average

Models which aggregate individuals into some kind of archetypal mean create a dead end. This is of a piece with Matthew Syed’s observation about the fighter cockpit designed for the average pilot, which turned out to fit no one.

Firstly, the average may fit no one: the spatial average of all the positions on a sphere is nowhere near any of them.

Secondly, the proxy for the average, the median, is the mediocre.

Thirdly, on the presumption, right or wrong, that the average is where you find the most people, the average is the point every other bastard is targeting too. As Cixin Liu put it, “In the cosmos, no matter how fast you are, someone will be faster; no matter how slow you are, someone will be slower.” For our purposes, the average is the cosmos. Per Anita Elberse there is a contraflow in the market system that militates against the long tail: the blockbuster effect: everyone is aiming at the volume end of their realistic segment of the market. (Elberse’s prescription is to go with it; Sutherland’s is to defy it.)

Don’t be logical when everyone else is being logical

This is a corollary of designing for the average. To be logical is to be predictable. To prioritise data-derived logic over imagination is to converge on the same spot as your competitors, leaving the rest of design-space to the unconventional thinkers.

While you and your fellow bald men race to the bottom in a fight over the same comb, someone else is eating all the pudding you didn’t have the imagination to see. It is to see the world as mediocristan, obeying a normal distribution, and able to be navigated by probabilities, which are better calculated by machine than human.

A flower is simply a weed with a marketing budget=

It is not about efficiency. Sometimes the very inefficiency is what marks out effectiveness — the peacock’s tail.

Logic kills off magic

People don’t perceive the word objectively. So, address people’s perceptions of reality, not necessarily reality itself.

A good guess which stands up to empirical observation is still science

See also Nassim Nicholas Taleb’s remarks about “lecturing birds to fly”. Random accidents can — and usually do — generate progress. On which Sutherland invokes the impish figure of Paul Feyerabend who, with Thomas Kuhn, is one of the JC’s favourite contrarians in the philosophy of science.

Test counterintuitive things because no-one else will

See also no-one got fired for hiring IBM. Create a small space where people can test things that don’t make sense. So, a Skunkworks. All the more reason if your competitors are too scared to go there.

Solving problems using only rationality is like playing golf using only one club

Rationality has its uses; but it is a naive model of the world, and what it leaves might be more important than what it leaves in. The moment you say is “the way to solve the problem is like this ~” you have defined the problem in a way that allows only a very small solution set. There are lots of reasons people behave as they do, and economic incentives only over a small part of them.

See also

• Behavioural economics