Unpredictable Patterns #73: The allure of analogy
Plausible mechanisms and assumptions as key elements in thinking creatively about everyday problems
Dear reader,
Spring is coming to a close, and summer is at the door. The beauty of these early summer nights is magical and the scents of lilac trigger memories from days gone by. The fact that our nose seems to keep so many memories for us is really intriguing, but I must say that for me that is very much the case. Scents and taste - I suppose this proves the Proustian cookie conjecture. Thank you for the comments to last week’s note! There is a lot to say about the kinds of things that we propose exist in our world, and the idea that the world has “natural kinds” - as some pointed out - is strong in us, and something we might want to be more attentive to. The way we make worlds, as philosopher Nelson Goodman noted, is after all primarily dependent on composition and decomposition, wholes and parts. Changing wholes and parts can reveal interesting things about our models of reality. In today’s note we will look closer at analogies and how they use us and we use them!
Plausible mechanisms
Often when we look at problems and try to solve them, we end up discussing different courses of action. Say that we want to change the image of the organization we work for, because we believe that it has been maligned in an unfair way: what typically ensues is a meeting in which different courses of action are proposed, and the pros and cons of those actions are discussed. Then an option is chosen and we go with that option. Sounds logical, after all - right?
The challenge with this is that it misses a step, and potentially an important step at that: we spend very little time discussing the plausible mechanism that connects our course of action with the desired result.
Let's continue with our example. One group is suggesting a marketing campaign, another is suggesting an interview with the CEO to clear things up. The pros and cons are discussed at length - marketing campaigns are expensive, the CEO does not have the time to be briefed right - and then the whole thing is put to a vote. But should we not also have asked the simpler question of why we think any of the options would work?
If we suggested that to the people in the room it is not unlikely that they would be irritated: we surprisingly often suppose that we have a shared model of reality and how things work - and we find it tedious to have to explore such basic things. This, in turn, has to do a little with how we think. We are pattern matchers - so we look for things that we have done before when other things have happened, and we even learn that way from others and look for what others do when things like this happen.
We may find, therefore, that the advocates of the marketing campaign argue that company X did a really successful marketing push and managed to increase their trust numbers or that the CEO at Y company rescued the company through an open-hearted interview on TV, and so assume that those same things will work for us.
This - the analogy trap - is a pernicious thing because it works so often for us in daily life. Carl wanted to lose weight and cut carbs, so I will cut carbs and then I will lose weight. Or Penelope was unhappy in her relationship and went to a therapist, and now she is happy, so I will now go to a therapist to fix my relationship. And it works! So why should we bother to examine the analogies closer?
The point of pausing and exploring the plausible mechanism is that it allows for the construction of a shared model of reality - one in which we spell out what we think does what, and the patterns of cause and effect - not just the patterns of correlation.
Let's see what that could look like. If we asked the question of the plausible mechanism in the marketing campaign case, our advocates might end up saying something like: typically the opinion of our company is determined by the way our consumers talk about us, they are the main source of our reputation and so a broad marketing campaign that highlights what we know the consumers want us to do will set that source straight and will fix our reputation issues.
Now it is obvious what the value of exploring the plausible mechanism is: we have just found a key hypothesis that was just implicit or assumed: consumers are the source of our reputation. Is that really the case? Is there any way we can substantiate that with data points? Or are is the organization's reputation based on a poly-centric network of different actors? If so, how should a marketing campaign be distributed over these many centers of reputation gravity?
The pause to look for plausible mechanisms is a way to interrogate our pattern matching and our analogies at a deeper level.
Analogical reasoning
Cognitive scientist Douglas Hofstadter once hypothesized that our thinking is, at its core, analogical - and that creativity and discovery depend on the careful exploration of this as that. Questions like how is X like Y are fascinating and really tease the best out of our intellectual capability - and we should in no way demean analogies. But we need to be on our watch for what they miss -- analogies are examples of Kahneman's system 1, and, frankly, it would be much better if we called them analogies rather than biases, because that would highlight how they have arisen in our minds.
Kahneman's biases are mostly analogies that have proven to be valuable over evolutionary timespans - and the way to approach them is not to condemn them, but to look for plausible mechanisms that underlie the analogical conclusions that we draw - and if they are lacking, build a better model on what we find.
Moving from analogical to logical thinking is not easy - but it is almost always rewarding. The exploration of the vast undergrowth of assumptions and images that hold us hostage in the subterranean analogical world is the perhaps best way we can find new ideas and solutions to really hard problems.
One approach to this that is specifically focused on planning is the so-called assumption based planning model. In this model, the planner lists the assumptions first - and then looks at various possible options. The trick is to make the assumptions explicit to start with, so that when the plan is executed the organization can keep an eye on the assumptions and see how they hold up. It is also a way to explore what would be true if the assumptions are wrong.
This is where we find another very powerful tool: the counterfactual.
Counterfactuals are devices that help us think in alternate realities - they are used for imagining what the world would be like if the facts were different in some way. Counterfactuals are a species of thought experiment, and they are particularly powerful when used to understand why we are where we are today.
One simple method of deploying them starts with looking back at history and identifying points of indeterminacy - points where things could have gone very differently - and then look at how the world would have evolved if, at any of these points, the world would have looked different than it actually did.
The counterfactual for our company, then, is simply this: at what points in our history could things have been different and led to a situation in which we did not have a reputation problem? The group studying the problem might end up noting that the release of one product with a lot of problems was a key point of indeterminacy, and if just this product had not been released, things would be immeasurably better - if so that is an interesting insight into the source of the problem!
Counterfactuals are sometime used to assign moral blame as well, but that is less helpful and as I have written elsewhere I find blame supremely uninteresting. That said, if the group can identify a consistent pattern of points of indeterminacy where the company has chosen the wrong thing to do - then maybe the entire decision process of the company can be improved with that data.
There are those that condemn counterfactuals as useless "whatifs", but this really misses the point: the use of counterfactuals helps you explore your own decision patterns and their robustness. It is equally possible to look at all the points of indeterminacy and say that there is nothing that we could have done that would have prevented this reputation backlash - it is actually the product of the more general pattern of our industry, or of technology or of politics in a certain economic cycle. That is fine - but if so, all of those patterns can be interrogated more closely to understand where the indeterminacy sits.
Counterfactuals are tools of assumption archaeology, and allow you to study how assumptions have changed and impacted a company or an organization over time.
Polling for analogies
Many companies use polling to understand their environment. They look for approval rates and try to understand who is a detractor and who could be a champion. These studies have their worth, but it could be at least equally interesting exploring the analogies you exist in.
Say that we want to understand social media and the general debate about social media. We can choose to poll social media users and ask them their opinion, but we can also explore the analogies that seem to be governing the field for now. A very simplistic, but interesting first draft way of this is to use a search engine's autofill feature. Here is what we find if we do this for social media:
The picture is harsh, but it at least gives us a baseline of analogies that currently seem to be the dominant ones. Knowing this we can then look at if there is a good way to counter them, or provide new analogies - or even recognize that these analogies are dominant and start from there.
Knowing the analogical landscape of an issue is key to understanding an issue. It is also the key to defining one. As noted in the famous book Metaphors we live by (1980), by George Lakoff and Mark Johnson, our language frames our understanding of the world on deeper levels that we often are oblivious of. Now, Lakoff speaks of metaphors and we have spoken of analogies - but I prefer to just use analogies because they are more fundamental. A metaphor is merely an analogy dressed up to go to an academic conference.
Studying analogical landscapes also allows us to figure out connections between different seemingly innocent words. Analogies are contagious on basic levels: the idea that social media is like a drug is certainly connected to the fact that we have - for ages - been speaking about internet "users" and so the user-language provides a bridge into the drug landscape.
In some cases public policy is almost reduced to two analogies battling it out. Here is what the analogical search engine test for data looks like:
Her we see data is like water and data is like oil - two very different analogies - close to each-other and we can note that in between them they also represent two very different mechanisms for how data creates value: through scarcity (oil) or abundance (water). It is also noteworthy that the "data is like a bikini" appears here - analogies are often powerful if they are a little bit absurd, funny or filled with innuendo. This is a good reminder that analogical thinking is system 1-thinking and vulnerable to the simple pun.
(The story behind that analogy is left to the reader to find!)
Know your analogical landscape well!
Playing with analogies
Analogies can be used generatively as well. Since we are hard-wired to think analogically, we can use this ability to explore problems creatively. One way to do this is imply to look at our problem, and use the standard analogical question:
How is my problem X like this other thing Y?
Now, the choice of Y can be anything, but there are a number of things that are quite general and generative (we have mentioned this in an earlier note). They include:
How is this like a game?
How is this like a musical piece?
How is this like mathematics?
How is this like war?
How is this like evolution?
How is this like a garden?
How is this like the human body?
These analogical questions are sure to force you to think about your problem in a new way, and if you pick other, more specific areas you may find even more interesting insights. A key to analogical creativity is to use something you know well - if you are an avid aikido practictioner, you may look at a problem and ask how this is like martial arts and how you would like to approach it. If you are a stellar musician, you may ask how the problem would best be described in music.
Analogical creativity can be used to dig even deeper, by looking at specific pieces of different systems. The general question then becomes:
What is X to Y as B is to A?
Examples are easy to find, and allows for a second level analogical exploration of different systems. What is to the information economy as pollution is to the industrial economy? A simple question, for sure, but one that holds a ton of implications depending on exactly how it is answered.
Analogies can be complex - and one of the many, many reasons for reading fiction is that fiction provides you with complex narratives that can be used analogically in different situations. Fiction is believable, as Aristotle noted, when it bears a resemblance to the real world and it is plausible.
So we are back to our plausible mechanism, again. And this is interesting - because when we descrived the plausible mechanisms we need to tell a little story about what the world is like, and that is something that then helps us navigate together as a team.
Plausible mechanisms outline possible worlds, as judged by their verisimilitude.
Analogies, then, are based on pattern likenesses at different levels - and so become a part of our pattern matching capabilities. But we are not just matching patterns, we are also abstracting them. In a sense you can say that an analogy is produced by taking an image and then reducing the resolution and then look for other images that match that pattern and increasing resolution again. But what we are reducing is not just resolution, it is something else - we are stripping a pattern of what is not seen as essential to it, we are reducing it to a core of some kind.
How we do this is interesting: there is infinitely many ways of reducing a pattern in complexity, and so the fact that we find some of them to match is not a simple feat.
Analogical thinking is a truly beautiful thing.
So what?
A few things from this:
First, it is worthwhile taking the time to dig beneath the analogies for plausible mechanisms and assumptions. Calling this out occasionally makes for good hygiene in thinking.
Second, it is always worthwhile applying more than one analogy to a case, and exploring what other analogies could teach us. A good analogy is the best response to a competing analogy!
Third, understanding the abstraction of pattern that makes an analogy possible is a great way to get insight into other people's thinking. The idea that data is like oil is based on an abstraction from raw materials extraction that focuses on the cost and extraction, not in the amount. Now, that tells us something: what worries the person who uses this analogy is extractive behavior and economic forces.
Fourth, it is important not to get stuck in analogies. They are tools for thinking, not reality. The worst thing that can happen is that you become so enamored with your analogies that you mistake them for ultimate reality.
Thank you for reading!
Nicklas
Dear Nicklas,
I love this piece and the proposition about moving from analogical thinking to logical thinking. And the focus on the plausible mechanism is a good framework. You acknowledge that moving from analogical to logical thinking is not easy, but how does it happen in practice?
I recall a very specific instance about a decade ago involving an internal conflict very much like the one that you described above: marketing, policy, the "L-Team," etc. I kept trying to intellectualize and rationalize everyone's absurd behavior and rushed decisions, and your advice to me was basically this, approximately:
"This kind of conflict in the company is like a knife fight. You're intellectualizing a knife fight and your weapon is a chess board, not a knife. You have the wrong weapon and the wrong field of play. You need to recognize appropriately and in kind. As [forgot who you attributed to, maybe Sun Tzu] said, never bring a chess board to a knife fight."
Your analogy about the knife fight and chessboard wasn't that I should take up a knife, but that his *was* a knife fight and dynamics of knife fights are emotional, not intellectual challenges (like a chess match is). I changed tactics and navigated the situation based on your advice.
Of course, every org conflict is unique, but is the implicit advice you're suggesting here sounds to me to be a general change in thinking, i.e., that maybe we should be insisting on more chess games even in the middle of knife fights.
Also, given the focus on analogies, your piece reminded me of another piece great advice that you gave me about use of analogies. I felt like my title might as well have been "The Recursive Department for Internet Analogies Department" because that's all we did was combat analogies, and no analogy worked (Internet is like the electric grid; Internet is like water; Internet knows no borders like the pre-Westfalian world didn't have borders, etc. . . .). The killer advice came when you prepped me for a "murder board" where I was in an analogy contest of some kind and you said that the best response in that case wasn't go come up with another analogy --- but to insist the best analogy isn't an analogy at all.
"No, I don't think your analogy for the Internet fits because the best analogy for the Internet, at this point, is the Internet itself. It doesn't need an analogy."
I've since totally weaponized this. You can insist in response to any public attack that "the better analogy for X is X itself." It throws people off and works every time. I gift that advice back to you, my friend. :)
"The point of pausing and exploring the plausible mechanism is that it allows for the construction of a shared model of reality - one in which we spell out what we think does what, and the patterns of cause and effect - not just the patterns of correlation." -- 💎
I also really liked your insight about analogical thinking being system 1. Very interesting. It also seems like mental models and patterns and analogies are due for some semantic disambiguation.