Rarely do we need what that we weren’t already looking for.
This is because there’s a crucial difference between
a) Realizing you have a specific problem, and then finding the solution to that problem.
b) Seeing an as-seen-on-TV ad and thinking “I do have that [extremely small problem that I never realized I had], I should buy this!”
a) Going to the store with a list and only buying what you put on the list beforehand.
b) Going to the store without a list and buying whatever seems like it would be useful as you see it.
a) Asking “Is this the solution that effectively solves my specific problem X?”
b) Asking “Is it possible that this could be useful to me in any way?”
(a) consisted of examples of “thinking forwards”, in a top-down manner using inductive reasoning. It’s valuing a priori, independently-generated thoughts. It’s thinking from first principles. It’s when the observed problem generates the solution, without outside interference.
(b) on the other hand consisted of examples of “thinking backwards”, in a bottom-up manner using deductive reasoning. It’s valuing a posteriori, dependently-influenced thoughts. It’s when the solution has to explain the problem which it solves.—But if you didn’t already know, how could it be important?
This is because, while there is an infinite number of solutions to the infinite number of possible problems you could have, there is only a limited amount of solutions to problems that you actually have. Thus, picking any random solution off of a shelf will probably yield a solution that fixes a problem you don’t really have!
Moreover, when the problem generates the solution, you tend to be in an environment with multiple options, where you can choose the best one.
But when the solution generates the problem, the solution’s marketing is often able to lay a deceptive argument for why you really need it, even if you’ve never realized you needed it before!
In general, if you really needed something to solve a problem you have, you would’ve known you needed it before the solution itself to you.
Data alone cannot be inducted into a theory about causation. You can’t merely ‘go up’ from data to theory and be done with it. “Correlation doesn’t necessarily imply causation.”
You must instead create a hypothesis, and then ‘go down’ by generating new data to test the hypothesis.
I am cautious of thinking backwards. Especially when the solution is trying to get me to contribute resources like time or money.
This isn’t to say, however that unsolicited recommendations or solutions to problems are never useful. Sometimes a solution exists that actually does solve a significant problem you didn’t know you had. It's just rare.
I’ve tried to articulate the idea at the core of this article as best I can, but I still find trouble with the way I’ve explained it. If you have any ideas for how I can communicate this better, please reach out to me.
‘Think backwards’ with me by reading some of my other short essays:
Posted 2020 August 2, last updated 2021 January 5.
Get notified when new essays are posted.