The Right Problems
"Sounds like AI" deletes the best thinking first. Using tools to create isn't inhuman.
Every tool since fire had its deniers, and history marks them irrelevant fast.
(Also your tools for detecting this kind of stuff are worse than you think. Pangram flags this note as AI-written. I typed it myself. Point proven.)
In the past, I haven't spent enough time picking the right problems to work on. Pretty quickly after starting to explore why, I noticed that I have loose definitions for both "right" and "problem".
What is a "problem"
Before we even have a chance to find the right problems, we need a way to think about what problems are to begin with.
To start — it's easier to understand problems in specific contexts like solving a math equation, finding a profitable way to deliver a good or service, handling traffic in a distributed system, etc., but it's a generic definition of problem which evades me.
The first step I took to find a generic definition was to make a big list of different problems in business, science, engineering, self-improvement, personal relationships, and economics, then read them over a few times and let them sizzle in my unconscious while I went about my life for a while. Eventually, halfway through watching 21 Jump Street with a couple of friends, a framework came to me and I'm taking a moment to write it down:
Defining a generic problem
It turns out you only need a few pieces to define a generic problem. There's a current state S and a function f that describes its dynamics, plus a desired state S' that satisfies some condition or goal, all subject to a set of constraints C.
The problem, then, is to find or design a transformation f' such that f'(S) = S', optimizing for C. In practice, finding f' tends to mean diagnosing (understanding f) before you can design anything (proposing f'), and only then does implementing (applying f') make sense.
I like this abstraction because it fits the essence of most problem-solving scenarios I've run into, though it still keeps its intuitive appeal.
Examples of problems that fit this spec
Everyday life: Say the current state is S = keys are lost. Then f' is whatever set of steps actually finds the keys, which is to say retracing your steps, checking the usual spots, calling the last place you remember having them, until you reach S' = keys are found. The constraints C here might be how much time you have to search, which spaces you can physically get into, and whatever state your memory happens to be in that morning.
Business: Here the current state S is one where f(customer engagement strategy) = high churn. So f' is some improved version of how you engage customers, maybe personalized outreach, better onboarding, more proactive support, that gets you to S' = low churn. Your constraints C are the usual suspects, budget and team capacity and whatever is technically feasible.
The right problems
This definition of problem even works for the problem of finding the right problems!
Problem: Current state S where f(all possible problems) represents our unfiltered view of problems. The solution function f' implements a filter based on our values, capabilities, and impact potential, leading to desired state S' = high-value problems worth solving. Constraints C include our time, resources, skills, and the requirement that solutions create lasting positive change.
My filters
When I think about the problems that I want to spend my life on, they share a few characteristics:
- if solved, the trajectory of society is permanently changed for the better (impact of
S → S'), - and the probability of success is so remote that even attempts are viewed as a fool's errand by most rational observers (The path from
ftof'is ambiguous).
Aside: Roles
That said, just having interesting problems in my vicinity isn't enough. My role in solving the problems is also important — I need real ownership to give my life to solving a problem. Ownership means something very specific to me: It means that my success or failure is judged on outcomes over a sufficiently long time horizon such that I can make unpopular and seemingly insane bets with enough time for them to win out in the end.