The Adjacent Possible + Low-Resolution North Star
Take only safe next steps — but always the one with the steepest gradient toward a 3-5 year North Star.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 95%
The adjacent possible — the set of actions genuinely within arm's reach that would almost certainly work — is much smaller than tech culture assumes. Rather than a flying leap, you take small safe steps; each one reconfigures the world and generates a new adjacent possible. But pure incrementalism random-walks you into a corner, so you pair it with a deliberately low-resolution North Star three to five years out and always choose the reachable step whose gradient pulls hardest toward it.
Origin
Komoroske credits 'the adjacent possible' as coming from design thinking — he encountered it via the Designing Your Life authors. (The term originates with complexity theorist Stuart Kauffman.) The North Star pairing and the gradient-selection rule are his own synthesis. He also references Buddhist practice — 'point your cart in a direction and start walking' — as the same idea.
Core principles
- 01Your adjacent possible is small. That is not nihilism — it means you have full agency within it.
- 02Every action you take reconfigures the world and generates the next set of options.
- 03Slicing decisions smaller strips out risk while keeping full exposure to the upside. Risk comes from jumping too far in an unknown-unknown environment.
- 04Incrementalism alone follows the steepest local user gradient and random-walks into a corner.
- 05Long-term vision alone builds castles in the sky. You need both.
- 06False precision — 'will it be 93 or 95 in five years?' — is a comfort blanket bought at great expense.
How to run it
- 1
Write a low-resolution 3-5 year North Star
Describe a future state deliberately at low resolution. Two tests must both pass: (a) plausibility — everyone with relevant knowledge, from a lawyer to someone who shipped a similar product at Microsoft 30 years ago, reads it and says 'I can see how that could work' rather than 'a miracle happens here'; and (b) worth-it — if you actually landed it, everyone would high-five, not shrug. 'I guess that could work' at the end means the North Star is not worth having.
Pro tip Low resolution is a feature: the star will drift slightly in the sky over time, and because it is far away you arc gently instead of jerking around.
Watch out A high-resolution five-year vision manufactures false precision and forces you to defend detail you cannot possibly know.
- 2
Enumerate your actual adjacent possible
List the actions genuinely within arm's reach right now — things that, if you did them, would almost certainly work or at minimum would not be too expensive. Be honest about how short that reach is.
Pro tip Slice decisions into smaller and smaller ones until each next step obviously pays for itself.
Watch out Do not confuse the adjacent possible with the aspirational possible. The whole method depends on the steps being genuinely safe.
- 3
Pick the steepest gradient toward the North Star
Among the reachable options, do not automatically take the one users demand most. Take the one that pulls hardest in your North Star direction — often the second-most-demanded option — provided it is still safe and reasonable on its own terms.
Pro tip This single substitution is what converts a random walk into an arc.
Watch out Following pure UXR demand means you will end up exactly where the steepest local gradient leads — which is rarely anywhere interesting.
- 4
Take the step, then re-read the board
After acting, the world reconfigures and a new adjacent possible appears, partly shaped by the step you just took. Re-enumerate and repeat. If the follow-on options you hoped for did not materialise, that path is done — go a different direction for a while. If they did, take another step.
Pro tip Cheap abandonment is part of the design: no single step was expensive enough to trap you.
- 5
Refuse false precision on the way
When someone demands the five-year number ('will it be 93 or 95?'), refuse the analysis if the answer does not change the decision. If the idea genuinely compounds, the outcomes are zero or a thousand — not 93 or 95 — and precision between 1000 and 1001 is worthless.
Pro tip Accept an inefficient path. Optimising for efficiency in unknown terrain leads to doing nothing at all, or to dangerous decisions that do not work.
Watch out Made-up numbers presented as data are worse than admitted uncertainty — they hide the uncertainty rather than manage it.
In the wild
Komoroske's argument against forecasting precision: things with a compounding return do not pay out 93 or 95. They pay out zero or a thousand. Whether it is a thousand or a thousand and one is irrelevant — it is orders of magnitude larger than the alternative either way, so the entire precision exercise adds nothing to the decision.
→ You free up the analysis budget entirely, and the decision gets made on order of magnitude rather than on invented decimal places.
Lenny's silent meditation retreat: he asked how to be successful while not clinging to a specific outcome. The answer was to point your cart in the direction you want to go and start walking — do not pre-compute the exact path. Afterwards you will see the curves and jogs and say it could have been more efficient.
→ The slightly-inefficient path taken beats the perfectly-planned path never taken, because you were navigating unknowns that could not have been known up front.
Common mistakes
The flying leap
Jumping to the end state in an unknown-unknown environment is where the risk actually lives. Slicing the same journey into safe steps removes most of the risk while keeping all of the upside.
Incrementalism with no star
Following only the steepest immediate gradient — what UXR says users want next — random-walks you into a corner with no coherent position.
A high-resolution vision
A precise five-year picture is a comfort blanket. It manufactures certainty where there is none, costs a fortune to produce, and cannot survive contact with the first reconfiguration.
Is it for you?
Best for
Product and strategy leaders building a roadmap in genuinely uncertain terrain, who need direction without pretending to know the path.
Not ideal for
Well-understood domains with a known destination and a known route, where planning the whole path up front is cheaper than discovering it.
From the transcript
“the frame is the adjacent possible which is I believe comes from design thinking”
“when you recognize that your adjacent possible is relatively small you realize that you actually have full agency to pick within”
“you need coherence about where you're going and the way you get that is by creating a kind of North Star for yourself”
“it should be very low resolution you should describe a thing that every single person who reads it who has any kind of knowledge that…”
“then what you do is you look down your adjacent possible and you look for the thing that has the steepest gradient that pulls you…”
“because if you don't do if you only do incremental then you'll end up random walking into a corner and if you only do the…”
From the episode
Thinking like a gardener not a builder, organizing teams like slime mold, the adjacent possible, and other unconventional product advice
Alex Komoroske (Stripe, Google)