Gardening Over Building (Farming for Miracles)
Plant many cheap compounding seeds instead of manufacturing outcomes with brute effort.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 95%
The Builder mindset makes a plan and forces reality to match it — which caps your output at the effort you put in. The Gardener mindset instead looks for things that grow on their own: seeds with a compounding or network-effect shape, planted cheaply, watered only where signal appears. You do not need to know in advance which acorn becomes an oak; you need the seeds to be cheap enough that being wrong costs almost nothing.
Origin
Alex Komoroske's own framing, developed across 13 years at Google (Chrome open web platform, Google Maps AR) and Stripe corporate strategy, and published in his 'Bits and Bobs' working notes. It draws on complexity/systems thinking — he credits Eric Beinhocker's 'Origin of Wealth' and Donella Meadows' 'Thinking in Systems' as his entrée into that lens.
Core principles
- 01Builder value is capped by effort in; gardened value is not.
- 02You cannot predict which seed becomes an oak — so make seeds cheap enough that prediction is unnecessary.
- 03Look for shapes that self-accelerate: network effects, ecosystems, compounding loops.
- 04Invest incrementally, only on signal. Stop investing the moment the signal stops.
- 05Done right it looks like luck — accept that the credit will be illegible.
How to run it
- 1
Truffle-hunt for compounding shapes
Before planting anything, scan the problem for dynamics that would accelerate if they worked: ecosystems, network effects, developer/user flywheels, anything where power grows with the number of participants. Komoroske calls this truffle hunting — the shapes are common, but you have to know what you are looking for.
Pro tip Ask: 'if this worked, would it work at an accelerating rate?' If the answer is no, it is a build, not a garden.
Watch out Not every problem has a compounding shape. Forcing gardening onto linear work wastes time.
- 2
Make the seed absurdly cheap
Shrink the bet until the downside is just the opportunity cost of a few hours — a small open-source tool, a tinkery prototype, a doc. Cheap enough that failure is genuinely fine and the person planting it enjoys the planting.
Pro tip If the planter gets energy from the work, the real opportunity cost is close to zero.
Watch out If a seed is expensive, you will feel obliged to defend it — that turns it back into a build.
- 3
Plant many, predict none
Plant a broad set of seeds rather than analysing which will win beforehand. The systemic claim is not 'this one will work' but 'a high likelihood that one of these will work in a transformative way'.
Pro tip Say out loud that you do not know which one will work — that is the point of the method, not a weakness in it.
Watch out Expect to be called a naive dreamer for refusing to forecast winners.
- 4
Water only what grows
Watch for signal that someone finds the thing useful. Each time signal arrives, invest one incremental unit of energy — never a large lump. Stop watering the moment signal stops.
Pro tip Constantly seek disconfirming evidence about your own favourites; direction should be set by what is actually growing, not what you hoped would grow.
Watch out The failure mode is sentimental watering: pouring effort into a seed you like that no one else is responding to.
- 5
Protect the sprout
Young ideas are fragile and look silly. Actively shield them from the organizational squirrels who dig up acorns — reviews, resource raids, 'what does that team even do' scrutiny. Give them space and time.
Pro tip Credibility elsewhere in the org is the cover that buys the sprout its space (see the 70/30 allocation).
Watch out Making the sprout legible too early is the fastest way to get it killed.
In the wild
Komoroske's canonical shape: spend three hours building a small open-source utility and put it out there. If nobody uses it, the loss was three enjoyable hours. If a developer uses it and a user follows, that attracts more developers and more users — the thing grows on its own, and you add one incremental unit of investment each time you see a signal of usefulness.
→ Bounded downside of a few hours; unbounded upside from a self-accelerating ecosystem loop, with zero upfront prediction required.
A PM on the team is excited about an idea the leader privately thinks is weak. Rather than saying 'no, we don't have time,' Komoroske says go for it — names his concerns, points at the one part he thinks is genuinely cool, and lets them run. The PM was going to be motivated by it anyway.
→ Either the seed grows into something real, or the PM stretches their agency, learns, and gets stronger. Both branches are positive; the only cost is the opportunity cost of work the person enjoyed.
Common mistakes
Analysing which seed will become the oak
The whole economic case rests on seeds being so cheap that prediction is unnecessary. If you spend real effort forecasting winners, you have reintroduced the cost the method exists to avoid — and you will still be wrong.
Planting seeds you cannot afford to lose
A seed with a heavy cost becomes a commitment. You will keep watering it out of sunk-cost pressure rather than on signal, which converts the garden back into a plan you are forcing on reality.
Expecting legible credit
When emergence works, observers say 'that was luck — where was the heroic moment?' If you need the org to attribute the win to your effort, you will drift back to the Builder mindset and lose the leverage.
Is it for you?
Best for
Product leaders, founders and PMs working on ecosystem-, platform- or network-shaped problems where outcomes are unknowable up front and small bets can compound.
Not ideal for
Deterministic, deadline-bound delivery work with a known spec — shipping a compliance feature or a committed launch. There, planning and executing beats planting.
From the transcript
“what I look for instead are things that can be gardened things that can grow on their own and that you can sort of direct…”
“plant a bunch of these suckers you know if one of these grows into an oak treat that's great don't try to analyze beforehand which…”
“it looks like getting lucky um because what you're doing is you're you're farming for miracles”
“compounding Loops are not rare they are they're they are there's like truffle hunting you have to know what you're looking for”
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)