Reset Through Better Abstractions
Don't fear throwing work away — a better abstraction compounds and recovers sunk cost fast.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 4
- Confidence
- 90%
When you're stuck on the wrong foundation, resetting is not just emotionally healthy — it's mathematically efficient. Because progress from a better abstraction compounds (like math or code), a fresh, smaller, cleverer kernel can catch up to everything you built far faster than linear human intuition expects. Humans underestimate this because we think linearly, not exponentially.
Origin
Ivan Zhao on why Notion repeatedly threw away its codebase and rebuilt, drawing on the Smalltalk/Lisp lineage where a ~100-line kernel produces complex compounding behavior.
Core principles
- 01Momentum, not merit, often keeps you on a path that doesn't matter
- 02Sunk cost feels large but a better abstraction recovers it quickly
- 03Kernels can be tiny yet compound into complex, valuable behavior
- 04Courage to reset is a core builder skill, not just self-help
How to run it
- 1
Notice when momentum, not value, is driving you
Audit whether you're continuing a direction because it's genuinely working or merely because inertia carried you there. Zhao warns 'often time you're working on don't matter but momentum just took you there.'
Watch out Momentum is seductive precisely because it feels like progress; separate motion from value.
- 2
Accept the sunk cost as recoverable, not lost
Reframe the work you'd discard: with a better approach you can regain everything you built much quicker than you fear, because better abstractions compound rather than add linearly.
Pro tip Toby Lütke's version: be comfortable with sunk cost — 'I have done all this already and I will throw it away and start again and that's okay.'
- 3
Find a smaller, better kernel
Look for the more powerful underlying abstraction — the tiny core (like Lisp's ~100-line kernel) that unlocks complex emergent behavior. Bet on the cleaner foundation even if it means restarting.
Pro tip The smaller the kernel, the faster it compounds and the more behavior it unlocks per line.
Watch out Don't reset onto an unstable new foundation either — Notion's Web Components bet failed because the tech was too new and bugs were untraceable.
- 4
Rebuild and let compounding catch you up
Execute the reset — reset the codebase, reset the company if needed — then let the superior abstraction compound past where the old path had reached.
Pro tip Pair the reset with a physical/mental reset (sleep, change of scenery) to sustain morale through the rebuild.
In the wild
About 8-10 years ago Notion bet its foundation on Google's Web Components before React had won. The technology was so new and unstable that bugs were untraceable between source code and underlying libraries, so the team reset the codebase and the company rather than run out of time on a broken foundation.
→ Notion rebuilt on a stable technical foundation and shipped the productivity product, later citing the reset as the reason it survived.
Common mistakes
Treating sunk cost as a reason to keep going
Because humans think linearly, they overvalue accumulated work and stay on a broken path; Zhao argues a better abstraction recovers that cost faster than intuition predicts, so refusing to reset is the actual waste.
Resetting onto another unproven foundation
Notion's original reset was itself triggered by betting on Web Components, which was too new and unstable — resetting is only valuable if the new kernel is genuinely more robust, not just newer.
Is it for you?
Best for
Founders and engineers stuck on a foundation that fights them, weighing a costly rebuild against incremental patching.
Not ideal for
Situations where the current foundation works and the urge to rewrite is perfectionism rather than a genuine abstraction upgrade.
From the transcript
“you can create progress through better abstractions and that thing compounds faster can catch up to all the thing you build much quicker than you…”
“your humans are not thinking not good at thinking in terms of abstraction word exponentials we're thinking in terms of linearly”
“the kernel of things could be really small but just like math it can compound”
From the episode
Notion’s lost years, its near collapse during Covid, staying small to move fast, the joy and suffering of building horizontal, more
Ivan Zhao (CEO and co-founder)