Monkeys and pedestals: do the hard uncertain thing first
Monkeys and pedestals: do the hard uncertain thing first
One-line summary: Astro Teller's moonshot rule, via Duke: the bottleneck is training the monkey to juggle flaming torches; building the pedestal is something you already know you can do. Pedestal-first is false progress that accumulates sunk cost and makes the later quit harder.
What it is
Teller (captain of moonshots at X, Google's innovation lab): if the side hustle is a monkey juggling flaming torches on a pedestal in the town square, do not start by building the pedestal. Pedestals have been solved for thousands of years. The possibly intractable piece is the monkey. "Monkey first."
Why it is a quitting tool, not just a project-management slogan:
- False progress. "There's progress and there's false progress or the illusion of progress. And if you build a pedestal, it's the illusion of progress, because you already knew you could do it." Low-hanging fruit feels like momentum.
- Sunk-cost snowball. Every dollar and minute on pedestals before the monkey is cost you will not want to waste, which then makes you less willing to quit when the monkey is unsolvable.
- Cheaper falsification. Teller: learning you cannot do it after $2M is not a waste of $2M; it is a saving of $7M versus learning at $9M.
California high-speed rail is her disaster case: the monkeys are the Diablo Range and Tehachapi Mountains; the pedestals are track on flat land (Madera–Fresno, then Bakersfield–Merced, then SF–Silicon Valley). Budget $33B → $81B → ~$120B; still building pedestals. Escalation of commitment dressed as "let's tackle the low-hanging fruit."
X itself was named later: they "just wanted to figure out whether this was going to work" first. Capsule: "do the hard thing first and beware of false progress."
ted-seides applies it to his own fund launch: the monkey was whether he could raise; the pedestals were legal docs and back office. He built pedestals first.
Why it matters to psychology
This is a mental model for where kill-criteria should point: at the uncertain bottleneck, not at tasks that were never in doubt. It also explains a mechanism of escalation — not just "I already spent money" but "I spent money on the easy parts, which produced a feeling of progress that was not information."
Evidence
- annie-duke in 2026-08-03-podcast-capital-allocators-best-of-decision-making-annie-duke-2022: "when you're approaching a problem, you want to think about what are the bottlenecks? What's the really hard thing that needs to happen in order for me to be able to unlock this problem? And that's the thing that you're supposed to go after first. You're not supposed to build the pedestals."
- annie-duke in 2026-08-03-podcast-capital-allocators-best-of-decision-making-annie-duke-2022: "If I build track between Bakersfield and Merced, it's the illusion of progress, because I already knew I could do it."
- annie-duke in 2026-08-03-podcast-capital-allocators-best-of-decision-making-annie-duke-2022: "every minute that you spend building a pedestal, every dollar that you spend building a pedestal before you've tackled that monkey, those are sunk costs that you're accumulating."
- annie-duke in 2026-08-03-podcast-capital-allocators-best-of-decision-making-annie-duke-2022: "If I can figure out after spending $2 million that we can't do it instead of figuring that out after we've spent $9 million, that's not a waste of $2 million. It's a saving of $7 million."
- annie-duke in 2026-08-03-podcast-capital-allocators-best-of-decision-making-annie-duke-2022: "I think it's do the hard thing first and beware of false progress."
- ted-seides in 2026-08-03-podcast-capital-allocators-best-of-decision-making-annie-duke-2022: "when I was planning to launch a fund earlier this year, the obvious monkey is how easy is it going to be? Raised the money. And some of the pedestals are getting your legal docs done and all the back office stuff... I built all these pedestals before I tackled the monkey."
Implications
- Sequence work so the first spend is a test of the uncertain node. Feasibility on the mountain range before track in the valley; LP interest before the PPM.
- "Low-hanging fruit first" is a recognized opposite heuristic and should be treated as a trap when the fruit is not the bottleneck.
- Pairs with kill-criteria: if the monkey fails the precommitted test, stop — do not "at least finish the pedestals."
Open questions
- How do you distinguish a real monkey from a romanticized hard problem that is not actually on the critical path?
- Is there a justified amount of pedestal-building for morale/team-coherence, or is that always the illusion Teller warns about?