Wandering as a Method
Jeff Bezos argues that genuine invention often requires the opposite of efficiency: time to wander without knowing exactly where the solution will lead. He applies that idea to problem-solving meetings, where he says a predetermined path or duration can be a sign that the answer is already understood. The same philosophy underpins his broader discussion of invention and Blue Origin, where long-term experimentation matters more than following the shortest route to an incremental improvement.
The habit traces back to a Texas ranch. Bezos spent every summer from age four to sixteen with his grandfather, a man who made his own veterinary sutures from heated wire and spent an entire summer rebuilding a broken-down 1955 Caterpillar D6 bulldozer bought for about $5, 000, mail-ordering transmission gears so heavy the pair had to build a crane just to move them.
What Bezos calls the great gift of those summers was learning that problems yield "with enough persistence and ingenuity, " a self-reliance he now treats as the psychological foundation of his inventive method.
He is candid that the process resists introspection. Asked how his own thinking works, he admits he honestly does not know, only that it involves lots of wandering and a tolerance for ideas that arrive as mere kernels.
When a new idea of his survives first scrutiny, he warns colleagues that objections will be easy to find but asks them to work with him anyway, an act he calls intuition, a defense of fragile ideas against the killing ease of early criticism.
Bezos's Rules for Fast, Honest Decisions
- Sort the doors firstClassify every decision as one-way (irreversible, scrutinize hard) or two-way (reversible, decide fast and move on).
- Disagree, then commitState your objection once, then actively help execute the decision instead of sniping or saying I told you so.
- Never resolve by exhaustionStubbornness contests and compromise both fail to reach truth; escalate to a decision-maker instead.
- Trust the anecdotesWhen customer stories contradict a metric, doubt the measurement before doubting the customers.
- Guard the proxiesMetrics invented years ago can quietly lose their connection to the underlying truth; re-examine why each one exists.
Deciding Fast, Deciding Well
Wandering produces ideas; Bezos's second discipline is deciding quickly what to do with them. At Blue Origin, where he now says the goal is to become "the world's most decisive company, " his framework is the famous one-way and two-way doors. Two-way door decisions (most decisions, he insists) should be made fast, by individuals or small teams deep in the organization, with full understanding that you can walk back through the door.
One-way doors, the irreversible ones, deserve deliberate slowdown, and as Amazon's CEO he often found himself cast as the "chief slowdown officer" for exactly those calls.
The framework pairs with an unusual honesty about disagreement. Rather than compromise, which Bezos calls a low-energy way to avoid truth, or a war of attrition that rewards stubbornness, he advocates escalation and "disagree and commit": state your case, then genuinely back the other person's call, gambling with them because they may be closer to the ground truth.
The real bottleneck, in his telling, is rarely analysis, it is the time organizations take to decide at all.
Keeping decisions honest requires what Bezos calls truth-telling, and he is clear-eyed about why it is hard: humans are social animals, not truth-seeking ones, and the village truth-teller historically risked getting clubbed in the night. His countermeasures are structural.
He speaks last in every meeting so junior voices arrive unfiltered, he wants the most junior person able to overrule the most senior one if they have data, and he is alert to optimism bias, when two interpretations of new data exist, he defaults to the unhappy one until evidence says otherwise.
- Most decisions are two-way doors: decide quickly, by individuals or small teams, knowing you can reverse course.
- One-way door decisions deserve extra analysis; Bezos as CEO would deliberately slow them down as the chief slowdown officer.
- Compromise and stubbornness are both low-energy ways to avoid truth; escalate instead, then disagree and commit.
- When data and anecdotes disagree, the anecdotes are usually right, meaning you are probably measuring the wrong thing.
- He speaks last in meetings so that hierarchy does not filter junior opinions before they are heard.
New Glenn and the Only Interesting Problem
All of this philosophy now funnels into one machine. New Glenn, Blue Origin's heavy-lift rocket under construction at Cape Canaveral, will carry about 45 metric tons to low Earth orbit on roughly 3.9 million pounds of liftoff thrust (a little more than half the thrust of the Saturn V) driven by seven BE-4 engines fueled with liquefied natural gas.
The expendable second stage runs on two BE-3U hydrogen engines, chosen because hydrogen's high specific impulse pays off most where tanks stay small. Getting to orbit itself, Bezos argues, is a solved problem from the 1950s and '60s; the only interesting problem is dramatically reducing its cost, because cheap access is what opens space to everyone else.
That framing makes the real challenge surprising. The first launch, targeted for 2024, matters less than rate manufacturing: to fly New Glenn 24 times a year, Blue Origin must produce an expendable upper stage every two weeks and a BE-3U engine every week, indefinitely.
Bezos considers building the first article merely the vehicle's test of the factory, the harder, quieter work is engineering facilities, inspection, and acceptance testing so the whole system can run at rate without losing the cost advantage that motivated it.
He is not pretending the stakes are comfortable. Asked whether he is nervous about the debut flight, he answers that a hundred percent nervousness is the only sane response, and that some problems can only be found in flight. The urgency is deliberate: he left the Amazon CEO role primarily to inject energy and speed into Blue Origin, which he says plainly "needs to be much faster."
The Six-Page Memo, Decoded
- Study hall firstMeetings open with 30 minutes of silent reading of a six-page narrative memo so everyone actually absorbs it.
- Narrative beats bulletsComplete sentences with topic sentences make sloppy thinking hard to hide; bullet points make it easy.
- Hard on the authorA good memo can take two weeks of rewriting and hole-poking, but it saves the whole room time.
- Messy discussion on purposeAfter the crisp document, the meeting should wander through questions nobody can yet answer.
Moon Machines and Thinking Machines
The cost thesis extends to the moon. Blue Origin's MK1 lander is an expendable cargo vehicle carrying 3, 000 kilograms to the lunar surface on a single New Glenn flight, while the MK2 lander for NASA's Artemis program is a single-stage reusable design meant to land, take off, and fly again, built to make lunar missions affordable because, as Bezos puts it, this time the point of going back to the moon is to stay.
Behind both sits a quieter bet: solar-powered cryo-coolers that could make boil-off-prone liquid hydrogen storable in deep space, which he calls a game changer for high-energy missions to the moon, Mars, and beyond.
Back on Earth, Bezos brings the same inventor's lens to artificial intelligence (and offers its most striking framing. Large language models, he argues, are not inventions but discoveries: unlike a fully engineered 787, whose behavior is designed and predictable, LLMs constantly surprise their creators with capabilities nobody specified. That uncertainty does not dim his optimism.
Even against real risks, from specialized AI weapons to the immaturity he sees dramatized in Oppenheimer, he believes these tools are "much more likely to help us and save us even than they are to unbalance, hurt us, and destroy us") perhaps by helping humanity avoid extinguishing itself.
He is equally direct about what remains unsolved. The human brain runs on about 20 watts where today's AI techniques burn many kilowatts, and a teenager learns to drive on far fewer miles than the billions self-driving systems need, so he does not think scaling alone has learned the last trick.
On the product side, he points to Alexa's installed base of hundreds of millions of devices that is "about to get a lot smarter, " and to AWS's Bedrock, built so enterprise clients can use powerful models with their own data without accidentally contributing it back.
A Trillion Humans and a 10,000-Year Clock
Zoomed out, the vision behind all the hardware is startlingly specific: Bezos would like to see a trillion humans living in the solar system, which he calculates would mean a thousand Mozarts and a thousand Einsteins alive at any given time.
The only way to get there, he argues, is giant O'Neill-style space colonies built from lunar and asteroid materials (planetary surfaces are simply too small) spun for Earth gravity and clustered near Earth so residents can vacation on the home planet the way tourists visit Yellowstone.
The motive, strikingly, is environmental. Earth, he says, is "the good planet" (the only one we know of among all the worlds our probes have visited) and the one thing moving backwards in an otherwise improving world is the pristine natural world we traded for industrial gifts.
Rising energy use per capita is incompatible with a finite planet, he argues, so heavy industry belongs in space not as an escape from Earth but as its preservation plan.
He frames his own role as infrastructure: just as Amazon stood on the shoulders of credit cards and the postal service, he wants to spend his Amazon winnings building the road to space so that, one day, a valuable space company can start in a dorm room.
That horizon explains his strangest project. Inside a west Texas mountain stands a 500-foot-tall mechanical 10, 000 Year Clock, conceived by Danny Hillis, that ticks once a year and chimes once a century (a symbol, he says, for stretching human thinking beyond the five-year windows most institutions manage. Long-term thinking, in his view, is "a giant lever": it lets you solve problems that are impossible to solve short-term.
He knows the clock's real influence may not arrive for hundreds of years, after it has taken on the patina of age) an invention, characteristically, designed to be wandered toward rather than rushed.
Charts & Visual Insights
New Glenn Thrust by Stage of Power
Liftoff thrust of the full New Glenn vehicle compared with the per-engine thrust of its BE-4 first-stage engines and BE-3U second-stage engines, in pounds of thrust.
| Component | Thrust (lbs) | Source |
|---|---|---|
| New Glenn liftoff | 3,900,000 | |
| BE-4 engine (each of 7) | 550,000 | |
| BE-3U engine (each of 2) | 160,000 |