The Right Hill
Build something, ship it, see what happens, adjust. It’s hill-climbing. When you’re on the right hill, this is incredibly powerful - every cycle compounds on the last, and the gap between teams that iterate fast and teams that don’t grows exponentially.
But hill-climbing has a blind spot. It finds the top of whatever hill you’re on. It can’t tell you if you’re on the right hill.
The longer I spend building things, the more I think the answer to both problems is the same: move forward, pay attention, and stay honest about what the movement is telling you.
You can’t see from here
The core problem is information. When you’re standing at the start of something - a product, a feature, a company, a career move - you don’t have the information you need to pick the optimal direction. You can’t, because the information doesn’t exist yet. It gets created by action.
You think you know what users want. You ship something, and you find out what they actually do with it. You think you understand the problem. You start building, and the real constraints reveal themselves. You think a technical approach will work. You try it, and you discover the thing nobody wrote about in the docs.
This isn’t a failure of planning. It’s a property of complex systems. The information you need to make good decisions is generated by action, not by analysis. Analysis helps - you shouldn’t move blindly. But past a point, more analysis has diminishing returns and you’re just stalling.
Herbert Simon called this bounded rationality - we never have complete information, we can’t process what we do have perfectly, and we don’t have infinite time. So we satisfice. We find something good enough and move. That’s not a flaw in human cognition. It’s the rational response to the actual conditions we work in.
Movement is signal
Movement generates the signal you need to correct course.
Ship a feature. Watch what happens. Users ignore it - that’s data. They use it in a way you didn’t expect - that’s data. The thing falls over under load - that’s data. Every one of those is information you couldn’t have gotten by thinking more carefully in a planning meeting.
This is why tight feedback loops matter so much. A team shipping weekly has roughly 50 chances per year to learn something and adjust course. A team shipping quarterly has 4. Not every cycle produces a breakthrough insight - sometimes you just learn that nothing broke. But the cumulative difference in signal is enormous. Over time, the fast team converges on something good even if their initial direction was worse.
It’s the same shape as desired state systems. You don’t need to predict the right path. You need a goal, a way to observe where you are, and the ability to adjust. The loop does the navigation. The faster the loop runs, the faster you converge.
But here’s the thing about convergence: it only works if you’re in the right neighborhood. A feedback loop will find the top of whatever hill you’re on. If you’re on the wrong hill, you converge beautifully on the peak of the wrong hill.
The blind spot
This is where it gets interesting, and it’s what makes the “move slow” argument feel compelling.
In optimization theory, this is the local vs global optima problem. Local search - hill-climbing, gradient descent, iteration - gets you to the nearest peak. It can’t tell you whether there’s a higher peak in a completely different part of the landscape. Peter Thiel makes the same point in business terms: iteration is a 1-to-n activity. It optimizes within a space. It doesn’t find the space. A startup iterating fast on a product nobody wants will converge on the best version of something nobody wants.
James March called this the “myopia of learning.” Exploitation - refining what you’re already doing - produces fast, clear, local feedback. Exploration - questioning whether you’re doing the right thing at all - produces slow, ambiguous, distant feedback. Any system that chases legible signal will naturally drift toward exploitation. You get better and better at climbing, and you stop asking whether you’re on the right hill.
This is the real risk. Not that you move too fast. Not that you move too slow. But that moving fast feels so productive that you never question the direction. The feedback is positive, the metrics are improving, the team is shipping - and none of it matters because you’re optimizing in the wrong space.
And the feedback loop can’t save you if the signal itself is bad. If you’re measuring the wrong thing, or getting feedback from the wrong users, or interpreting noise as signal, speed just makes you wrong faster with more confidence. The loop is only as good as the observation step - which means the quality of your attention matters more than the speed of your iteration.
And there’s a reason people stay on the wrong hill even when the signal is telling them to move. When you’ve invested months in analysis and planning, you’re anchored. You’ve built consensus around the plan. Changing course feels like failure rather than learning. The sunk cost is real, and it’s not just time - it’s emotional and political. The more deliberately you chose the hill, the harder it is to admit it’s the wrong one.
So the discipline isn’t about speed. It’s about the kind of attention you’re paying while you move.
Run
Here’s the other half of the argument, and the part I feel most strongly about: once you’ve found the right hill, speed is everything.
When you’re iterating on the right problem, for the right users, with real signal coming back - every cycle compounds. You’re not just learning anymore, you’re building on what you learned last cycle. The team that ships weekly on the right problem doesn’t just learn 50 times a year - each thing they learn makes the next cycle more valuable. The product gets better, which generates better feedback, which leads to better decisions, which makes the product better. It’s a flywheel.
And you can afford to run because most of the decisions you make while running are reversible. Bezos calls these two-way doors - you walk through, look around, and walk back if you don’t like what you see. Most code can be changed. Most features can be reverted. Most architectural choices can be migrated away from. The one-way doors - shutting down a product line, signing a long-term contract - deserve deliberation. But people consistently overestimate how many of their decisions are one-way doors, and slow down on choices they could have just tried and reversed.
This is where the “move slow, get direction right” advice does the most damage. Teams that have found the right hill but move cautiously - over-planning features, over-deliberating technical choices, waiting for certainty before acting - are leaving compounding gains on the table. They’ve done the hard part. They’ve found the hill. And then they walk up it when they could be running.
The best teams I’ve worked with have an almost uncomfortable bias toward action once they have conviction about the problem. They ship things that aren’t perfect. They make decisions at 70% confidence. They fix things in the next cycle instead of getting them right in this one. And they pull away from competitors who are just as smart but move half as fast, because compounding is exponential and the gap gets wider every cycle.
One honest caveat: speed creates its own irreversibility over time. Technical debt is the mechanism - every shortcut converts a two-way door into something a little more one-way, and they compound. Reversibility is an asset you can spend down without noticing. Run fast, but keep an eye on whether you’re burning the option to change direction.
The real discipline
Direction matters. The “move slow” advice is right about that. But you don’t figure out direction by thinking harder at the start. You figure it out by moving, paying attention to what the movement tells you, and being honest about whether you’re on the right hill - even when the hill you’re on feels productive.
That’s the hard part. Not the speed. Not the planning. The honesty. Being willing to look at a thing that’s working locally and ask whether it’s the right thing to be working on at all. Being willing to abandon a hill you’ve been climbing successfully because you’ve realized there’s a better one.
And then, when you find the right hill - run.