The received wisdom of the last two decades of product development was built on a single premise: that building things is slow and expensive, so the discipline lies in building less, sooner, and learning from it.
Everything followed from that premise. The minimum viable product exists to reduce the cost of being wrong. Iteration exists because the first version cannot be right. Speed to market is an advantage because the competitor is also slow.
The premise has weakened considerably, and the conclusions have not been revisited.
When everyone is fast
A functioning prototype that once took a team a quarter can now take a competent individual a few days. Interfaces, copy, imagery, integrations and plausible demonstrations of a concept are all available at a fraction of their former cost.
The first-order effect is the pleasant one: more can be tried. The second-order effect is the one that changes strategy. If building is cheap for you, it is cheap for everyone, and speed of execution stops distinguishing anyone. An advantage that every competitor can obtain by adopting the same tools is not an advantage; it is a new baseline cost of participation.
Meanwhile the constraint has moved to the step that did not get cheaper. Deciding what is worth building requires understanding a problem well enough to know which version of it matters — and no tool has made that faster.
The cost that did not disappear
There is a further asymmetry that firms are discovering late. Building became cheap. Maintaining, supporting, explaining, securing and eventually retiring what was built did not.
Every feature shipped is a permanent obligation: documentation that must stay true, an interface that constrains the next redesign, a support burden, a security surface, and a promise to whoever came to depend on it. Those costs were always the larger share of a product's lifetime expense, and they are now a much larger multiple of the cost of creation.
The practical consequence is that cheap building makes over-building more dangerous, not less. A team that can produce twenty features a quarter and evaluates them on the cost of production will produce twenty obligations and mistake the fact that they were easy for evidence that they were wise.
Testing more, learning less
There is a subtler failure that follows from abundance. When experiments are expensive, teams choose them carefully, and a chosen experiment tends to be one that tests a belief someone actually holds. When experiments are nearly free, they proliferate, and most of them test nothing anyone believed in the first place.
The result is a great deal of measured activity producing very little knowledge: hundreds of variants tested, small effects detected, and no improvement in anyone's understanding of why customers behave as they do. Learning requires a hypothesis worth being wrong about, and hypotheses did not become cheaper.
What becomes scarce
If execution is commoditised, the differentiators are the inputs that resist it.
Knowing which problem matters — the kind of understanding that comes from prolonged contact with customers rather than from a research summary. Judgement about what to refuse, which is the only defence against an infinite feasible set. And the willingness to work on something that takes longer than a competitor's imitation window, because a project that can be copied in a week was never a strategy.
The uncomfortable version of the argument is this: a firm whose distinctive capability was shipping quickly has just had that capability distributed to everyone, for a subscription fee. What it does next depends entirely on whether it also had a view about what was worth shipping.