When production gets easier, advantage shifts toward choosing the right problem, selecting what deserves attention, and judging whether the result fits its context.
It is now possible to produce a plausible first draft of many things in minutes: a product description, a page layout, an illustration, a working prototype. Yet the important decisions around that draft have not become equally easy. What should exist? Who is it for? What should be removed? Is the result polished, or useful?Production and value are not the same thing. Lower production costs create more options, but do not tell us which deserves attention. In some domains, the scarce input is shifting toward the ability to frame a problem, select among alternatives, and understand where the work will live.That is not a universal economic law. It is a practical pattern worth noticing, especially for founders and designers using generative tools.Abundance moves the bottleneckWhen a team can generate ten directions almost as easily as one, its bottleneck may become evaluation. More choices can reveal possibilities, but they can also hide weak thinking behind competent-looking output.Consider an explicitly hypothetical product team preparing a new onboarding page. A generative tool might quickly offer copy, diagrams, and several visual structures. The team still has to decide which concern a new user needs resolved first, which promise the product can honestly make, and which details can wait. If nobody can answer those questions, faster production only creates more material to review.The useful unit of progress is therefore not output volume. It is a resolved decision. A draft has value when it helps someone understand, choose, or act. A prototype has value when it exposes an important uncertainty. Production serves those ends; it does not define them.Taste is selection with reasons“Taste” can sound mysterious, as though some people simply possess it. I prefer a more useful definition: taste is the practiced ability to notice differences, select deliberately, and explain why one choice better serves the work.The Louvre photograph used with this essay offers a concrete example without requiring a story about how or why it was made. Diagonal beams cross repeated glass geometry. The escalator and people provide another strong line and a change of scale. Dark structural forms divide brighter areas of glass. These observable features direct attention; the photograph is not just an inventory of everything visible inside the building.That is what selection does. A frame includes some relationships and excludes countless others. Editing continues the process: keep this image, reject another, sequence it here rather than there. Generative tools can widen the field of candidates, increasing the importance of knowing what to keep.Supporting artwork: repeated fragments become a considered composition through selection. AI-generated editorial illustration.Context turns quality into usefulnessAn artifact is not valuable in isolation. The same clear explanation can be essential for one audience and distracting for another. The same prototype can be useful early and dangerous if mistaken for a finished system. The same image can feel precise in an essay and arbitrary in a sales deck.Imagine, again hypothetically, two teams producing identical technical guides. One publishes its guide when customers are encountering the problem, uses their vocabulary, and provides a clear route from the relevant screen. The other places the guide in a general archive under an internal term. The production quality is equal. The usefulness is not. Timing, language, placement, and trust shape whether good work reaches the people who need it.This is why distribution should not be an afterthought. Distribution is part of the design. Choosing a channel means choosing a context: what the audience knows, how much attention they can offer, and what claim they are prepared to believe. Judgment connects the artifact to that reality.Cheaper making is still real progressThere is a reasonable counterargument. Production remains difficult in many fields. Physical manufacturing, healthcare, infrastructure, and dependable software involve constraints that a quick draft cannot remove. Even in media and software, finishing, maintaining, and taking responsibility can be much harder than generating a first version.Lower barriers also have genuine benefits. More people can test an idea, express a point of view, or build a small tool without waiting for scarce specialist time. Exploration that once felt too expensive can become ordinary. New makers will develop judgment by making more, not by standing outside the process and discussing taste.I agree with that case. The point is not that making no longer matters, or that gatekeeping should replace craft. Cheaper experimentation changes where disciplined effort pays off. If producing another option takes seconds, generating options is less likely to be decisive. Asking a better question, seeing a meaningful difference, and accepting responsibility for the final choice become more important.A working practice for...