The Floor, the Ceiling, and the Template

I was listening to Lulu Cheng Meservey speak with David Senra about originality, beauty, and templates. Her criticism of templates landed immediately.

As I heard it, the complaint was not that templates save time. It was that they save time by treating work as interchangeable. The client has particular circumstances. The service should carry the judgment of the person providing it. Force both into a standard form and you may save twenty minutes while quietly devaluing the client and the service.

I agreed wholeheartedly.

Less than twenty seconds later, I turned back to my own work and started vibe coding the creation of a template.

That reversal was too fast to excuse. I had not changed my mind. I had discovered that I held both positions at once.

Templates remove something human from work. Templates are also how work scales.

The useful question is not whether templates are good or bad. It is what they do to the range of possible outcomes.

A template controls variance

A template raises the floor and lowers the ceiling.

It raises the floor because the basic structure is already there. Important elements are less likely to be forgotten. Work becomes easier to review, repeat, teach, and improve. A weak performance is pulled toward an acceptable result.

It lowers the ceiling for the same reason. The structure is already there. The person doing the work is encouraged to fill the available spaces instead of asking whether those spaces belong. A strong performance is pulled toward the same acceptable result.

This is the bargain behind commoditization. Classification makes scale possible. To process many things, a system must decide that they are similar enough to receive similar treatment. The wider the class, the more variance the system removes.

That is valuable when failure is expensive and exceptional quality is not the point. It is costly when distinction is the point.

A template is therefore not merely a productivity tool. It is a decision about which tail of the distribution matters. Are we trying to prevent the worst outcome, or preserve the possibility of the best one?

Most organizations template by default because embarrassment is easier to measure than missed excellence.

A standard is not a template

This argument appears to conflict with The Standard, where I argue that reliable improvement requires explicit criteria.

It does not.

A standard says what must be true. A template says what form the answer should take.

Those can support each other, but they are not the same. A standard for a proposal might require a clear problem, an honest account of risk, a defined outcome, and evidence that the proposed work can achieve it. A template can help ensure those matters are addressed. But if the template also determines the order, emphasis, language, and shape of every proposal, it begins to answer a different question: not whether the work is good, but whether it looks familiar.

The distinction matters because standards can preserve judgment while templates can conceal its absence. A named person can apply criteria to an unusual case and explain why the result passes. A form can produce a familiar-looking result without anyone deciding whether it fits.

Standards make defects visible. Templates make variation manageable. Confusing the two is how an organization mistakes consistency for quality.

Template the invariant parts

The practical answer is to decompose the work.

Some parts of a service genuinely do not change from client to client. Recreating them is not craftsmanship. It is charging repeatedly for reinvention. Those parts should usually be standardized, automated, or templated.

Other parts carry the reason the client hired a particular person. They depend on context, judgment, taste, or a view of consequences. Those parts should resist the template.

This is the boundary most template debates miss. The anti-template position treats every repeated structure as dehumanizing. The pro-template position treats every variation as waste. Both positions avoid the harder responsibility: deciding which parts of the work are actually distinctive.

That decision cannot belong to the template.

It belongs to the professional who puts a name behind the work. In Work Assignments as Contracts, I argue that an assignment should make intent, constraints, support, and acceptance clear. The same discipline applies here. A template may carry the stable terms. Someone still has to judge the actual matter and own the division between what can be reused and what must be considered anew.

If that person cannot identify where their judgment enters the work, the template may be revealing something uncomfortable: the service was already a commodity.

Beauty lives above the floor

This is why the question belongs in the Artistic Revolution series.

In From Function to Taste, I argued that functional success is not enough. Human work also carries delight, meaning, and care. In Studying Taste, I went further: taste matters precisely because it resists standardization.

A template can prevent ugliness. It cannot guarantee beauty.

Individuality creates the possibility of both. Once a person leaves the safe structure, the work may become excellent, or it may collapse. The same freedom that allows a distinctive proposal, product, article, or experience also permits incoherence and failure.

The template eliminates part of that risk by eliminating part of that possibility.

This does not make templates hostile to beauty. It makes them a floor beneath it. The problem begins when the floor is mistaken for the finished room.

AI changes the old bargain

Historically, bespoke work was expensive because variation required human time. Templated work was cheap because the cost of producing the next instance approached zero.

Generative systems weaken that trade-off. As The New Baseline shows, the cost of producing a coherent first instance has fallen sharply. It is now possible to generate work that appears bespoke at a scale previously reserved for templates.

But appearance is not distinction.

AI-generated work may be the most templated artifact ever produced: personalized at the surface, statistically averaged underneath. A thousand outputs can differ without any of them carrying judgment about why the difference matters.

So the template does not disappear. It moves.

The old template was a visible document with blanks. The new template may be the distribution of patterns from which a model generates. The responsibility remains the same: a human must decide which parts can be generic, which parts require context, and which differences deserve to survive the system's pull toward the average.

Lower generation cost creates room for more individuality. It does not create taste, responsibility, or judgment automatically.

Choose the range you are willing to own

There is no virtue in refusing every template. There is no intelligence in templating everything.

Use templates where uniformity protects the work, where repetition adds no value, and where the floor matters more than the ceiling.

Break them where the circumstances are genuinely particular, where the client is paying for judgment, and where the possibility of excellence justifies the risk of variance.

Then own the division.

That is the part no system can decide on your behalf. A template is a tool for controlling outcomes. Choosing which outcomes to permit is an act of judgment.

Template what must repeat.

Judge what must remain distinct.