TALAMANA · THE AI LITERACY MAP FOR ARCHITECTURE AND DESIGN · Judgment · AGE 18—21 · POSITIONAL · HELD
Words for critiquing output
A fault you can name is a fault you can fix.
Our position
"I like it" and "it feels off" are not yet critique. A feeling can start a critique; it cannot finish one. Ask what in the work produced it, and which criterion it breaks. Critique of generated output needs words that carry a fix: the light comes from nowhere · the scale is unbelievable, that door is two and a half metres · this material is wrong, laterite does not do that at a corner · the caption flattens the design into a style. Precise critique is what makes the next cycle possible, because the next brief is written out of it.
> A fault you can name is a fault you can fix. A mood is just weather.
Why we hold it
The studio crit has always worked this way. The useful juror names the fault and the useless one reports a taste. Generated output raises the stakes, because the next cycle is only as good as the words that describe what was wrong with the last one. A vocabulary of fixable faults (light, scale, material, structure, story) is the Lab's working critique register, and it is what lets a markup become a brief.
The strongest objection
Some of the most important judgments do not break down into named faults. "This is the wrong building for this street" is true, can be acted on in the largest sense, and cannot be reduced to a light or a scale. A vocabulary that demands a fixable fault for every objection may train students to see only what the vocabulary can name, and to wave through a coherent, well-lit, believably scaled image of the wrong idea.
What would make us revise it
The vocabulary already includes idea and story as faults that can be named. "The idea is missing" is a sentence. We would revise the position if we saw critique-trained students becoming fluent at local faults and blind to the whole: cohorts whose crits got sharper at the detail and weaker at the scheme. That can be checked at jury, and we check it.
Try it
Critique one output using only points a peer could act on in the next cycle. Then take three taste-verdicts ("nice," "off," "boring") and replace each with a named fault that can be fixed. Notice which of the three was hardest to convert. That is where your eye is least trained.
Take it to crit
In crit, does the student name a fault the maker could fix, or only report a feeling? When they say "it doesn't work," ask: what, exactly, and what would you change?
How it works
The Lab's critique register (light, scale, material, structure, idea, story) is its own working vocabulary, developed in teaching and kept as internal material. The words on this card are drawn from it. The register works because each word points at a layer of the image that can be checked on its own, and because each one maps onto something the next brief can say. Taste-words point at the viewer. Fault-words point at the image.
What this idea builds on
What this idea opens up
- Nothing yet names this as a foundation.
Sources
- collaborative-red-pen
- judgment_register
- What Counts as Architectural Thinking After AI
Open this idea on the map · The complete map · Logika · RBDS AI Lab, India · revised every edition.