In 1969 the astrophysicist Fritz Zwicky published Discovery, Invention, Research Through the Morphological Approach, which laid out a procedure he had been using for decades on problems from jet engines to telescope design. The procedure is almost insultingly simple, and that is the point.
The method in four steps
First, state the problem precisely. Second, identify the dimensions that any solution must take a position on. Zwicky called these parameters; we call them axes. For a propulsion system the parameters might be energy source, medium of operation, thrust direction, and so on. Third, list the possible values along each dimension. Fourth, and this is the move that makes it morphological, consider every combination of one value from each dimension. The full set of combinations is the morphological box.
Most combinations are nonsense. A great many will be physically impossible, illegal, or absurd. Zwicky's claim, borne out across his own working life, is that the survivors, the combinations you cannot immediately rule out, are the real candidate solutions, and that enumerating them systematically surfaces options you would never have reached by free association. The discipline is in the enumeration, not the inspiration.
Why enumeration beats listing
When you brainstorm a list, you generate the ideas you can already reach. The combinations you never think to write down are simply absent, and you have no way of knowing they are missing. A morphological box inverts this. It generates the whole combinatorial field first, mechanically, and then asks you to eliminate. Absence becomes visible because every cell exists whether or not you have an idea sitting in it.
The catch is that the full box is large. Five dimensions with four values each is 1,024 combinations, which no one can hold in their head or usefully stare at. Zwicky worked with cross-consistency tables to prune pairs of values that cannot coexist, but the raw combinatorics remain unwieldy on paper.
How Concept Space renders the box
We do not try to show you the whole box at once, because a 1,024-cell field is not something a person can read. Concept Space renders the box two dimensions at a time: you pick one axis for the rows and one for the columns, and you get a flat grid you can actually look at. Your existing projects land in the cells whose coordinates they match. The empty cells are the combinations along that pair of axes that you have not occupied.
Swapping the row or column axis is how you walk the larger box. Audience against tone is one slice; audience against delivery mode is another; tone against effort is a third. Each slice is small enough to read, and rotating through them lets you traverse a five or six dimensional space without ever facing more than one rectangle of it. The grid is not the whole morphological box, then, but a series of honest two dimensional sections through it, which is the only form in which a box that size is legible to a human reader.
The AI proposals follow the same logic. When you click an empty cell, the candidate it proposes is constrained to that cell's coordinates: it has to fit the row value and the column value you are looking at. That constraint is exactly Zwicky's discipline, applied one combination at a time. You are not asking for an idea in general; you are asking what would have to be true for something to live at this specific intersection, given the projects already sitting next to it.
That is the whole mechanism. List your dimensions, give each a few values, look at the combinations a slice at a time, and pay attention to the cells nothing has reached yet.