Something to work on for the next four years

I guess it's become about spirals.

"Pure" spirals (i.e., with one center point) are difficult because they are intrinsically figural: sea shells, ram's horns, etc. The tendency is to escape from this by becoming looser and more expressionist. However, in the context of architecture, such moves are unsatisfying because they are formally shallow even if they are content-deep. The initial investigation has been about escaping this dilemma by corrupting the construction of the pure spiral.

The above spiral was achieved over several steps with minimal means. The initial mesh was soft-selected. Then, at each step the center of rotation was moved, then the selected mesh was rotated a certain number of degrees clockwise or counter-clockwise. The faces closest to the selection are rotated more strongly than faces farther away, which "lag" and form echoes of the previous loops. At step 8, above, the mesh was "unwound," tightening older loops into hyperbolic points.

The exact steps were:

1. (initial)
2. move to A, rotate clockwise 533 degrees.
3. move to B (z+3), rotate clockwise 533 degrees.
4. move to C (z+3), rotate clockwise 514 degrees.
5. move to E (z+6), rotate clockwise 233 degrees.
6. move to F (z+3), rotate clockwise 265 degrees.
7. move to G (z+3), rotate clockwise 164 degrees.
8. move to I (z+6), rotate counterclockwise 610 degrees

Different degrees of rotation, sequences of center points, etc. produces immense variation in the size, proximity, and "tightness" of loops. For example, adding 1 degree of rotation to steps 2-4 above produced the following (original in red, variant in black):

Below are several superimposed series of spirals, each series consisting of three meshes initially spaced 10 units apart, and the only difference among series being the degree of counterclockwise rotation at the final step.

The interesting thing about these spirals is how a single spiral forms a series of interior and exterior spaces simply by intersecting itself:

So, I think the final house might be formed by taking the above primary spiral (red) and combining it with parts of several secondary spirals (black and blue).

The main problem with the spiral technique is that it creates extremely distorted meshes that are impossible to work with. But, the side effect is that these meshes have an interesting texture when rendered--a pure expression of "Maya tectonics" i.e. the formal characteristics of Maya's tools and geometry.

For about a week I was trying to produce the "right" mesh by experimenting with the variables, but unfortunately that approach is insane and I've decided to approximate it by tracing the spirals with NURBS curves and then lofting--NURBS curves being something I, at least, can edit/design. The potential benefit here is I can add a second level of geometric complexity in elevation, such as these hyperbolic corners:

I've tried faking the mesh-distortion effect by hand (by pulling on edge segments), which doesn't look as good as the real thing but does begin to bring back some aspects of my earlier "school of fish" texture.

No floor, no ceiling, no roof, no openings, no house.

Sketch plan: