A very naive materials-engineering approach. Materials are very quirky.

Carpenters must think in terms of "too long not good, but too short terrible." Wood is not isotropic, not in strength, flexibility, thermal growth, or brittleness.

Metal formers think in terms of "shape it once; reshaping can induce fatigue or unpredictable warping". Unfatigued metal is completely isotropic.

Tailors work with a material that is not only not isotropic, it's not even linearly inhomogeneous. It can be stretched, especially along its diagonals, but sometimes more along weft than warp. If you iron it first its dimensions will change. Ditto if you wash & dry it, or if it is mercerized and then washed after cutting. Most importantly, we construct 3-dimensional curved structures out of it, which end up looking like straight cylinders at a glance (spoiler: they never are!). Oh, and the junctions of these not-cylinders are never right angles, nor even straight lines.

Now, let me begin my thesis on why 2-D pictures will never produce accurate measurements of a 3-D object...

160k over three years. That’s like 1 or 2 junior devs and zero hardware budget.

The fact this failed is not surprising, I’d be amazed had it succeeded.

This business can't be that important to this guy if he let his "post mortem" be written by an LLM.