Hacker News Reader

Comment by logfromblammo | Hacker News Reader

And the professional digital modeler that just found a way to cut 20 triangles out of the trunk wireframe just rolls their eyes and goes back to fine-tuning the skeleton so that the active forelimb doesn't pinch in the "skull-crush execution" animation....

The described technique is interesting, but elephants are 3-dimensional objects with a somewhat more detailed contour, so I'm going to have to declare that the well-known saying remains unimplemented. I think Dyson could have retorted, "Yes, but my model will be finished long before you have discovered your fourth parameter."

View on HN

> Yes, but my model will be finished long before you have discovered your fourth parameter

This attitude is one of the issues I encounter among data scientists which limits their impact in business to lower ticket decisions (like individual recommendation systems) vs. bigger budget allocation decisions in business and the same is the case with physics. Being able to explain drivers and protecting against catastrophic overfit failure is much more important than getting a great predictive fit in a limited dataset.

Well, if they can use complex numbers, I think quaternions[0] are fair game for a 3D elephant.

[0]https://en.wikipedia.org/wiki/Quaternion

That seems like cheating. If you can use quaternions to squeeze out more degrees of freedom, you could also use octonions. You could use any number of additional dimensions that square to -1.

We lose features too, as we climb in complexity. Lucky 10000: https://en.wikipedia.org/wiki/Cayley–Dickson_construction

but if you want a normed ring division algebra you must stop at 8 dimensions.

I thought those things weren't so easy to come by -- that was why Hamilton was excited enough to carve it into a bridge wasn't it?

I think the difficulty is actually in arranging them such that they combine to make an interesting algebra.

Reply on news.ycombinator.com