undefined | Better HN

0 pointsangersock9y ago0 comments

Like, what special sauce do you bring to the table that ANSYS or Solidworks or whatever FEA package people use for optimization studies doesn't already do--and probably do well enough?

You dismiss twitter, snapchat clones, whatever when they do deliver actual value to people just because they don't meet some arbitrary feelgood measure of "innovative". A measure that coincidentally happens to somehow make your cute little tool worldchanging! Industry, especially in mech and AEC, is hilariously slow to change workflows once they have something that works. Your product is probably going to get lost in the noise, all with you puttering around going "but but but I was innovating! Why don't people support innovation?!".

You're coming off as a holier-than-thou "srs bsns" engineer when you should have some humility. Maybe post under your real account.

0 comments

fjdhcjjenxncnd9y ago

It's still very much a research implementation, but I'm making global black box combinatoric methods which can be used where analytic optimization methods are not available, e.g. multi-objective optimization.

>Maybe post under your real account.

I have no real account. I always make a new one every few posts partly for privacy reasons, and partly because I enjoy reading posts like your own. :^)

arcanus9y ago

As someone who did numerical optimization for my doctoral thesis (also mechE), this sounds more amenable to publications than a start up.

Are you trying to commercialize?

mappu9y ago

> I have no real account. I always make a new one every few posts partly for privacy reasons

This is the logical next step after random site-specific passwords. The login/logout workflow does make this inconvenient though. Browser extension idea?

Chan boards use thread-specific IDs or thread-specific tripcode salts to achieve much the same thing - privacy by default, but identifiers when it helps you follow a conversation.

j / k navigate · click thread line to collapse