[Max-q]

Ignoring performance is what gives you slow code, costing you a lot if the code you write will be a success because you have to throw a lot more hardware at it. Think back to early Twitter that crashed and went down often hours each day.

Most optimization will improve on all or some VMs. Most will not make it slower on others.

If you write code that will be scaled up, optimization can save a lot of money, give better uptime, and it’s not a bad thing, the better code is not less portable in most cases.

[skybrian]

Sorry, I phrased that badly. By “ignoring performance,” I meant something more like “writing resilient code that can handle swings in performance.” For example, having generous deadlines where possible. Performance is fuzzy, but fast code that has lots of headroom will be more resilient than code that barely kees up.

[wavemode]

To play devil's advocate: I don't think, in hindsight, Twitter would choose to have delayed their product's launch in order to engineer it for better performance. Given how wildly successful it became purely by being in the right place at the right time.