[simonw]

You're right. Here's the full context:

> At one point, an ebullient engineer had declared the entire rewrite could run on two servers and, our minimalist QA environment being much larger to the contrary, we got remarkably close to launching with two servers as our most accurate estimate. The week before launch, the capacity planning project was shifted to Rich and I. We put on a brave farce of installing JMeter and generated as much performance data as we could against the complex, dense and rapidly shifting sands that comprised the rewrite. It was not the least confident I’ve ever been in my work, I can remember writing a book report on the bus to school about a book I never read in fourth grade, but it is possible we were launching without much sense of whether this was going to work.

> We had the suspicion it wouldn’t matter much anyway, because we weren’t going to be able to order and install new hardware in our datacenters before the launch. Capacity would suffice because it was all we had.

Given that description I'm not ready to blame the capacity problems exclusively on Will!