Didn't they recently publish a blog post saying "we have caches at every layer?" Yes:
https://signalvnoise.com/posts/3112-how-basecamp-next-got-to...1. I wonder how large their CPU time budget would be without all the caching.
2. Of course they're right. Being twice as fast is generally irrelevant, unless CPU is the bottleneck. We call it O(n), not kO(n), because the k means so little.
