While you're correct to think that latency does indeed matter, I think what the article is trying to point out is that very few services are latency sensitive b

While you're correct to think that latency does indeed matter, I think what the article is trying to point out is that very few services are latency sensitive beyond what is traditionally expected. In many applications, there are certain expectations of performance around certain actions, sure. That said, if things go long the user may get annoyed but they won't be deterred. But in some cases, having anything but the highest regard for latency is not only a deterrence, it's basically completely unacceptable to the user. Sure, you want your blog to load fast. But if your blog takes 300ms or 900ms very few people are going to notice the difference despite the 3x disparity. You're unlikely to get a nastygram on Twitter about your blog being unusable. Hell, even websites that take greater-than-one seconds to load are routinely not taken to task by the average user because, quite frankly, while fast is nice it's not a need. But apply that to a competitive, real-time game and if your service takes 70ms RTT for one of the roughly several hundreds of thousands of commands issued to it over the course of a 20 minute span you're performing about average. At 250ms - dramatically less than the kind of latency bound available to a blog post - you're rapidly approaching "completely unplayable" and going to get nasty responses.