I wonder how much of Ferris' latency is related to net speed and rendering speed / computational horsepower of their local machine. I've seen several different evaluation regimes which try to objectively measure web-site "responsiveness" independent of net and rendering speed. Maybe the OP could use one of them to show how different sites were using techniques that unduly reduce the perceived performance of their site.

Think about what happens if you're on a site that loads hundreds of images and has keep-alive explicitly turned off. Or if you're doing a TLS handshake across a high latency network (TLS, for all it's benefits, requires at least one back-and-forth to setup the secure transport before sending "content.") Or you have a weird / inefficient dynamic loading process. Each of these will cause perceived latency, but have different solutions.

I tend to agree with the OP, it certainly seems like sites I use on a regular basis are laggy, but we may need a more objective evaluation framework than "ugh. the web is slow." And who knows, maybe most of the problem could be solved by getting the OP a better ISP and a faster machine.