For those interested in numbers, here is a table of numbers detailing the time savings per year by milliseconds page load time:
Looking at worldometers.info we see that the world has passed two billion Google searches by tea time EST (~4:00 p.m.). Using that as a ballpark, and multiplying by five to estimate the number of page loads per day, we arrive at an estimated 10 billion page loads per day. This gives us something to work with. We are not concerned here with accuracy at this point, but simply at finding something that we can viscerally relate to.
Looking at only a one hundred millisecond drop in page load time, we see that works out to approximately 30 years worldwide per day. If an average of an entire second were shaved off of page load times worldwide, the total time gained would be about 300 years, per day. Since the advantage gained drops off past below about 300 milliseconds (which we are near achieving), it seems doable that only starting at an average of a 1.5 second page load and going down to the physical minimum we have determined so far (~300ms), we could collectively gain almost four hundred years per day.
Putting this together with information gleaned from elsewhere, we should also see a shift from dissatisfaction to satisfaction with page load times in specific, and with web sites in general. Even looking at this site, I have to pay attention, or I wouldn’t even notice the page has changed. This is the effect we are after, that we have to think about adding back in a transition effect, rather than the other way around.