Page A loads content for X seconds and Z seconds loads ads. uBlock overhead is Y second but saves Z seconds by blocking ads. Total time with uBlock is X+Y-Z. If page has zero ads the time will be X+Y-0 which is greater than X+0.
Just to clarify, this is desirable because "fewer/no ads" is preferred over "faster page load times"?
Yes. Pages with ads are unreadable. I'm horrified every time I see friends navigate with no adblockers. I always recommend one. I'm going to start recommending Firefox again, as I did at the times of IE6.
It's also probably entirely random fluctuations in this case. uBlock doesn't add anywhere near that much latency normally.
Presumably because uBlock's baseline overhead is approximately 1 second. Therefore on other sites there is enough overhead from ads or tracking that it is still worth it to use uBlock.
So it has 1 second of startup cost? I have a hard time believing uBlock would add that much time to every request. Ideally this would make a more realistic measurement that doesn't include startup cost.
Or maybe it's just random variation.
Here's evidence[1] that uBlock adds less than 100ms to "an atypically large page from a nice site" (i.e. one that doesn't have any ads). Some other effect is at work here, it's probably just random network latency differences.