I wouldn't call it "malicious," but way more than 50% of what you download is stuff you don't want. For example, a random NY Times op-ed has 6730 characters of text, bur just the initial request is 454kb, so 98.5% of what you downloaded is overhead. Some of that is formatting markup, but probably 95% of it is junk. And that's just the initial request, which in turn pulls in the images, tracking scripts, and ads. In the end, probably 99% of what you downloaded to read those 6730 characters is stuff you didn't want.