The first two do not have low latency high bandwidth connections (or often any connection at all) to the user to collect data, so it's not a reasonable comparison. As for the third, I understand that duck.com claims this, but I don't buy their claims, considering their behavior. None of these sites that claim to not collect user data should be trusted until they submit to a transparent audit that exposes all their infrastructure. As for the fourth, this is a tiny exception to the rule held up by some individual with standards.
Apple doesn't fit the pattern of any of your examples.
Had anyone actually substantiated the claim about the unconfirmed non-disabling device ID beyond a single packet and theorizing about encrypted packets?
Because the whole article is based on this evidence and peoples’ desire to dog pile on apple.
* bill boards and similar
* printed media ads
* ads that respond to keyword searches - without relying on any other user data
* display ads on topical websites (e.g. guitar ads on guitar related websites) - without relying on any other user data