Web servers however aren’t smart enough to decide without fail who the recipient of any given copy will be. You can apply heuristics to make sure a large enough number of people get included and excluded from receiving a copy, but you can never reach everyone. To give a simple example: you can try and block other servers from leeching your images, but only if the thieving party is honest enough to tell your web server who they are (Referrer).

We’ve tried that with 24 Oranges, and the problem is you either need to build a blacklist (which with a popular site can get bloated), or assume everybody is a bad actor and accept that some good actors will be blocked.

In other words, most webservers will serve copies to anybody who’s asking.

But publication is not the same as copying, and I am not sure why you assume the same rules have to apply. To get back to your example of a picture book where the user has to provide the photos: a website with embedded videos is more like a picture book where the picture book provides the photos.

In the cable TV rulings by the Dutch Supreme Court the TV providers were republishing the works via cable. This I can understand: they retransmitted the works as they were picked up from the air. In the Poortvliet case Hovener redistributed the photos. If he had merely sold the calendar with empty placeholders for Poortvliet pictures, would he infringe?

Do I infringe copyrights if I publish a book with boxes where you can place stickers of football players? In my view you only infringe if you sell the stickers as well.