Well, it's a complicated question, and I'm not that well educated.
To my best knowledge...
1. You can't make a single cent if your bot visited facebook.com 100 million times. You can make some serious money if your bot visited some-exciting-domain.com, which belongs to you, and there was 5 ads displayed on each visit.
With this incentive you have all the reasons to make your bot very human-like (think headless chrome, realistic mouse movements, having old cookies, etc) so fighting fraud gets extremely hard.
It's easier to serve the ads and let advertiser figure out anti-fraud measures by himself. Being responsible for measurements and lack of fraud on open web is a huge PITA without clear path to huge uplift in revenue.
2. Facebook optimizes UX (or claims to), and calls to other servers make site slower, especially on mobile, lowering user engagement. This argument obviously does not work for some-exciting-domain.com. So, you can call whatever your want from your ad on some-exciting-domain.com, but on facebook.com you play by facebook rules.
In fact, some-exciting-domain.com can probably ban ads which call other domains, but it will just kill his revenue (programmatic systems will label him as "non-performing", because nothing is properly measured and stop buying ads there).
To my best knowledge...
1. You can't make a single cent if your bot visited facebook.com 100 million times. You can make some serious money if your bot visited some-exciting-domain.com, which belongs to you, and there was 5 ads displayed on each visit.
With this incentive you have all the reasons to make your bot very human-like (think headless chrome, realistic mouse movements, having old cookies, etc) so fighting fraud gets extremely hard.
It's easier to serve the ads and let advertiser figure out anti-fraud measures by himself. Being responsible for measurements and lack of fraud on open web is a huge PITA without clear path to huge uplift in revenue.
2. Facebook optimizes UX (or claims to), and calls to other servers make site slower, especially on mobile, lowering user engagement. This argument obviously does not work for some-exciting-domain.com. So, you can call whatever your want from your ad on some-exciting-domain.com, but on facebook.com you play by facebook rules.
In fact, some-exciting-domain.com can probably ban ads which call other domains, but it will just kill his revenue (programmatic systems will label him as "non-performing", because nothing is properly measured and stop buying ads there).