I have not seen these things abused yet. The user agent prevents it.

There are already countless websites abusing the web notifications API to push spam to end users. You're probably savvy enough to effortlessly avoid it, but the average end user often is not. We can't keep ignoring the the UX issues with a lot of these web APIs if we want them to be successful. Users don't care to philosophize about the open web, they just want their shit to work. That's what Apple iOS is providing them, and they will continue flocking to iOS until they're presented with a better alternative.

Reasonable implementations let users easily block the notifications after receiving them. It doesn't require much savviness.

Of course you haven’t seen them abused yet, iOS doesn’t allow it. I remember developers abusing notifications in Android Cupcake.

I don't use iOS. I use only mobile and desktop OSes that allow these things. I still haven't seen them abused.

I really have no idea what you mean by your comment, but humorously enough user agents are frequently one of the most widely abused parts of the browser.

The "User Agent" is the browser itself the header just how a "User Agent" identifies itself to the server. That language really only shows up in the RFCs though.