Many of the complaints here can be solved via a good implementation of infinite scrolling.
The URL should auto-update to reflect the current state of the content that has been loaded onto the page (the only exceptions are when the content is highly dynamic and pagination becomes somewhat meaningless over even short time intervals).
Either manual invocation of the scrolling should be necessary (so access to the site footer is not restricted) or else the site footer should be re-designed so it is still accessible (Twitter, for instance, just puts an information box on the left with links that would normally be in the footer).
It is also very important that clicking on a link on the page and then going back should restore the full state of the page, including scrolling position, at least to the greatest extent possible.
Ideally a visual separator should be inserted between different pages of content. This helps the user maintain context of where they are in the content stream.
Or, it could just be different pages of content and the back button actually works in every instance and the scrolling position is automatically (and accurately) preserved by the browser. As opposed to infinite scroll with updating URLs that inserts 10 different items in your browser history, breaking proper back button functionality to fit the view of what a designer thinks are all the pages you visited while you, as a user, just scrolled a single page and think the back button should take you back to the site you just came from.
You're describing poor implementations of infinite scrolling. Like I mentioned, in most cases it's possible to avoid all the issues you just described. Whether this is done or not depends really on how much effort the programmer wants to put into it. Also, just to be clear, this:
> As opposed to infinite scroll with updating URLs that inserts 10 different items in your browser history
is avoided with correct usage of the HTML5 history API. You can update the URL without inserting a bunch of extra pages into the browser's history.
I don't think I've ever seen an infinite scroll website implemented correctly. Nor have I ever seen one that works well on my phone. Which is probably why I generally never go back to infinite scroll websites aside from the social media biggies.
The URL should auto-update to reflect the current state of the content that has been loaded onto the page (the only exceptions are when the content is highly dynamic and pagination becomes somewhat meaningless over even short time intervals).
Either manual invocation of the scrolling should be necessary (so access to the site footer is not restricted) or else the site footer should be re-designed so it is still accessible (Twitter, for instance, just puts an information box on the left with links that would normally be in the footer).
It is also very important that clicking on a link on the page and then going back should restore the full state of the page, including scrolling position, at least to the greatest extent possible.
Ideally a visual separator should be inserted between different pages of content. This helps the user maintain context of where they are in the content stream.