The article is flawed, because for example it takes for granted that the footer needs to scroll with the infinite scrollable content pane, which is just an absurd thought to begin with. Similar statements hold true for memory management, of course a naive implementation will eventually fill memory, but the same is true for a bad implementation of an single page web app with paging UI. Same for bookmarks, of course if you bookmark page 1 of a multi page list, eventually the articles you saw will be pushed to page 2. And if you have a single page application and forget to have the url change to the articles you read, you will have issues, on the other hand if you read an article in an infinite loading list, it would be no problem to change the url as you read articles and make them linkable. If you take away all of these statements you end up with the following: a) it is easier to mess up infinite scrolling than to mess up paging, don't think its easy or you can build it quickly 'on the side'. (my take on that point is) -> We need better best practices and tooling to build infinite scolling for popular frontend frameworks. In my opinion it is on the level of complexity as building a state library such as redux including the react integrations and redux tooling and needs a similar amount of contributors, support, application feedback, bug reports etc. until it can provide a satisfactory mainstream user experience. b) the article traps into the same mistake most medium/hackernoon articles make: not more explicitly discriminating between real web apps and web pages that just happen to use a js framework. A real web app is something like calender, mailclient, taks, slack, fb messenger, trello etc. They have completely different requirements and expectations and it would be absurd to click to older pages of messages on slack and is really annoying to do so in gmail. Even though the article just mentions the term "web site", it does sell its advice as a kind of universal truth independent of context. For a web page, something like new york times or medium articles or some informational page for a company offering, i agree that especially autoloading of next articles at the end of one opened post can be really annoying and in the context of articles infinite scrolling is even easier to get wrong, but not because of an universal law, but because the user interface has more things to take into consideration and requires better planning.