If you click on an item after significant scrolling, view the item, then click back, where are you? almost never where you were. Usually just at the top of the list.
The scroll bar and find function can be mitigated, but not as elegantly.
The real problem is that, when you load a page in the middle, it’s easy to scroll down, but scrolling UP is glitchy. That one can’t be fixed. https://meta.discourse.org/t/scroll-jank-when-scrolling-up/1...
I don’t think there is an easy answer here. All options have some trade offs.
I always thought the infinite scroll pattern was designed for analytics purposes. So that you can track user engagement. Not because technology constraints necessitated it or that it was a good user experience.
I love infinite scrolling for random content. There is no “position” so to speak. And there is no concept of “find” since you don’t know what you’re looking for. “Back” continues to work as long as your browser cached the previous page state.
Sometimes you really do want to simulate an infinite list of 4,000,000 entries!
An example: https://wikiscroll.blankenship.io
The item you wanted to search might get unloaded, or not loaded yet, in some bizarre attempt at optimization when the javascript library resposible is an order of magnitude larger than the actual content anyway.
My phone has gigabytes of memory. If you want to present five hundred items in a web shop, just list them and do not bother with any type of pagination, invisible or otherwise.