The Potential Merits of an Itemized OS by Alexander Obenauer

Each item is some data that can be consumed by a view (not necessarily a visual interface component, but in this article we will stick primarily to that case). (View Highlight)

items are universal reference containers for any number or type of other items. (View Highlight)

In this way, the basic structure of data that a user stores in their system is an item graph. (View Highlight)

browsing context displaying opened items side-by-side, in a horizontally-scrolling path. The user is able to work through their “train of thought” visually, on screen. This browsing context is specifically called a browsing path, or simply, a path. (View Highlight)

When the user has a new train of thought, they might begin a new browsing path. Should they want to return to a past browsing path, a recent paths list displays the items in each recent path, allowing the user to pick back up where they left off. (View Highlight)

Items are rendered by views, with more or less detail depending on whether they are being viewed alone or within a broader context.
The user is able to choose a view as the default for all items of a certain type, and they are able to quickly flip between applicable views on any open item. (View Highlight)

Items can be shared with other users, alone or in entire subgraphs. Since view definitions are items as well, a user can share some set of their items, along with their custom views, to allow others to browse within an environment of their own design. (View Highlight)

Choosing new views for certain item types does not affect the service used to bring that data into the system; the two are separate, and developers can independently provide a new view for a certain item type without having to replicate the implementation of the service used to bring that data into the system. (View Highlight)