I've actually planned to implement this myself. I know this is under the Episerver radar, and similar functionality exists in Commerce.

You and me both then :)

It's a race then hehe

Isn't that really an information architecture anti-pattern, though? Even though you COULD make a giant node load reasonably fast, it will not be a great way for the editor to work with content. 

You would have to couple that with a powerful content filtering (think PowerSlice) to make it usable, no?

Well, my solution involves UI as well. E.g. not listing news articles in the page tree and instead show them in a sortable, filterable and paginated grid. Maybe powered by Find, but not rely on it. Nodes loads fast in Episerver, the problem is the UI, the page tree can't handle thousands of nodes in one level. By having a paginated grid we don't have to load all nodes, which we have to do in a page tree.

This is already solved in Commerce. 

This is actually a feature I've been missing since I started working with Episerver back in 2008. Some content don't fit in a tree structure.

Yeah, instead of using a tree pattern you make it searchable etc. Works great for stuff like comments, press releases, products etc. 

I can of course use DDS for those but then I need to build a user interface for editor / admin etc. 

Sitecore has a similar concept for this called item buckets (read: page node where you can store 1000s of pages without messing up editor ui...).  And I do like the word "item bucketability". 

I understand the bucketing concept, but I still struggle to see when it's *really* the best choice to have hundreds/thousands of pages under one node, instead of introducing some kind of hierarchical folder structure relevant to the type of content (e.g. year-folders for news archive, category-folders for press releases, etc). 

Even with an hierarchy, you could omit the folders from the URL and achieve a "clean and short" URL. 

With bucketing, you would have to force a new UI concept into Episerver.
You would also have to make sure the bucketing is created in a way that's easy to export/migrate later on.

Not saying it shouldn't be done, just highlighting some of the challenges :-)

Behind the scene, thats how it's done Arild. So making it possible to automatically create subfolders that are segmented alphabetically or by date would probably be 90% of the solution. Sure, I can do this myself by reacting to the Published event etc...

Well the problem editors have, is to actually find the content. Especially when there are thousands of pages in the tree. Powerslice is not the solution either. My solution would also involve containers, but they would never be visible in the page tree. The content is still located in the tree, so no export/import issues. I don't see the URLs as a problem either, the main problem is the editing experience really.

I don't find a grid 'a new concept' though, nor does the editors. Some of them even preferres grids, especially for certain kind of content.

Looking forward to the protoype then :-)