Q: If I wanted to go back to the "root view" of level 2. Is it intuitive enough to just be able to UNSELECT the left pane item that I previously selected? As users, would you expect to be able to unselect that row?
A: Maybe not. I suspect deselecting is something users do when they’re thinking of canceling something, not executing something (e.g., get meta-data), so it may not be the first thing users think of. On the other hand, it helps that the page opens with no selection, so many users may notice that (a) no selection is possible and (b) non-selection shows the meta-data. But that’s asking for a lot from the user. Also, deselecting can be difficult because there’s no explicit target to deselect things –the user needs to find “nothing” to click on. I know I’m been frustrated with Windows Explorer in trying to get the file list pane (second level) in a non-selected state in order to get the folder statistics in the properties pane (third level). Here are two alternatives:
Include an explicit selectable summary item in the second level list at the top. Being no bigger than any other item, it shows the most critical meta-data. Clicking it populates the detail pane with all meta-data. Such a summary item may also be used to select all items for a single action like move or copy (which may or may not be a good idea, depending on how often it’s needed). There’s always a selected item in the item list. It’s just that the “item” might be the summary item. You might find that, with less use of gratuitous graphics, you can actually fit all the meta-data in such a summary item and you don’t need to worry about showing meta-data in the detail pane.
Put the meta-data in its own pane. Put both the item detail pane and the meta-data pane in respective expanders so the user can show one, both, or none. This gives the users a lot of flexibility (e.g., they can see and compare the meta-data without losing their selection). Always have at least one item selected so the detail pane is always populated. Default arbitrarily to the first item. This helps your users realize that the items are selectable and that they can get item details by changing the selection. Depending on how useful the meta-data is, it might even make sense to make the meta-data pane be the second level (top level on page), and the item-list pane is a third-level detail of it. The item detail becomes a fourth level.
Q: The "level 2" in the crumb could also be clickable to reset the view, but I don't think that is a good way to get back to that root view. I guess the browser back button could work as well?
A: Users may reason that clicking on Level 2 will bring them to Level 2, which they may not do because they realize they’re already on level 2. Like clicking nothing to deselect everything, clicking Level 2 does not explicitly say “give me meta-data for all of Level 2.” Compounding that is that users don’t use breadcrumbs much even when they should. They may look at them for orientation, but it doesn’t occur to them to click them. The Back button is no substitute. Users expect it to move them back in pages, not items selected. And that’s how it should work. You don’t want users having to click back 15 times to cycle back through 15 selected items just to get back to the previous page. For that matter, you don’t want them having to click back 15 times to get back to the meta-data. Think of Back as the Close button, not the Undo button.
Q: I have a select box on items for bulk selection 1 to many actions on items (move, copy, delete). Should selecting the row select the checkbox as well? If it did that unselecting that item would be visible, just unselect the checkbox.
A: Probably not. A checked checkbox implies that clicking another results in two selected items (checkboxes are n-of-many selection control). Users may think they have to uncheck the current item before clicking the next item and never realized that it’s unnecessary. Mixing click-selection-box selection (n-of-many) with click-item-to-select (one-of-many) is an interesting idea, and I’d like hear how it turns out in a usability test. Having selection boxes is probably much more discoverable than expert shortcuts for multi-selection like shift-clicking, ctrl/command-clicking, and dragging. But will the selection boxes be such a strong cue that users assume it’s the only way to select? Will they never discover “normal” selection? What happens when the user clicks one selection box, then clicks another item (not selection box), does the check box uncheck? What will users expect? I don’t know. If executing the same action on multiple items is rare (usually the case), it may be best to have no selection boxes, and rely on the expert shortcuts. It’s okay to have only low-discoverability expert shortcuts because you’re not relying on them to accomplish the task –they’re just a short cut. Non-expert users can carry out the same task without multi-selecting by selecting and acting on each item one at a time.