you’re making a lot of interesting points in your comment. We agree on most of it – the part about the speech recognition and AR is maybe a bit too far away. We believe that indexing is useful but often not precise enough to tag objects, therefore we’re relying 100% on the user’s activity to tag the files (tagging themselves + auto-tagging based on name/location and more….).

While writing this I’m checking your blog – amazingly interesting stuff! What we believe is that (at least for the next years) semantic can go up to certain level in “understanding”, and thus categorizing contents, therefore “manual” categorization will (always) be needed…

Anyway, we’d LOVE to hear what you think of our GUI and the solution we found to make Tabbles as straightforward as possible. The GUI is indeed crucial for our project, cause among the main challenges was to produce a “2 dimensional file-manager” – meaning that relations can be both vertical (father/child = container/containee) and horizontal (needed for objects with similar characteristics but not in the same folder) – where file-managers are traditionally “1 dimensional”.
Tabbles does now allow you to navigate both vertically and horizontally with the exact same UX: it dynamically generates a tree of tabbles (virtual folders) depending on what you’re looking at, and placing the static children (objects contained) and the dynamic children (objects related but far away) on the same level.

Well, enough babbling for today I guess – we’d be really happy to hear your opinion and to send you a license or two (please drop us a line using the contact form on tabbles.net!).

Thanks,

Andrea

They all have a notion of folder/container that has some content inside. You need to remember what this container has inside without opening it, and you have to remember it’s exact location in other container/containers.

Automatic tagging involves semantic computing, which itself turns into the need for object-relational DB where you could be able to store information about relationships between information entities that you should extract from stored data. And to capture these entities, you need to make indexing. Essentially, WinFS was exactly that thing – object-relational DB with indexing + processing of relationships extraction etc. Though internal organization of that system was not perfect (it was a belief that actual entities should be stored, without notion of context in which they are to be used – “one schema for all”). Anyway, having semantics extracted from your personal/web data is not enough – you need then to build an entirely new User Interface on top of it; and here comes the need to make that UI as much natural as possible. That should involve touch, speech recognition, gestures, maybe augmented reality effects.

And there is a lot of work to be put into this area, to make the turn to the Non-Desktop Metaphors.

Thank you for submitting this cool story – Trackback from DotNetShoutout…