(Hi Joel, your last post and this one crossed like ships in the night!)
On reflection, what I found inhibiting about Tinderbox is that it seems to be necessary to have to press the spacebar after focusing on one of those cute little boxes (Notes) in order to read what's inside. Unlike PB, one can't hover over a TB note and see a label, or view its contents in a side panel. Please correct me if there are ways around this.
Without the presence of containers and adornments to signify where Notes really belong, TB immediately becomes more opaque than PB. This is the old, old problem of how to make the relationships between sets of objects visible at a glance. PB does this by making its Notes children of parents.
I agree with you that, despite all the proclaimed deficiencies of PB's links, linking is actually better developed PB than in TB. Coloring and thickening links, attaching comments to links (by hook or by crook!), can actually be achieved (better in Expanded view than Normal view, as I discovered), making the connecting of Thoughts by "lines of reasoning" more immediately apparent; whereas in TB, relationships are implied more by deciding which container or adornment they are in. Horses for courses.
This leads to an interesting dilemma, an offshoot of the "If your only tool is a hammer, everything looks like a nail" problem. If links are so well developed with their own comments, etc., that it becomes as natural to store relationship data in them as in PB Thoughts (TB Notes), would that be more intuitive than confining data to text fields in notes with just rudimentary links as TB and most concept mapping applications do? It's a hard one to answer. Directional links in PB can only be constructed with difficulty, which detracts greatly from their functionality, as jostber has observed. On the other hand, you have come to use links more sparingly - is that as a result of reading Mark Bernstein's book?
I can't see a correct answer here, but I can see that whichever application (even a piece of paper) we use for depicting our internal memory map can have significant feedback on the way we visualize it to ourselves. In a similar fashion, the words available to us in our native language constrain the way we conceptualize things. Sometimes it requires effort to resist thinking like a software application - too much like the tail wagging the dog.
At all times we should keep in mind the fact that PIMs, like paper, should be no more than aids to help us create larger maps than we can otherwise comfortably visualize. At least TB doesn't make crude attempts to confuse Brain and Thought with the way we naturally imagine things.
(Added since reading Joel's previous post)
Thanks for your response to my earlier post. Yes, I'm glad we seem to understand each other. Each application has its strengths and weaknesses. If Mark Bernstein ever decided to port Tinderbox over to Windows, I'd be in trouble riding two horses.
As you say, the crux of TB's design is the concept of the (isolated) Note as the "atom", whereas in Normal view, where PB began, it's the linked Thought. The enforced hierarchy that PB imposes may not be for everyone, but at least one can escape it to some extent in Expanded view.
My Expanded view demo is really just a static map and is pretty rough round the edges compared with TB's polished appearance. Also, the time I spent on it, (particularly crafting enclosures!) might be appropriate for presentations, but impractical for daily use unless Harlan comes up with some box drawing functions.
Thanks for shedding light on TB's development history. I suspect PB began life as an application for crime analysis, since I seem to recall early ads by Natrificial mentioning its use by forensic experts. It would be ideal for dredging up otherwise unsuspected links between events and crime figures; I bet it's still used for this purpose. Crime writers would find it invaluable.
TB lacks this associative background, though its Agent feature (shown in the screencast Planning) seems just the ticket for finding what's missing from a set of relations, similar to applying Boolean functions to tags in Reports filters in PB.
I look forward to reading The PersonalBrain Way by H. Hugh, though it might take some time to appear.