Lastly, it seems that Ranking, Key-wording and Captioning are never completely done; at some point
you just decide that it’s time to put those originals to bed. To what extent are you “locked in” if you
want to continue massaging the metadata in the DNG’s but not have them sitting around ‘pre-archived’
This is a question that interests me as well, and is probably relevant to anyone who is fairly new at this process. Until we completely work out the kinks in our keywording process, we'll probably continue to find, for example, legacy images we imported a bit too hurriedly and are therefore light on keywording.
As I am not using Bridge or iView, I can't really address the specifics of how you might want to go about actually accomplishing these updates. Just from what I remember reading in Peter's book, iView most certainly can write keywords to IPTC (be sure to sync!), and of course virtual set info has no meaning to Bridge anyhow, so really, ratings are the only thing I think you might worry about about transferring to Bridge. I would say it seems to me that once you've done your initial processing with Bridge, you would probably be better off just using iView for as much as possible, and not worrying if Bridge can correctly see the new ratings, but if there is a straightforward and automatable to communicate the info back to Bridge, you might as well do it before running another backup to DVD.
In general, I think this is mostly an issue for legacy images and the first few batches of new images you bring in. I suspect that once you've developed a good process, you'll get better at doing your initial kewording. If nothing else, you won't be spending as much time keywording when you doing only new images as you may be now, when you might also be trying to import legacy images, so you won't be as tempted to skimp. I'm sure that's one reaosn Peter recommends not bothering with legacy images until you've worked out a process, but as someone to doesn't shoot hundreds of images a day, I felt that looking at my legacy images was important in figuring out what kinds of metadata make sense for me - working out my own "controlled vocabulary", for example.
I know that this means your write-once backups become obsolete - does anyone run into this issue and do
you then ‘re-burn’ the project if you make any later modifications to its metadata?
The approach I am taking right now is to expect this and use re-writable media for backup until I feel things have stabilized. I began this whole DAM process a few months ago after having my DSLR for about a year, and because my previous camera was only 2 MP and I shot moderately compressed JPEG, my entire archive - around five years' worth of pictures - fits on four DVD's. I've updated these backups several times as I've added more metadata, and expect I'll do so at least a couple more times before I feel like it would be worthwhile to burn them to write-once media. Realistically, I seem to have stopped active work on those files a couple of weeks ago, so I could probably just sync everything to IPTC and burn it now. But there doesn't seem to be any harm in waiting, and since I only just last week brought everything online on my wife's computer, now she gets to spend some time browsing around and maybe we'll find more organization to do.
Ultimately, though, it is a mistake to assume that because you are using "write-once" media, that means you really literally only *need* to write it once. We use write-once media because it is more stable long-term than re-writable, but still, I would not want to trust a backup I made to write-once media today to still be readbale in twenty or even ten years. Sure, it's possible we'll still have devices capable of reading DVD. But I would think it would be smart to expect to completely replace one's DVD backup library periodically. How often - well, I'm sure you'll get vastly different recommendations there.