Devonthink and tags
The Devonthink saga continues.
I posted on the DT forum asking a question about tags. The response was not quite what I suspected.
It turns out that my use of tags (i.e. tags as provenance, used to store the only copy of a document, the full original source) is officially frowned upon. It appears that tags and groups are not essentially the same thing after all. The developers’ concern was that if tags change in the future (either through DT or the Mac OS), that could have unexpected results on the tagging system in DT. A reasonable concern, but I didn’t realize that tags were such second-class citizens in the DTverse. Which reminds me of why I liked my custom-built Access database so much – I created it so I knew exactly how it worked, warts and all. It’s ability to make explicit relationships between various pieces of data and easily add or edit dozens of pieces of metadata to thousands of records in a few seconds made it really powerful for quickly assigning keywords and finding all records which met a whole slew of criteria (Author=English AND letter written before 1706 AND content about the Flanders theater…). Of course it was also slow (relatively speaking), unable to view various file formats, and generally had difficulty analyzing text.
Though I take no responsibility for anyone else’s data or system, if anyone has adopted my system and is of the paranoid variety, you can easily conform to the standard usage by:
- Creating a new Source group in the groups section
- Select all the provenance tags (and, by extension, their documents), and Move them to the Source group.
- That will copy all of the documents to the Source group, keeping the tag hierarchy intact. At least it did in my small-scale test.
You could then use tags for more standard tag-like things – you know, get taggy with it. I really don’t like mucking up the AI by combining all of Marlborough’s letters into a single group, since he talked about 50 different subjects in different parts of different letters, and I want to analyze by those subjects, not by whether Marlborough wrote the letters or not. So you might want to check Exclude from Classify for those provenance groups, especially if you have un-OCRable PDFs that may still have hidden metadata text somewhere.
I’ve been using my system for months without a hiccup (OK, one minor hiccup) and will continue to do so, but of course you should always have a backup (e.g. Time Machine), and the ability to roll back in case some future change does muck things up.
So I’m still waiting for the perfect historian’s note-taking software. If only I could combine the advantages of a textbase like DT with the advantages of a relational database (and throw in a bibliographic database like Zotero while we’re at it). That would be the bomb. I guess I didn’t appreciate how complicated the historian’s note-taking needs are.
And why doesn’t everybody have flying cars and robot maids by now?