Glad my post helped!
Picking how many databases to use is a very personal choice that depends a great deal on how you work and the type of data you work with.
One major consideration is that the AI/See Also/Auto Classify features only work within the same database, and not across databases.
I think organization and retrieval is really the major consideration when thinking about number of databases. There are a couple main considerations related to this:
- False positives in the See Also feature (that is – items that the AI thinks are related but aren’t related according to you. This can happen if you have a lot of material that is unrelated by which might use the same words)
- Mis-classification in auto-classify (auto classify either puts something in the wrong place according to you, or is unable to classify because it cannot identify a clearcut group to send the file to. This is both an issue of the overall relatedness of your database, as well as your chosen classification scheme – that is, this issue can just as easily arise in a small but badly organized database.)
- False positives or difficult to parse search results (while the number and relevance of search results is, of course, also a product of your query, having a lot of files of questionable relatedness together can muddy results, especially if you are using a fairly vague query.) (also worth noting that the big ⌘-shift-f search window sill search all open databases (if you so choose, and is the default), so you CAN perform manual searches across all databases).
But all of the above considerations depend a great deal on how you see your data and how you use it. Two people might choose to treat the same set of data very differently and see merit in consolidating it into one database, or separating it into several focused databases.
Database SIZE (both file size and number of files) seems to have a very high upper bound before any performance degradation takes place. Database performance has been, for me, rather stable regardless of the size. For example, I have a database of 1800 files, predominantly PDFs, totally 35,000,000 words with a size on my disk of 3.1gb. This database opens rather quickly and navigating through it is effortless. I know this is far from the largest database that users on here have, but this library of files has given some other applications issues, but not DEVONthink. I feel confident that this database can grow probably several times its current size before I begin to see any issues at all.
The beautiful thing is, moving things into, and out of, and splitting up/consolidating databases is actually really easy. I recently went from about 9 databases to 4. Part of this was simply deleting unneeded files, but a large part of it was consolidating databases that worked just fine together – where their being in the same database enhanced retrieval and search because data that was originally split across databases. (related: Tips for splitting a database)