Steven Berlin Johnson talks in this much cited article about his research workflow using DevonThink. In short, he advocates for creating 50-500 word focused text snippets from your research readings and placing them in DevonThink for future use. Reason: short, focused snippets allows DevonThink's "See Also" AI feature to come up with many overlooked associations between your snippets. These associations allow you to get a much more in-depth use of your previously read research that you may have forgotten about or never made a link to on your own.
My question: is it worth the effort? Is the pain of accumulating enough 50-500 word snippets in DevonThink during research worth the gain you receive from it in the writing process? Converting previously-read articles to 50-500 word snippets can be very time consuming. And the effort required to create a DevonThink snippet database large enough to make the "See Also" function useful is a substantial investment.
Has anyone used both Johnson’s DevonThink workflow and another database applicaton with standard search features? Can you offer a comparison? Are the associations and “new leads” that are uncovered using Johnson’s DevonThink 50-500 word snippet technique worth the effort? Or will you get, for example, 80-90% of the same results just using whole articles dumped into Evernote, EagleFiler etc and relying on its built-in search functions when retrieving research ideas? Are these two research techniques comparable or entirely different in their results? If so, how?