Submitted for your consideration

Posted by Alexander Deliyannis
Aug 18, 2012 at 07:28 AM

Steve, great job, and I will disagree with you that there is no benefit from this analysis; I expect that at least people who come here looking for the right tool for a certain job will have a much more clear starting point—and CRIMPers will be able to see in which quadrant they are missing out…

Now, in respect to the taxonomy:

- Re the Main Purpose, I believe that the Collect-Create axis covers it quite well. I can also think of applications whose main purpose is to “Create via re-Organisation of Collected information” and I would place these closest to the middle of the axis. Brainstorm, Connected Text (e.g. break down imported text to smaller chunks and inter-associate these) and possibly Tinderbox would fit in this description.

- Re the Organisational Scheme, I am not so sure. For starters, I don’t think of tags as saved searches; the critical aspect of search is that it is machine-driven. Tags, just like the Topic approach, rely on the user to proactively classify the information. The fact that they are often in flat lists and not hierarchical is secondary. By contrast, search-based is just that: the user collects the information but is not concerned with its organisation, they just expect the software’s indexing to be able to locate it quickly when needed.

As an example, take Evernote. As discussed often in this forum, it boasts a very powerful search but lacks hierarchical folders; as a result, one would be tempted to classify it as a Search application. However, it also offers hierarchical tags, with which info can be organised as in a folder tree (but items can very easily belong to more than one categories). In this respect then, tags are just a more flexible take on Topics.

So, I would initially maintain the Search-Topic axis but I would restate the description to include tags under Topics, i.e. the distinction would be something like “Search (Automatic)” vs. “Topics (User-defined). Artificial intelligence applications which can themselves extract Topics would fall somewhere in the middle.

I also see an additional ‘dimension’: this is the organisation of information items via relationships. It is the approach taken by TheBrain and all wikis, including ConnectedText. Sure, these tools also provide tagging, but this is complementary. At their core lies the original concept of hypertext.

How would this be depicted? I believe that a triangle would be in order, i.e. Search - Topics - Relationships. I expect this is not so easy to produce, but I think that it more accurately represents the various approaches.