Organizing vs. searching

Posted by Paulo Diniz
Nov 23, 2006 at 02:20 AM

While i agree that scraping the outline is the way to go (for all the reasons you described, and also because they tend to get huge with time, so even if i remember where i put such info, it still take ages to navigate through the tree and all its levels), and also that being able to search all the data and being able to filter and actually find stuff in a snap is a central feature of a great PIM, I think that while searching and finding works when you need to mine a specific information, sometimes you want to see things information in a context, alongside other bits of similar information, so there is a big room for some kind of organization.

So, instead of the hierarchical approach, which is deeply flawed in a information overload scenario (good for digesting information, but not for capturing and making sense of it in a quick way), i think that the core organization feature of a PIM should be tagging (metadata). Tags are good because you can classify stuff in a very direct, cognitive way, which is designating categories to the information on an equivalent way your own brain would label the same information. It is very convenient too that you don’t need to put that new info in the structure of your universe of information. You just capture it, give it some tags and done. Tags also don’t have parents or children. They don’t have levels. They just intersect with each other, which is a great way of filtering similar information, and viewing it in ways that you’d never imagined when you tagged the info. It subtracts from the cognitive cost of having to choose a single place on the structure to place the info. With tags, an info can be on many places at the same time.

My ideal software for handling information would be such a PIM, with an ‘capture agent’ sitting unobtrusively on background, just waiting to be called by a keyboard shortcut (or by clicking on an icon at the tray) for new information to be filed. If you have a selection when you call it, it is acknowledged, and you just need to tag it and make minor edits to the text. If you don’t have a selection, you can always call the capture agent, type the new entry manually and then tag it. It can be anything, a phone number, an URL (del.icio.us style here), an inspiring prose, etc. It’s all stored in the same place, and at the same time it is on a multitude of different places, because you can always choose to just see a particular tag on the main window. This can be very good to implementing GTD contexts.

Speaking of the main window (and since the thread starter asked for our thoughts on the subject) i think that my ideal PIM would be very similar to a web browser. But instead of an address bar, you have a tag bar, on which you can enter the desired tag or tag intersection to enforce. Similarly to an Internet URL, each and every possible tag or combination of tags could be treated, as a matter of fact, as different locations, each one being customized for displaying its data on different ways. So, in some, you can just have your entries as a simple list on reverse-chronological order, like a blog. On others, you can display your entries alphabetically. Yet, for another tag or intersection of tags, you can create ‘specific data fields’ for the entries that bear those, and having the respective location to show the entries on a grid, just like a spreadsheet, with a variety of ways of sorting the grid based on the specific data fields of the entries on that location. This way, one could create specific mini-databases inside the general purpose knowledge base. And better, if you are filing a new entry in the ‘capture agent’ described on the preceding paragraph, and you manage to tag an entry with tags where you have created specific data-fields, the software could acknowledge that and let you fill all of those fields on-the-fly, just because you have tagged the entry on a particular way. (Zoot meets del.icio.us?)

Not letting the webbrowser analogy go, it would also be good if the user could bookmark some locations (tags or combination of tags), so he doesn’t need to type an often-used long tag intersection every time he needs to reach it. He could even make an outline hierarchy out of his bookmarks if he wanted, and this, IMO, would be acceptable because each bookmark would really act as a smart folder, and not like a regular node on a hierarchic structure. The cost of deciding where things go would still be low because it would still be all about what tags you do give to each entry, and not to where you put the new info on a huge structure. Of course, outliners are great for their own stuff, mostly when you want to structure a specific subject and can expect the outline to reach a given size and not grow beyond that. But i don’t think you can manage all your life’s information on an outliner.