The erratic non-progress of the personal information manager

The Personal Information Manager (PIM) has had a difficult 24 years, since Borland's "Sidekick" more or less launched the genre. We're coming up on the 25th anniversary of Sidekick, and I think it's fair to say a geek of 1983 would be shocked by how little progress we've made. The iPhone has no tasks. What more can I say?

The PIM has been a longstanding interest of mine. At various times in my life I've had a pre-web listserv dedicated to the personal information manager, a now-defunct blog dedicated to the Palm and its alternatives and an abandoned web page or two on related topics. I thought of the PIM as I cleaned out some old files, with clippings about (some of these were groupware too) the golden years from 1983-1994. It was in 1994 that the reign of Sauron began.

• Arabesque's Ecco (much mourned)
• Lotus Organizer (ok, so it wasn't too fancy)
• Act for Windows (still around I think)
• CrossTies (object oriented model)
  
• MeetingMaker (cross-platform)
  
• Lotus Agenda (a classic)
• Attain Corp's "In Control": outliner/calendar combination
• GrandView: calendar/task/outliner/spreadsheet
• Ascend
• Commence
• Arrange 2.0 (Mac - bit of an object oriented database I think)
• InfoDepot
• FullContact
• First Things First (outliner, calendar)
• NewtonOS: a PIM that was an Operating System

Outlook came later, and the combination of Outlook/Exchange crushed the genre on the PC -- and finished off Palm as well (though by then Palm's owners had shot both feet off). Reinvention continues on the Mac, with a vast array of small vendor products that have various combinations of features of all of the 1983-1994 PIMs. On the web we have Backpack and a range of Web 2.0 apps, most of which will vanish in the next few years. Along with all your data.

Fifteen years ago I thought the salvation of the PIM would be application embedding, what we then thought of as OpenDoc. We'd have applications for projects, tasks, calendars and the like, and they'd all seamlessly interoperate with one another. That was a bit before I got into the interoperability business myself, building applications that tried to talk to one another about lab studies, diseases, genetic history, procedure history, consultations, etc. In that world software is relatively easy, the hard part is "meaning" -- having a common, or at least reasonably interoperable way to represent knowledge about things between systems. It starts with being able to generate a common data model (even if it's only used for communication), but it gets much harder than that when you need to store and create bits of data. That's when you get into really painful things, like formally maintained ontologies. (Engineers love emergent ontologies, which is more like the way our minds work, but interoperability between minds requires more CPU power than we have on the desktop.)

I think the 25 years of non-progress in PIMs springs from the same roots as 25 years of very slow progress in interoperable clinical systems (whether you want them to actually be able to share data is another matter - one of which I've blogged before). The domain of the PIM is far simpler semantically than that of the clinical record, but there's far less pressure for the grindingly hard work of common semantics and mutually agreed upon data models. I think we'll be at roughly the same point in 2033 that were were in 1983 ...