Free 802.11 here in San Diego (with a very good signal). A nice addition to both the conference hall and the hotel room. The connection in the hotel room may not be intentional .. but it's been a welcome addition to a nice trip.
Dinner last night in the hotel with the computer zoo staff: Nancy Clark, John Epling, Richard Neill MD, Meredith Oliver, Bruce Seaton, and Howard Rubenstein. Beth Milligan couldn't make it. We walked over to Kansas City Barbecue and brought back a great meal, and sampled some of the local beer. As the hotel had run out of rooms when I arrived, Bruce and I ended up sharing a suite on the 11th floor with a balcony, living room, two bedrooms, to bathrooms, etc .. so it was the right place to have a little gathering. A relaxing way to spend the end of a busy day.
Most of use have worked together in various capacities before. Almost exactly 4 years ago, Nancy, Beth and I were on a panel at the 1998 AMIA PCIWG meeting.
Andy Ury also gave a talk at that meeting. He and I spent some time together yesterday .. and he paraphrased something that I think holds true for medical informatics — and possible all good businesses: "It's harder to decide what NOT to do than it is to decide what to do." We have many opportunities … both in software development and in our daily lives. Making good choices about business alliances, software implementations, even treatment decisions for/with patients. The choices are vast. The right choices are few.
With apologies to Slawson & Shaughnessy for borrowing their usefulness equation .. I would suggest that P = V / W. What do I mean by this? The priority of a project is a function of the value over work. This will one day be referred to as "Reider's Rule" at Harvard Business school ;-). Value, of course, is a subjective measure. Making a judgment about how valuable a palm client is to our oncalls application was a subjective decision. On a scale of 1 to 100 … I give it an 80. Why? Physicians use palmpilots. Pulling the call schedule into the palmpilot would add significant value to those physicians who use them (most). Adding "pre-loaded" US holidays to the schedules automatically has been another user request. Sure .. it has some value .. but not nearly as much as the Palm client. So the "holidays" feature gets a 30. But we can't stop there and go ahead to do the palm client. This is where most project teams seem to falter. the marketers assess value and the engineers assess work. Without doing the prioritization together … accurate prioritization can't occur.
Work is assessed as a function of time and/or money. One should always use a common factor for work. When we started using this at Albany Medical Center, we agreed to use dollars as our W. If a project was to be implemented by a vendor, the vendor's fee was the denominator. If internal resources were to be used, we would multiply the number of hours that were estimated by $125. So long as we're consistent, it doesn't really matter what we use.
A problem with the model started to arise when high-budget items started being prioritized. The multimullion $$ replacement of the radiology information system would NEVER attain a "P" value over 1 because the W was simply too big. So we started using share-root of W to normalize very high budget projects.
Back to our examples. Palm client implementation was outsourced .. with some internal coordination. Let's say this cost $10,000. So P = V/W .. P = 80/10,000. P= .008. Hmm … let's try the Reider's Root Rule for high budget items: 80/100 = .8. Now let's compare that to the "holidays" request we estimate that it will take 20 hours to implement this (at $125/hr): P = 30/50 = 0.6 (I get 50 as my dominator by taking 125 * 20 = 2500 .. then I take the square root of that = 50).
So we implement the larger palm client project before the "holidays" feature .. even though holidays could be done more easily.