||[Jan. 25th, 2006|12:28 am]
Quick update: |
On Sunday, Bryree, Kate, Mike, and I helped Neal playtest a module he's editing. It was fun! Warlocks are cool. We killed a vampire aboleth. And it provided useful feedback, so that's spiffy. (e.g., a room filled with water and a rusty iron bridge going from doorway to doorway just screams "trap!" and is unlikely to be entered...)
Yesterday was the first day of the Decision Analysis workshop at work. It's pretty cool. How to explain it... Okay, say you've got a hydro power plant that's coming up for licensing, and there are a bunch of interest groups wanting various changes to how it runs. Assuming you can monetize the social value of the proposed changes, decision anaysis lets you figure out that, well, we don't actually need to know exactly how much a person-day of recreational fishing is worth, because as long as it's somewhere in this general range, it's always worth it to make the change. Likewise, water policy B is never worth it, and knowing the existence value of this chunk of fish habitat is important, but we shouldn't spend more than X dollars finding out.
Extra-special neatness is that the coursework includes the phrases "value of clairvoyance" and "value of wizardry".
Today I successfully massaged climate datafiles into a new format, then went off to Star Wars at Jeff's. I got to selectively turn the gravity off on the interplanetary shuttle for everyone but the party. ...Of course, I accidentally disabled half the ship's engines doing it, and I didn't so much turn gravity off as reverse it, but still, it was cool.
Decision analysis is a totally weird field. The company that I've been working at is all about decision analysis, and I've been working there for eight months, and I'm still not sure I believe in it. It's kind of neat, in the sense of "Hey, if we assign numbers to all of these different quantities, we can extract conclusions", but whether those numbers actually mean anything is something I'm still very skeptical of. When I think of it in my head as a giant strategy dice-rolling game, it's fine, but when I think that people are actually using these numbers to make real decisions, I get the heebie-jeebies.
But maybe I'm just a wuss.
One of the things the teacher emphasized is that a lot of people do it wrong, and don't spend nearly enough time on problem-structuring and deterministic analysis phases.
I think the monetization / assigning numbers thing work if you think of it as a way of quantifying things so you can compare them in a sort of fuzzy way. But yeah, I don't know that they *mean* too much.
Apparently the biggest benefit is not the recommended decision, but the fact that everyone involved can see that their concerns are being addressed by the process. In the dam example, the energy company ended up making the concessions that the decision analysis recommended, plus a couple extras, mosly because it got them to see the social benefits people were getting out of it...
Oh, I see the benefits of doing it. My company tends to focus on portfolio management, which is trying to figure out which of a number of investments (e.g. potential drugs going into clinical trials) gets the best return on investment. So decision analysis is helpful in playing with the numbers (putting in best case, worst case, and most likely case scenarios) to give people an idea of what numbers have the most impact on the desired results. If it turns out that time to market is the biggest driver of profits, it lets them know where to focus their efforts. If it turns out that uncertainty in projected market size has a huge impact on results, they know they need to spend more time getting better numbers on that, etc.
So it's got its uses in your case, where putting everything into a common unit of dollars allows comparison, and in my company's case, where it's used as basically a investment maximization game. And if decision analysis were always used for those purposes, I'd be less nervous. But there's still way too many people in this world that go "Oooh, numbers! It's quantitative! It must be THE TRUTH!" And ways to manipulate numbers to get any result that people want, as described in Priceless
. And the combination of those two factors makes me nervous. More education, dammit.
I practice a form of this when I'm trying to recommend software features for our product. Unfortunately, most of the new (and cheap to implement) features I recommend are completely ignored.
I suspect one of these days the only way to convince them to try to listen is to sneak in features and see how customers like/dislike them. :-)