Sobering thoughts for us techno-geeks.
On 13 May 2010, Marc Ambinder (The Atlantic) wrote The Fires This Time–Joe Flood on Managing New York City”, interviewing Joe Flood on NYC’s 1970’s fetish on efficiency and “how its overreliance on smart guys and computer formulas turned out be a disaster, especially when it came to the withdrawal of fire protection from poorer neighborhoods” with an abundance of fires.
One of the big appeals of using numbers to understand complex problems is getting counterintuitive results, which by definition go against common sense. After all, why spend all the time and money on a study that will only tell you what you already suspected? (…) Those are the kind of results the city hired RAND to produce, and that’s what they got.
Quoting Bill James in Michael Lewis’s Moneyball,
“Any new metric should tell you 80% what you already knew, and 20% what you didn’t. Less than 20% and it’s not very useful, more than 20% and there’s probably something wrong with the numbers.”
Perhaps Governor Deval Patrick and the Massachusetts supporters of subsides for casinos should read Joe Flood’s forthcoming book, The Fires: How a Computer Formula, Big Ideas, and the Best of Intentions Burned Down New York City-and Determined the Future of Cities.
-RichardH
Humor (Only read if you’ve got time to kill):
A Mathematical Model for the Dynamics and Synchronization of [WHAT?].
As mentioned in “Metric Mania,” Einstein said, “Not everything that can be counted counts, and not everything that counts can be counted.”
“Metric Manua” was an interesting article.
This may be a case of the media trying to undo their part in Greenberg’s Law of the Media.
You might also be interested in John Allen Poulos’s 16 May 2010 article in the NY Times Magazine, Metric Mania.
A corollary to what I wrote is that “when you find an obvious error in a program and correct it, you cannot always understand why that correction fixed every symptom that it did fix.”
Much as you hate to let these unanswered questions go, you often cannot afford to spend the time answering them.
As an engineer, I have run across many counter-intuitive results in my career.
When first discovered, this is a case of “Who ya gonna believe?.”
I always made it a practice to understand why my intuition failed. If I could not figure that out, it was just as likely that my computer program was incorrect. Frequently enough, it was the program that was wrong and my intuition was right. That is how you test and how you debug software, by cross-checking it against your intuition. Only a fool of an engineer would fail to do so.
Or as I like to say when trying to find a bug in a program, “When you see some behavior that cannot happen based on all your assumptions, then at least one of your assumptions must be wrong. When you figure out which one of your assumptions is wrong, then you have found the bug.”
I think it irresponsible to debug by trial and error and be satisfied when you make the symptom go away.
That’s probably why I didn’t fit in the software industry anymore as I grew older and the management grew younger.