| Monday, May 23, 2005 |
15:01 - Whisper numbers
http://www.chicagoboyz.net/archives/003157.html#more
|
(top)  |
There's an old, old Dilbert strip—from the days before the Pointy-Haired Boss, even—where Scott Adams illustrates how numbers in business can be manufactured from thin air. It goes like this:
RANDOM STOCKBOY: I have no idea; it could be anything from one to a million.
MIDDLE MANAGER: They say it could be a million.
EXECUTIVE: Experts say one million.
That's what it's felt like in recent months watching the number of Iraqi dead climb and climb in the media—first latching onto the "100,000" Lancet figure and gnawing on it to this day in prime time, over and over, long after it was debunked; then upping it higher and higher to 250,000 or even 300,000, with seemingly no criteria for verification beyond "It's bigger than the last number, so it must be right."
So here's Shannon Love (via JMH) on the idea of the "number gut", something the people repeating these numbers don't appear to possess:
Why couldn't 250,000 be dead from violence? Well, the first clue is that the total population of Iraq is around 25 million, so 250,000 dead represents 1% of the entire population. That means if LIMS is accurate then 1 in every 100 Iraqis were killed in the war up to Sept 2004. So what? After all, it's a war and lots of people die in wars right? Well, not as many as most people think.
For example, during WWII the Japanese mainland suffered the most extensive aerial bombardment in history. Every major urban area save one (Kyoto) was burned to the ground. On march 10th, 1945 the great Tokyo fire raid burned down a third of the city and killed 100,000 people. Two major cities were nuked. Japan at the time had a population of 78 million, so 1% of the population would have been around 780,000. Now, what is your guess as to the number of Japanese killed on the Japanese mainland?
Did you guess around 500,000? Under 1%? Well, that is in fact the number (note: that's only dead, not dead-and-wounded).
So, with the Falluja cluster included, LIMS asks us to believe that Iraq has suffered a worse proportional aerial bombardment than did Japan during WWII. Common sense compels us to ask: does Iraq look like it suffered such a fate? Where are the mass graves? Where are the leveled cities? Where are the hundreds of thousands of walking wounded? Where are the millions of refugees that such intense fighting must have inevitably produced?
I get the feeling that there's nothing more to these numbers people are tossing around than to the "whisper numbers" that ruled the stock markets during the dot-com boom; we lived in terror that someone named ShadowKnight718 would post a message on some discussion forum somewhere that said "17 cents per share profit" or "2 cents per share loss"—from an anonymous IP and with no supporting research or anything—and traders would immediately set expectations accordingly. Suppose a company was about to announce a 15-cent-per-share profit, a number that in a vacuum would indicate fabulous breakout results? Well, someone on some chat board said 17 cents, so get ready to tank. Tough beans.
All you have to do is make up some number that one-ups the last one they said on the news, whether it's true or not, and it instantly attains the mantle of truth. It's almost as though they have an agenda to promote.
|
|
Brian Tiemann
