On Numbers

U.S. troop fatalities in Iraq, from the Brookings Institute's Iraq Index:

Can you see the steady upward trend of the past four years?

If so, you're seeing an optical illusion.

The deadliest year for U.S. troops in Iraq (thus far) was 2004, and the numbers actually decreased slightly in each of the following two years.

Actual numbers as reported by Brookings:

2004: 848
2005: 846
2006: 823

If you squint at the graph, you'll also see that December, 2006, was (at the time) the third highest monthly total of the war. So even as 2006 closed out as the second consecutive year with fewer deaths than the previous the media could ignore it and instead write headlines about that monthly total.

I'll repeat something I said about monthly totals at that time:

The variability of the numbers are chaotic, graphed they resemble nothing more than a saw's edge. Anyone who touts the peaks or valleys as representative is a fool. The media looks only at the peaks and declares them "trends". When the death toll plunged (predictably) after this year's Ramadan surge the media ignored it. When it rose again in the past month the death toll became headline worthy again. When the annual totals turned out to be lower this year it was reported under a headline about the monthly total being almost as high as it was back during my first tour.

That's certainly something to celebrate, but if recent history is any indication, next month will probably see an upturn (a bus crash would do it). And even if numbers continue to fall, this year will eclipse 2004 as the deadliest of the war. At that point, an amazing thing will probably happen - the headline writers will ignore the monthly totals* and discover the annual. Increases are newsworthy, decreases are not.

*And likely ignore the increase in combat troops, too.