For Want of a Nail
Posted by rbpasker on August 15, 2007
One of my favorite books in the world is David Hackett Fischer’s Historians’ Fallacies : Toward a Logic of Historical Thought because it brings the rigor of logic to the study of history. Today I get to take Fischer for a tour of the computer room and show him the errors of his ways.
Under the heading Reductive Fallacies, Fischer writes:
In the computer room, however, broken nails are responsible for lost battles all the time.
In 1985, I was working at the Bank of New York on their government securities clearance system, which failed when a signed 16-bit index overflowed from
-37,768, and started overwriting memory. BoNY had to take out a $32 billion overnight loan from the Federal Reserve, and pledge the bank itself as collateral. (Although I was not involved in the application which had the bug, I spent the night in the computer room with the folks who were because I had responsibility for the underlying TP monitor software.)
A few days ago, Los Angeles International Airport was shut down because of the failure of a failed U.S. Customers computer system. Was it a terrorist? A hacker? Nope, it was a failed LAN card. It took 9 hours to fix, and stranded 17,000 travelers.
On the battlefield, there is a tremendous amount of redundancy: troops, ammo, tanks, etc. One guy falls, and there are 5 more to take his place, and the reductive argument doesn’t work.
In the computer room, however, there are lots of nails, the failure of any of which is sufficient to cause catastrophic failure.
As computer professionals, we have a responsibility to build redundancy into our systems.
This entry was posted on August 15, 2007 at 2:49 pm and is filed under flying, work. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.