Month: May 2005

An anonymous commenter pointed me to this comic panel:

Athough it sounds like something the Head Counselor at the Micahel

Jackson Sleepover Camp does, a butt-stroke is actually a rifle-combat maneoeuvre in

which you strike someone with the butt of your rifle. Butt-strokes were apparently on the menu at Abu Ghraib.

If you are in what Douglas Adams liked to call the Shouting and Killing People industries, you can learn butt-stroke techniques here.

…was not this. “Icy Eight” or “Special K” on their own? Fine. But together? That’s just overkill, man.

Evan Robinson, author of the IGDA paper Why Crunch Mode Doesn’t Work, commented on my entry about it. I thought it was worthy of getting promoted to its own entry. Thanks, Evan!

I didn’t hammer the point home in Why Crunch Mode Doesn’t Work, but sleep deprivation is much, much, much worse than long-term Crunch Mode. The analogy that I draw in something I’m writing now is that long-term Crunch Mode is “shooting yourself in the foot”, while sleep deprivation is “shooting yourself in the head.”

Military studies (I cite Colonel Belenky’s work in Why Crunch Mode Doesn’t Work) show that each 24 hours without sleep reduces cognitive function by 25%. Other studies report the equivalence of 21 hours awake to legally drunk.

The details of Belenky’s work are fascinating: sleep deprived soldiers can shoot stationary targets as well as ever, but their performance against popup targets (which require constant observation, decision-making, and precise movement under time pressure) degrades rapidly. Belenky’s conclusion is that “In contrast to complex mental performance, simple psychomotor performance, physical strength and endurance are unaffected by sleep deprivation.” Unfortunately for us, programming is not primarily “simple psychomotor performance”.

Belenky concludes that the mechanism for this failure is the slowdown of decision-making. When a decision has to be made, but the sleep-deprived soldier can’t make a decision in the time available, essentially an intermediate result from the ongoing calculation is output as the decision. Naturally, an incomplete decision is often not the correct one. Friendly fire incidents can result. Belenky’s description of what happens to a sleep-deprived artillery unit would be funny if it weren’t so serious. At GDC I spoke to an Army Captain back from Iraq who was concerned that this information wasn’t made available to him and his men.

Unfortunately, recognition of sleep deprivation as a problem isn’t widespread, especially in programming. We’re still proud of pulling all-nighters or 100 hour weeks, and remain (sometimes willfully) ignorant about the effects on our bodies and our projects. Dr. Dement’s 1999 book The Promise of Sleep provides a good overview of the general problem. As the founder of the Stanford Sleep Center in Palo Alto, Dr. Dement is better position than many to understand the consequences of sleep deprivation.