Re: Tufte seminar

["Matthew B. Hicks" <matt -at- UNIDATA -dot- UCAR -dot- EDU>]

On Fri, 27 Jan 1995, Jay Cherniak wrote:

>     Tufte spent a good deal of time discussing the data on O-rings that was
> reviewed by Morton Thiokol's engineers a few days before the Challenger
> disaster. (For the young folks reading this, Challenger was a space shuttle
 that
> exploded a minute after takeoff in January 1986, killing all of the
 astronauts.
> The explosion was later traced to failure of an O-ring seal in a fuel tank.)
> Tufte made a compelling case that if the engineering data on the O-rings had
> been displayed differently, so as to make clear the relation between ambient
 air
> temperature and ring failure, the Morton Thiokol engineers would not have
> given the go-ahead for the shuttle launch. That is, the Challenger was
 launched
> when the ambient temperature was low, and buried in the data was the clear
 fact
> that the O-rings tended to fail in cool temperatures.

I saw this talk when Tufte was presented with an award at a SIGDOC
conference a couple of years ago. The graph he made from the available
data indicated that an O-ring failure was not only likely, it was
virtually guaranteed. The problem with the graph of O-ring failures
versus temperature was that only instances of failures were plotted on
the graph; instances of launches without failures were left off. Because
of this omission, temp and failures seemed to be unrelated, but when the
missing data points were added in, the resulting curve shot up as the
temperature fell so that it was approaching 100% at 0 degrees.

As Jay says, he made a convincing case that the Challenger tragedy should
have been averted. I'd be interested in hearing him speak again.

One of my favorite charts that he used in his talk is still hanging on my
wall. It's a table used by John Gotti's defense team to earn him one of his
acquittals, and it shows the criminal histories of the people who testified
against Gotti (most of whom I assume have probably since been "whacked" :-)
in this particular trial. I'll try to scan it in later today and put it on my
Web page if anyone is interested in seeing it.  (I started trying to
reproduce it in ascii, but it just doesn't do it justice.) I can also scan
the chart showing the withering of Napolean's forces as they marched on
Russia and then stumbled home, if anyone is interested in seeing that.

-------------------------------------------------------------------
Matt Hicks, Tech. Writer, Unidata * I may not agree with what you
Boulder, CO, (303)497-8676, ******* say, but I'll defend to the
matt -at- unidata -dot- ucar -dot- edu ************* death my right to mock you.