A day to remember

[gar]

111207-1107 EST

This note is about information theory, and signal detectability.

About 8 years ago I was searching the Internet with my psychology professor's name, Wilson P. Tanner, and found the following:

I found http://www.wwiivets.com/IssueXV/PearlHarbor.htm referencing this name. Is this the same person I knew? Spike I believe was in the Navy. If so he never mentioned the following occasion. Quoting from this site --- Dec 7, 1941 --- " "At 0640 I received a Teletype from the naval air station communication center which said, in effect, Ensign Wilson P. Tanner had sighted and sunk an enemy submarine one mile south of Pearl Harbor." According to Thomas the messages, which were sent in code, caused some initial confusion. One of the radiomen had attempted to alert Pearl Harbor that they were under attach, Thomas said. But the reply came back, "You guys must have had a wonderful Saturday night. Can't you keep those drunk radiomen off the air on Sunday morning?" ". Sometime in the future more on Pearl Harbor. As you probe the Internet in different ways various bits of information come together.

Continuing with my comments:

Here is a good illustration of making assumptions instead of verifying information. The same problems exist today and illustrates the failure to use a priori information (the knowledge that existed prior to and early on 7 Dec 1941 in that event, like 3 am or the days and months before).

More on Spike Tanner. Further searching indicates that the PBY pilot was an Ensign William P. Tanner. But relative to Spike Tanner's work if you search on "W. P. Tanner" Swets, or "W. P. Tanner" Green, or "W. P. Tanner" Birdsall, or "W. P. Tanner" Machol you will find many references to his psychophysical work.

There may be more to Dec 7th as indicated in the site
http://www.independent.org/tii/news/001207Stinnett.html

However, if you read in the following site it provides a different perspective. There seems to have been a great deal of information available to the Pearl Harbor commanders. Maybe some immediate information preceding 7 December was not available, but at least two or three significant items of information were essentially ignored before 8 am. Considering the knowledge that an attack from Japan was possible, possibly in early 1941 or earlier, then the top command should have instructed subordaniates to not ignore small bits of information. Signal detectability theory should have been applied. This would have indicated that a high false alarm rate would be acceptable to raise the probability of a true positive. On my page 35 of a printout of this site is a paragraph starting with:

:"4. On November 28, orders were issued to bomb unidentified submarines found in the operating sea areas around Oahu. ......."

On my page 62 is reference to "The "Code Destruction" Intelligence". My page 69 --- "The "Mori Call" ".

Page 70 detection of sub 0342 am. Although this may have been a noisy signal (meaning not sure of a positive detection) the information should have gone to the highest level. By not doing so more noise was effectively added to the communication channel. The sub is sighted and sunk about 0640 am. Again the information was ignored. The sinking was not noisy information. Page 72 on the Army Radar. To indicate a large number of planes is again not a noisy signal. The noise was again added in the communication channel to higher command by ignoring the information.

These are only limited bits of information. To get the real import you should read this site. This year, 2003, December 7 is again on Sunday.

www.ibiblio.org/pha/pha/congress/part_3.html

Some important points.

The disaster at pearl could have been minimized if small bits of information had not been ignored. Remember that in your daily work. Do not make assumptions when you do not have to. Signals (information) may be noisy but make use of the information.

New electronic equipment (radar) information was ignored or the wrong assumptions made about its signal. No real analysis or thought was given to the signal.

Correlating information from different points is very important. Sub sightings, sinking, radar, and the general knowledge that war was possible should have been put together. They weren't.

Spike Tanner's PHD was on signal detectability. I was subject in his visual experiments in early 1953, and by summer I was developing equipment for his audio tests and the invention of electronic random number generators.

See http://en.wikipedia.org/wiki/Detection_theory

See http://deepblue.lib.umich.edu/bitstream/2027.42/3602/5/bab2686.0001.001.pdf for some switched audio signals. These were tube days and to create a gating circuit with little switching transient and no DC offset was a difficult task.

This is an important read if you have the time www.ibiblio.org/pha/pha/congress/part_3.html

.

 

[mbeatty]

Great stuff Gar! Thank you. :thumbsup:
Mark

 

[handy10]

111207-1107 EST

Some important points.

The disaster at pearl could have been minimized if small bits of information had not been ignored. Remember that in your daily work. Do not make assumptions when you do not have to. Signals (information) may be noisy but make use of the information.

This is good advice in so many ways. I used to read mountaineering books. One of the more interesting is called Scrambles amongst the Alps by Edward Whymper. He was the first man to climb the Materhorn. On the way down, a rope broke and four members of his party were killed. Afterward he wrote something like this: "Climb
if you will, but remember from the beginning what the end may be for the indiscretion of the moment may destroy the happiness of a lifetime." Whymper's
advice seems to me to be in the same category as gar's applying to a much broader swath of life than its original context. Thanks