Skip to Content.
Sympa Menu

nafex - [NAFEX] Three strikes and you are out

nafex@lists.ibiblio.org

Subject: North American Fruit Explorers mailing list at ibiblio

List archive

Chronological Thread  
  • From: "Joe Hecksel" <jhecksel@voyager.net>
  • To: nafex@lists.ibiblio.org (North American Fruit Explorers)
  • Subject: [NAFEX] Three strikes and you are out
  • Date: Wed, 26 Mar 2003 07:53:01 -0500

Three strikes and you are out
A statistical justification for a common rule-of-thumb

One of the activities that we teach our children is to build
campfires. Building campfires are lessons in craftsmenship, materials,
luck, statistics, safety, patience and persistance. One of the
beautiful things about a campfire is that it provides unfiltered
feedback to the lighter. It either lit or it fizzled.

Sometimes patience is learned by the parent. My boy was reaching for
his 18th match when I felt a need to comment. "You know, chances are
that the 18th match will not work if the first 17 did not. A lot of
people would have made some changes to the set-up after the third try."

My son, who tends to be legalistic asked, "Why three?"

Indeed, why three?

************************************************************

Imagine you want to grow ABC berries, a new kind of fruit. You have
been successful with many other kinds of fruit although you never tried
ABC berries. You assume that you have a 50% or better chance of
success based on your other ventures.

For the purposes of this essay 50% chance of success means that if you
were able to plant 1,000,000 plantings over 100 years time, randomly
sprinkled over your site, that half of them would live long enough to
produce fruit.

This is called the hypothesis. The null hypothesis is that less than
half of them will succeed.

So you make your first planting. They die. Hey, based on your
hypothesis it was a flip of the coin whether they would survive or
not. No big deal.

Then you make your second planting. They die. Given your hypothesis,
the chances of that happening are 0.5 X 0.5 or 25%. Hmmmmm.

Then you make your third planting. They die. Chances of that
happening are 12.5% given your hypothesis. Playing fast and loose with
statistical terminology, you now have enough information to say that
you are about 90% certain that you do not have a reasonable (+50%)
chance of successfully growing ABC berries.

So you stop wasting money on blueberry plants. Ooops, ABC berries.

Caveats:
-I still plant M&M and Skittle seeds because the cost is so low and the
potential payoff is so high. Sometimes it makes economic sense to
exceed three strikes. Matches are cheap if you have a box of
them.....so maybe three is not written in stone for lighting a fire.
-Random rewards train people to be more persistant. Suppose you have a
site with a great deal of variation in soil and exposure. You may have
had many experiences where you were not successful until the second or
third (or 4th or 5th) attempt. You *know* you can be successful if you
try enough spots.
-Extreme uniformity trains people to be less persistant. Suppose you
have a site that is extremely uniform. What advantage is there to
trying out 2 other locations on the site?



--
Joe Hecksel
Personal Webpage http://my.voyager.net/~jhecksel




Archive powered by MHonArc 2.6.24.

Top of Page