Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site watmath.UUCP
Path: utzoo!watmath!jagardner
From: jagardner@watmath.UUCP (Jim Gardner)
Newsgroups: net.sf-lovers
Subject: Re: Sequels
Message-ID: <15015@watmath.UUCP>
Date: Tue, 11-Jun-85 11:53:23 EDT
Article-I.D.: watmath.15015
Posted: Tue Jun 11 11:53:23 1985
Date-Received: Wed, 12-Jun-85 01:44:51 EDT
References: <2188@topaz.ARPA> <242@azure.UUCP>
Reply-To: jagardner@watmath.UUCP (Jim Gardner)
Organization: U of Waterloo, Ontario
Lines: 47

[...]

Saying sequels are automatically bad is silly...remember that
Lord of the Rings is a sequel to the Hobbit (and I refuse to
listen to anyone who says the Hobbit is superior, much as it's
a nice story).

There's an interesting article in the latest issue of Discover
about why sequels are often let-downs (amongst other things).
It's due to a statistical principle called "Regression toward
the mean" which says that average results happen more often than
extreme results (either good or bad), so any extreme result is
usually followed by an average result.  This means that if an
author writes a very good book, it is more likely to be followed
by a less good book, simply because the average happens more often.
Of course, one author's average may be a good deal better than another
author's average, but almost no one stays at the peak consistently
(sad to say, for example, that Sword of the Lictor is noticeably
weaker in my opinion than the other books of the New Sun, even though
the four books taken together blow most of the rest of the field out
of the water).

By the way, the same principle (regression of the mean) has a very
interesting effect in education.  We know (from many many experiments)
that praise and positive reinforcement are better teaching methods in
the long run than punishment and negative reinforcement.  However, the
opposite is frequently perceived to be true by teachers.  When a student
does something very good, he/she will be praised; but the next time, the
student probably won't be as good, simply because average performances
usually follow extreme performances.  On the other hand, when a student
does something really dumb, he/she will probably be punished; and the
next time, the student's performance will probably not be as bad, because
again, average performances usually follow the extreme.  From the teacher's
point of view then, the punishment produced a better performance while the
praise was followed by a less good performance.  The teacher is therefore
inclined to believe that punishment is more effective in producing
results.

The article in Discover is full of all kinds of nifty things like
this: faulty reasoning that assumes statistical effects are due to
other causes.  Worthwhile reading for anyone in a decision-making
position.

				Jim Gardner, University of Waterloo