From: hal@finney.org
Date: Wed Nov 01 2000 - 00:10:06 MST
Another part of the paper that I disagree with is sections 4.2 and 4.3,
on procrastination and addiction.
In procrastination, we put off an unpleasant experience. Later we wish
that we had done it sooner. In addiction, we cause a pleasant experience
which will have unpleasant consequences in the future. At that time we
regret our earlier choice.
Obviously both of these phenomena have to do with how we make and evaluate
choices over time. However, as I mentioned earlier with reference to
another part of the paper, discounting as typically understood cannot
be considered to play a significant role in these phenomena.
The problem is simply that the time frame is too short for the kinds of
discount rates which I have seen suggested to play a significant part.
Typical discount rates are proposed to be on the order of a few percent
per YEAR. Yet with procrastination we are usually putting off an
unpleasant event by days or weeks at most. And with addiction, the
unpleasant consequence comes within days or even hours.
None of the models I have seen would give a reasonable explanation of
addiction in terms of discounting, when the time span between pleasure
and pain is only a few hours.
Now, you could probably come up with a model that would work for this,
but it would not involve exponential discounting. You might have a sort
of a quasi delta function, where the present weighs extremely heavily,
and events just a few hours away are heavily discounted, with the rate
of additional discounting tapering off as we get farther away from the
present. However I think this model would be inconsistent with results
from economic experiments which show more moderate discount rates.
My conclusion is that our perception of time, and the impacts our choices
make over time, is complex. Phenomena like addiction and procrastination
really don't fit into the simplistic model offered by the authors.
Rather, they show that the actual situation is much more complicated.
Hal
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:31:48 MST