WEBVTT

00:00:00.000 --> 00:00:01.936 align:middle line:90%
[SQUEAKING]

00:00:01.936 --> 00:00:03.388 align:middle line:90%
[RUSTLING]

00:00:03.388 --> 00:00:04.840 align:middle line:90%
[CLICKING]

00:00:04.840 --> 00:00:12.120 align:middle line:90%


00:00:12.120 --> 00:00:13.980 align:middle line:84%
RICHARD DE NEUFVILLE:
This is the first

00:00:13.980 --> 00:00:17.130 align:middle line:84%
of a couple of sessions
revolving around decision

00:00:17.130 --> 00:00:18.650 align:middle line:90%
analysis.

00:00:18.650 --> 00:00:24.960 align:middle line:84%
And what I want to do here is
to introduce a useful technique

00:00:24.960 --> 00:00:28.500 align:middle line:84%
in this way of thinking
about analyzing how

00:00:28.500 --> 00:00:30.570 align:middle line:90%
we respond to uncertainties.

00:00:30.570 --> 00:00:33.960 align:middle line:84%
And it's appropriate for
simple uncertain situations.

00:00:33.960 --> 00:00:35.520 align:middle line:90%
I emphasize that.

00:00:35.520 --> 00:00:40.950 align:middle line:84%
So the other aspect
I want to emphasize

00:00:40.950 --> 00:00:46.320 align:middle line:84%
is that this is a method that
was largely popular starting

00:00:46.320 --> 00:00:49.590 align:middle line:90%
around 40, 50 years ago.

00:00:49.590 --> 00:00:52.890 align:middle line:84%
It was developed
by a person called

00:00:52.890 --> 00:00:55.530 align:middle line:84%
Howard Raiffa, who was
a great exponent of it,

00:00:55.530 --> 00:01:01.050 align:middle line:84%
and Ralph Keeney,
who I taught with.

00:01:01.050 --> 00:01:03.810 align:middle line:84%
At around that
period in the '70s,

00:01:03.810 --> 00:01:05.200 align:middle line:90%
we did some work together.

00:01:05.200 --> 00:01:09.480 align:middle line:84%
And it was very popular
and was a staple

00:01:09.480 --> 00:01:16.380 align:middle line:84%
of most analytic business school
courses, which it is now not.

00:01:16.380 --> 00:01:21.660 align:middle line:84%
That-- it took a while,
but what has happened

00:01:21.660 --> 00:01:24.340 align:middle line:84%
is that a number of
companies didn't address it.

00:01:24.340 --> 00:01:27.000 align:middle line:84%
And it is certainly
an advantage over not

00:01:27.000 --> 00:01:28.800 align:middle line:90%
addressing uncertainty.

00:01:28.800 --> 00:01:32.370 align:middle line:84%
And for example,
I know, since I've

00:01:32.370 --> 00:01:35.550 align:middle line:84%
been working with Chevron
a bit, that they have,

00:01:35.550 --> 00:01:38.580 align:middle line:84%
and I have a copy of their,
decision analysis manual.

00:01:38.580 --> 00:01:43.500 align:middle line:84%
And it goes on to
articulate why it

00:01:43.500 --> 00:01:48.210 align:middle line:84%
is said that this is to
be a company-wide approach

00:01:48.210 --> 00:01:52.110 align:middle line:84%
to dealing with
questions of uncertainty.

00:01:52.110 --> 00:01:54.840 align:middle line:84%
I would be interested
at some point

00:01:54.840 --> 00:01:57.150 align:middle line:84%
to discuss with
the Chevron people

00:01:57.150 --> 00:01:58.710 align:middle line:84%
to the extent to
which they actually

00:01:58.710 --> 00:02:02.190 align:middle line:84%
see it being applied
because I have my doubts

00:02:02.190 --> 00:02:04.680 align:middle line:90%
from the discussions so far.

00:02:04.680 --> 00:02:10.199 align:middle line:84%
But the point is, it is a thing
that you ought to know about.

00:02:10.199 --> 00:02:13.500 align:middle line:84%
And it has some advantages
and provides you

00:02:13.500 --> 00:02:16.330 align:middle line:84%
with some insights that would
not otherwise be available.

00:02:16.330 --> 00:02:18.330 align:middle line:84%
So I think that's
useful to know about.

00:02:18.330 --> 00:02:21.540 align:middle line:84%
But in terms of
practical applications,

00:02:21.540 --> 00:02:27.780 align:middle line:84%
it's inherently limited to
some relatively simple cases.

00:02:27.780 --> 00:02:29.620 align:middle line:84%
That doesn't mean
it's not important,

00:02:29.620 --> 00:02:33.065 align:middle line:84%
it's just that it is not
a competitor, in my mind,

00:02:33.065 --> 00:02:34.440 align:middle line:84%
with the kind of
things that I've

00:02:34.440 --> 00:02:39.560 align:middle line:90%
been trying to share with you.

00:02:39.560 --> 00:02:45.440 align:middle line:84%
So I think it's important to
think about the kind of things

00:02:45.440 --> 00:02:50.730 align:middle line:84%
which become fairly obvious
when you use decision analysis.

00:02:50.730 --> 00:02:52.970 align:middle line:84%
So I call them the
three conclusions

00:02:52.970 --> 00:02:56.780 align:middle line:84%
that any decision analysis
will basically lead you

00:02:56.780 --> 00:03:00.920 align:middle line:90%
to, or generally lead you to.

00:03:00.920 --> 00:03:03.410 align:middle line:84%
And I think that they are
also the ones that you

00:03:03.410 --> 00:03:06.530 align:middle line:90%
see in flexibility analysis.

00:03:06.530 --> 00:03:09.350 align:middle line:84%
And I've been talking
about it, but it is not

00:03:09.350 --> 00:03:13.680 align:middle line:84%
something that is so obvious
when you do a simulation

00:03:13.680 --> 00:03:14.280 align:middle line:90%
and so forth.

00:03:14.280 --> 00:03:16.370 align:middle line:84%
You don't see the
structure, but you do see it

00:03:16.370 --> 00:03:19.320 align:middle line:90%
in a decision analysis.

00:03:19.320 --> 00:03:26.180 align:middle line:84%
So the first one is
that it is good to have

00:03:26.180 --> 00:03:30.560 align:middle line:84%
a strategy for altering
your choices as a future is

00:03:30.560 --> 00:03:31.550 align:middle line:90%
revealed.

00:03:31.550 --> 00:03:35.750 align:middle line:84%
That is, that when we deal with
uncertainty, instead of saying,

00:03:35.750 --> 00:03:39.350 align:middle line:90%
I have a plan to do x--

00:03:39.350 --> 00:03:42.080 align:middle line:84%
build out my supply
chain, for example,

00:03:42.080 --> 00:03:49.730 align:middle line:84%
or build up the factory, or
prepare for sea level rise--

00:03:49.730 --> 00:03:52.280 align:middle line:84%
to have a unique plan
that says, phase one

00:03:52.280 --> 00:03:54.720 align:middle line:84%
is this, phase 2 is
that, phase 3 is this,

00:03:54.720 --> 00:03:58.610 align:middle line:84%
and so forth, that is
not the best solution,

00:03:58.610 --> 00:04:01.940 align:middle line:84%
but you want a strategy,
a strategy which says,

00:04:01.940 --> 00:04:06.860 align:middle line:84%
not that I have a unique
choice, but first I do this,

00:04:06.860 --> 00:04:10.520 align:middle line:84%
and then I do the next
move depending upon what

00:04:10.520 --> 00:04:12.450 align:middle line:90%
I see happening in the world.

00:04:12.450 --> 00:04:18.050 align:middle line:84%
So it's an analogy, if
you want, to a chess game.

00:04:18.050 --> 00:04:20.959 align:middle line:84%
That is, you think
ahead, you have a plan,

00:04:20.959 --> 00:04:23.540 align:middle line:84%
but depending on what
your opponent does

00:04:23.540 --> 00:04:25.730 align:middle line:84%
or what happens
in the real world,

00:04:25.730 --> 00:04:28.350 align:middle line:84%
you decide to move
this way or that way.

00:04:28.350 --> 00:04:34.290 align:middle line:84%
So that you have a
strategy, not a fixed plan.

00:04:34.290 --> 00:04:40.590 align:middle line:84%
The second aspect of it is
that often your better choice

00:04:40.590 --> 00:04:43.400 align:middle line:90%
is never your first choice.

00:04:43.400 --> 00:04:50.390 align:middle line:84%
That is, that if you buy
insurance against a downside,

00:04:50.390 --> 00:04:53.660 align:middle line:84%
at the end of the year you've
bought, say, car insurance

00:04:53.660 --> 00:04:55.250 align:middle line:84%
and you never had
an accident, you

00:04:55.250 --> 00:04:57.830 align:middle line:84%
could say, well,
that wasn't so good.

00:04:57.830 --> 00:04:58.820 align:middle line:90%
Why wasn't it good?

00:04:58.820 --> 00:05:00.903 align:middle line:84%
Well, I payed all this
money for my car insurance,

00:05:00.903 --> 00:05:04.430 align:middle line:84%
and I didn't have an
accident so I didn't need it.

00:05:04.430 --> 00:05:07.190 align:middle line:84%
But having the insurance is
still a better choice because,

00:05:07.190 --> 00:05:08.820 align:middle line:90%
what if you did need it?

00:05:08.820 --> 00:05:12.140 align:middle line:84%
But on the other hand,
you could say, well,

00:05:12.140 --> 00:05:16.610 align:middle line:84%
buying the insurance was not
my best choice if I never

00:05:16.610 --> 00:05:17.820 align:middle line:90%
had an accident.

00:05:17.820 --> 00:05:19.640 align:middle line:84%
On the other hand, if
you had an accident

00:05:19.640 --> 00:05:23.330 align:middle line:84%
and you got really hurt, you
said, well, my better choice

00:05:23.330 --> 00:05:25.730 align:middle line:84%
might have been not
to drive at all,

00:05:25.730 --> 00:05:28.910 align:middle line:84%
not just drive with
insurance and take some risk.

00:05:28.910 --> 00:05:32.150 align:middle line:84%
The point being that
the better policy,

00:05:32.150 --> 00:05:35.420 align:middle line:84%
for example, if you
agree but as an example,

00:05:35.420 --> 00:05:40.850 align:middle line:84%
that you have insurance
against the downside situation,

00:05:40.850 --> 00:05:44.760 align:middle line:84%
it is never the best
choice in retrospect.

00:05:44.760 --> 00:05:47.930 align:middle line:84%
It could have been, I did buy
insurance and I didn't need it,

00:05:47.930 --> 00:05:50.840 align:middle line:84%
or I shouldn't have been
doing that full thing anyway

00:05:50.840 --> 00:05:52.910 align:middle line:84%
and I shouldn't
have had insurance,

00:05:52.910 --> 00:05:54.450 align:middle line:84%
I should have just
been doing that.

00:05:54.450 --> 00:06:00.020 align:middle line:84%
So in that sense, your
better choice may be this,

00:06:00.020 --> 00:06:03.110 align:middle line:84%
your second best
choice is what may

00:06:03.110 --> 00:06:05.990 align:middle line:84%
be best given the
range of things

00:06:05.990 --> 00:06:08.450 align:middle line:90%
that are going to happen.

00:06:08.450 --> 00:06:14.480 align:middle line:84%
The third thing, which is
one of the outcomes of doing

00:06:14.480 --> 00:06:19.910 align:middle line:84%
a decision analysis is education
of the client, that is,

00:06:19.910 --> 00:06:22.790 align:middle line:84%
very often the person that
you're trying to help--

00:06:22.790 --> 00:06:25.970 align:middle line:84%
it may be yourself, somebody
you're consulting with,

00:06:25.970 --> 00:06:31.160 align:middle line:84%
a boss you're trying to please,
some person you're trying

00:06:31.160 --> 00:06:32.330 align:middle line:90%
to work with--

00:06:32.330 --> 00:06:36.120 align:middle line:84%
may focus on an outcome
and say, oh, if we do this,

00:06:36.120 --> 00:06:40.670 align:middle line:84%
here's what we may get and
not understand the process

00:06:40.670 --> 00:06:42.410 align:middle line:90%
and the risks involved.

00:06:42.410 --> 00:06:45.290 align:middle line:84%
If you lead them through
a decision analysis,

00:06:45.290 --> 00:06:48.470 align:middle line:84%
as I will be suggesting
to you in a moment,

00:06:48.470 --> 00:06:51.980 align:middle line:84%
you make them think about
whoa, wait a second, this could

00:06:51.980 --> 00:06:54.230 align:middle line:84%
happen, that could happen,
we should take those things

00:06:54.230 --> 00:06:55.230 align:middle line:90%
into account.

00:06:55.230 --> 00:06:59.300 align:middle line:84%
So very often the process of
going through a distribution,

00:06:59.300 --> 00:07:01.940 align:middle line:84%
or explaining your
distribution-- explaining

00:07:01.940 --> 00:07:05.450 align:middle line:84%
your setup, helps
educate the client

00:07:05.450 --> 00:07:09.020 align:middle line:84%
to understand what the
uncertainties imply.

00:07:09.020 --> 00:07:11.180 align:middle line:84%
So it's in line
with what I've been

00:07:11.180 --> 00:07:12.500 align:middle line:90%
emphasizing in this course.

00:07:12.500 --> 00:07:16.670 align:middle line:84%
It's step one is to
recognize the uncertainties.

00:07:16.670 --> 00:07:19.940 align:middle line:84%
And it's often the case
that clients or people

00:07:19.940 --> 00:07:23.850 align:middle line:84%
you're working with don't really
think very clearly about it.

00:07:23.850 --> 00:07:29.120 align:middle line:84%
So these are three general
advantages of the decision

00:07:29.120 --> 00:07:30.215 align:middle line:90%
analysis approach.

00:07:30.215 --> 00:07:33.380 align:middle line:90%


00:07:33.380 --> 00:07:35.880 align:middle line:84%
Now there's a
motivation for this,

00:07:35.880 --> 00:07:39.500 align:middle line:84%
which is that, by and
large, people when they're

00:07:39.500 --> 00:07:42.800 align:middle line:84%
acting intuitively deal
poorly with complex uncertain

00:07:42.800 --> 00:07:43.790 align:middle line:90%
situations.

00:07:43.790 --> 00:07:48.470 align:middle line:84%
There is all kinds of
data or experiments

00:07:48.470 --> 00:07:53.990 align:middle line:84%
with groups and so forth about
how people don't react well,

00:07:53.990 --> 00:07:57.260 align:middle line:84%
in the sense that they don't
understand probabilities,

00:07:57.260 --> 00:08:00.920 align:middle line:84%
they don't update
information, et cetera.

00:08:00.920 --> 00:08:03.590 align:middle line:84%
So we'll go into some
examples of that,

00:08:03.590 --> 00:08:06.020 align:middle line:90%
but they process it poorly.

00:08:06.020 --> 00:08:08.670 align:middle line:84%
Very often they look
at a complex situation,

00:08:08.670 --> 00:08:12.270 align:middle line:84%
and they simplify it
either by, for example,

00:08:12.270 --> 00:08:14.450 align:middle line:84%
focusing on the extremes--
this could happen,

00:08:14.450 --> 00:08:17.330 align:middle line:84%
that could happen, it's
very bad or very good--

00:08:17.330 --> 00:08:21.440 align:middle line:84%
and they don't look
on the entire process.

00:08:21.440 --> 00:08:25.280 align:middle line:84%
That is, it's not just that you
have a one-shot bet typically,

00:08:25.280 --> 00:08:27.980 align:middle line:84%
but you're doing
something over time--

00:08:27.980 --> 00:08:30.350 align:middle line:84%
say, building out your
supply chain, your factory

00:08:30.350 --> 00:08:35.090 align:middle line:84%
or whatever-- and that
there are intermediate steps

00:08:35.090 --> 00:08:38.700 align:middle line:84%
they're going through and
those count for a lot.

00:08:38.700 --> 00:08:44.370 align:middle line:84%
So decision analysis provide
a structured, efficient way

00:08:44.370 --> 00:08:47.760 align:middle line:84%
to help people recognize
this complexity,

00:08:47.760 --> 00:08:51.650 align:middle line:84%
to recognize the probabilities,
and to deal with it.

00:08:51.650 --> 00:08:58.000 align:middle line:90%