WEBVTT

00:00:00.000 --> 00:00:01.940 align:middle line:90%
[SQUEAKING]

00:00:01.940 --> 00:00:04.365 align:middle line:90%
[RUSTLING]

00:00:04.365 --> 00:00:12.620 align:middle line:90%
[CLICKING]

00:00:12.620 --> 00:00:18.350 align:middle line:84%
RICHARD DE NEUFVILLE:
Basically, something like 8, 10.

00:00:18.350 --> 00:00:24.330 align:middle line:84%
10 moves at the beginning
with my pawns and my rooks,

00:00:24.330 --> 00:00:30.360 align:middle line:84%
I would have 10 possible choices
or 12 possible choices here.

00:00:30.360 --> 00:00:33.310 align:middle line:84%
And then I could have
what my opponent does,

00:00:33.310 --> 00:00:36.750 align:middle line:84%
and a realistic problem
can explode on you.

00:00:36.750 --> 00:00:39.630 align:middle line:84%
So one of the issues
here, again back

00:00:39.630 --> 00:00:43.980 align:middle line:84%
to my emphasis at the beginning
or my remark at the beginning,

00:00:43.980 --> 00:00:46.830 align:middle line:84%
is that decision analysis
is good for simple problems.

00:00:46.830 --> 00:00:50.595 align:middle line:84%
For complicated problems,
it just gets too messy.

00:00:50.595 --> 00:00:54.560 align:middle line:90%


00:00:54.560 --> 00:00:58.550 align:middle line:84%
So I'm going to illustrate
this with a simple example.

00:00:58.550 --> 00:01:02.260 align:middle line:90%
So the situation is--

00:01:02.260 --> 00:01:04.870 align:middle line:90%
and I get this each Fall--

00:01:04.870 --> 00:01:06.430 align:middle line:90%
it's autumn at MIT.

00:01:06.430 --> 00:01:07.795 align:middle line:90%
You can expect rain.

00:01:07.795 --> 00:01:09.670 align:middle line:84%
You're about to leave
your apartment and head

00:01:09.670 --> 00:01:11.980 align:middle line:90%
to campus for class.

00:01:11.980 --> 00:01:16.460 align:middle line:84%
What's the probability of
noticeable rain during the day?

00:01:16.460 --> 00:01:19.480 align:middle line:84%
So again, this is part
of decision analysis.

00:01:19.480 --> 00:01:24.670 align:middle line:84%
It's fairly simple that
what's a rainy day?

00:01:24.670 --> 00:01:26.560 align:middle line:90%
Does it rain for a few seconds?

00:01:26.560 --> 00:01:29.740 align:middle line:90%
Does it rain for 10 minutes?

00:01:29.740 --> 00:01:32.230 align:middle line:84%
Is it a downpour--
good many states,

00:01:32.230 --> 00:01:34.870 align:middle line:84%
but let's just say
it's a yes/no answer--

00:01:34.870 --> 00:01:36.820 align:middle line:84%
that's either going
to rain or it's not

00:01:36.820 --> 00:01:39.580 align:middle line:84%
going to rain, or
noticeable rain.

00:01:39.580 --> 00:01:43.060 align:middle line:84%
So you have four possible
outcomes-- you take your coat

00:01:43.060 --> 00:01:46.270 align:middle line:84%
and it rains, you take your
coat and it's not raining,

00:01:46.270 --> 00:01:48.190 align:middle line:84%
you didn't take your
coat and it rains,

00:01:48.190 --> 00:01:50.200 align:middle line:84%
you didn't take your
coat and no rains.

00:01:50.200 --> 00:01:54.730 align:middle line:84%
Now, I've given
values to the outcome

00:01:54.730 --> 00:01:56.860 align:middle line:84%
that a person might
feel for this.

00:01:56.860 --> 00:01:59.410 align:middle line:84%
I could I have a
cover story for them

00:01:59.410 --> 00:02:01.000 align:middle line:84%
as to why I invented
the numbers.

00:02:01.000 --> 00:02:03.010 align:middle line:90%
But they're all fake anyway.

00:02:03.010 --> 00:02:08.229 align:middle line:84%
It's more to show you a simple
example of how you set it

00:02:08.229 --> 00:02:09.620 align:middle line:90%
up and do the calculations.

00:02:09.620 --> 00:02:11.980 align:middle line:84%
So bear with me
that these numbers

00:02:11.980 --> 00:02:16.770 align:middle line:84%
are OK for the purpose
of a demonstration.

00:02:16.770 --> 00:02:19.590 align:middle line:84%
So first, I create
the decision tree.

00:02:19.590 --> 00:02:24.290 align:middle line:84%
So my first decision is, do
I take the raincoat or not?

00:02:24.290 --> 00:02:26.420 align:middle line:84%
Then whether it rains
or doesn't rain,

00:02:26.420 --> 00:02:28.220 align:middle line:84%
it so happens that
nothing to do.

00:02:28.220 --> 00:02:31.820 align:middle line:84%
Of course, nothing to do
with my taking the raincoat

00:02:31.820 --> 00:02:36.440 align:middle line:84%
affects the amount of rain,
so these are the same here--

00:02:36.440 --> 00:02:40.250 align:middle line:84%
probability of rain is on one
case, probably of no rain,

00:02:40.250 --> 00:02:41.610 align:middle line:90%
and so forth.

00:02:41.610 --> 00:02:48.500 align:middle line:84%
Now, it may well be in general
that the decision you make

00:02:48.500 --> 00:02:51.350 align:middle line:84%
affects the competition,
so their response

00:02:51.350 --> 00:02:52.740 align:middle line:90%
depending on what you have.

00:02:52.740 --> 00:02:57.980 align:middle line:84%
So the probability
associated with one choice

00:02:57.980 --> 00:03:00.230 align:middle line:84%
is not necessarily
the probability

00:03:00.230 --> 00:03:02.270 align:middle line:90%
associated with another choice.

00:03:02.270 --> 00:03:06.080 align:middle line:84%
That is, if you be
aggressive in the market,

00:03:06.080 --> 00:03:10.160 align:middle line:84%
your competitors may respond
by price cuts themselves,

00:03:10.160 --> 00:03:12.050 align:middle line:84%
and the whole situation
may be different.

00:03:12.050 --> 00:03:14.960 align:middle line:84%
If you're not aggressive there
might be not the price cuts,

00:03:14.960 --> 00:03:18.590 align:middle line:84%
and you have a
different situation.

00:03:18.590 --> 00:03:24.140 align:middle line:84%
But for the simple
demonstration, I have,

00:03:24.140 --> 00:03:26.450 align:middle line:84%
there's no dependence
between your choices

00:03:26.450 --> 00:03:29.540 align:middle line:84%
and the outcomes, which
is contrary to the most

00:03:29.540 --> 00:03:30.260 align:middle line:90%
general case.

00:03:30.260 --> 00:03:34.470 align:middle line:84%
But again, I'm trying to
make this very simple.

00:03:34.470 --> 00:03:39.580 align:middle line:90%
So first step, set up the tree.

00:03:39.580 --> 00:03:45.850 align:middle line:84%
Secondly, now the process is
to calculate the expected value

00:03:45.850 --> 00:03:47.390 align:middle line:90%
of the outcomes.

00:03:47.390 --> 00:03:51.130 align:middle line:84%
So given the data I have
for the probability of rain

00:03:51.130 --> 00:03:53.890 align:middle line:84%
and not rain, I can
calculate the expected value

00:03:53.890 --> 00:03:54.880 align:middle line:90%
of the raincoat.

00:03:54.880 --> 00:03:59.290 align:middle line:90%
So 2-- that is, 0.4 times 5--

00:03:59.290 --> 00:04:03.730 align:middle line:84%
minus 1.2-- that is,
0.6 times minus 2--

00:04:03.730 --> 00:04:06.400 align:middle line:90%
and it's 0.8.

00:04:06.400 --> 00:04:09.330 align:middle line:84%
I can do the same thing
for the other alternative.

00:04:09.330 --> 00:04:13.600 align:middle line:90%
It is 0.4 times minus--

00:04:13.600 --> 00:04:16.450 align:middle line:90%
whoops, I didn't notice that.

00:04:16.450 --> 00:04:18.940 align:middle line:90%
It's minus 10.

00:04:18.940 --> 00:04:21.950 align:middle line:84%
Somehow when I transferred
it from PowerPoint,

00:04:21.950 --> 00:04:24.670 align:middle line:90%
that got moved there.

00:04:24.670 --> 00:04:25.630 align:middle line:90%
Can I change it?

00:04:25.630 --> 00:04:26.890 align:middle line:90%
Probably not right now.

00:04:26.890 --> 00:04:29.590 align:middle line:90%
Well, I won't right now.

00:04:29.590 --> 00:04:31.840 align:middle line:90%
And that's a 4.

00:04:31.840 --> 00:04:35.410 align:middle line:84%
And no rain, which is
0.6 times a value of 4.

00:04:35.410 --> 00:04:40.360 align:middle line:84%
It is with a plus, so here I
have the two values' outcome.

00:04:40.360 --> 00:04:43.780 align:middle line:84%
This says, right, I've done
the decision tree on it.

00:04:43.780 --> 00:04:46.670 align:middle line:84%
With these values,
my better choice,

00:04:46.670 --> 00:04:51.070 align:middle line:84%
the expected value of the
decision to take the raincoat

00:04:51.070 --> 00:04:53.110 align:middle line:90%
is 0.8.

00:04:53.110 --> 00:04:55.270 align:middle line:90%
It's better than -1.6.

00:04:55.270 --> 00:04:57.970 align:middle line:84%
Decision marked is
to take the raincoat.

00:04:57.970 --> 00:05:01.970 align:middle line:90%


00:05:01.970 --> 00:05:05.400 align:middle line:84%
Now, I can also do this,
if you're interested in it,

00:05:05.400 --> 00:05:09.320 align:middle line:84%
I can do it not as a tree
but in an Excel layout.

00:05:09.320 --> 00:05:12.030 align:middle line:90%
So here's an example of it.

00:05:12.030 --> 00:05:13.790 align:middle line:84%
So I have the
decisions, and you can

00:05:13.790 --> 00:05:15.440 align:middle line:90%
imagine I leave the space here.

00:05:15.440 --> 00:05:18.410 align:middle line:84%
I haven't had the grid
behind it, remove that.

00:05:18.410 --> 00:05:19.850 align:middle line:90%
I can go through.

00:05:19.850 --> 00:05:23.660 align:middle line:84%
I can automate I can set the
equation so that probability

00:05:23.660 --> 00:05:26.390 align:middle line:84%
times outcome,
probability times this,

00:05:26.390 --> 00:05:28.850 align:middle line:84%
is the sum of the probability
times outcome here

00:05:28.850 --> 00:05:31.380 align:middle line:84%
plus a sum of the probability
times outcome here.

00:05:31.380 --> 00:05:35.210 align:middle line:84%
I get 0.8, which I put
in bold to indicate

00:05:35.210 --> 00:05:39.170 align:middle line:90%
that it's the better solution.

00:05:39.170 --> 00:05:40.770 align:middle line:90%
And I do the same thing here.

00:05:40.770 --> 00:05:43.460 align:middle line:90%
So it can be automated.

00:05:43.460 --> 00:05:45.680 align:middle line:84%
And there are
programs to do this.

00:05:45.680 --> 00:05:48.470 align:middle line:84%
The classroom
demonstration of the model

00:05:48.470 --> 00:05:50.900 align:middle line:84%
might be where
I'm tossing coins,

00:05:50.900 --> 00:05:53.480 align:middle line:84%
or how many do I get in a row,
or something, where everything

00:05:53.480 --> 00:05:54.440 align:middle line:90%
is discrete.

00:05:54.440 --> 00:05:57.030 align:middle line:84%
The real world, however,
is quite complicated,

00:05:57.030 --> 00:06:02.060 align:middle line:84%
so that if you're thinking of
using it to model a business

00:06:02.060 --> 00:06:05.240 align:middle line:84%
expansion plan or
some kind of activity,

00:06:05.240 --> 00:06:07.430 align:middle line:84%
you'd have to have some
kind of way to say, OK,

00:06:07.430 --> 00:06:11.510 align:middle line:84%
if I did this and this-- if I
took the raincoat or I did this

00:06:11.510 --> 00:06:13.060 align:middle line:90%
action and this happened--

00:06:13.060 --> 00:06:16.420 align:middle line:84%
I have some way of
calculating the result.

00:06:16.420 --> 00:06:18.640 align:middle line:84%
And I didn't want
to show something

00:06:18.640 --> 00:06:21.080 align:middle line:84%
like that for this
simple example,

00:06:21.080 --> 00:06:27.000 align:middle line:84%
so I invented these
numbers to fulfill it.

00:06:27.000 --> 00:06:33.068 align:middle line:90%