WEBVTT

00:00:00.000 --> 00:00:01.948 align:middle line:90%
[SQUEAKING]

00:00:01.948 --> 00:00:03.409 align:middle line:90%
[RUSTLING]

00:00:03.409 --> 00:00:06.331 align:middle line:90%
[CLICKING]

00:00:06.331 --> 00:00:16.400 align:middle line:90%


00:00:16.400 --> 00:00:19.400 align:middle line:84%
RICHARD DE NEUFVILLE: So the
reality of design analysis

00:00:19.400 --> 00:00:22.820 align:middle line:84%
is that the inputs
and the context--

00:00:22.820 --> 00:00:25.970 align:middle line:84%
that is, demand for whatever
the system is producing

00:00:25.970 --> 00:00:27.860 align:middle line:90%
or delivering-- are uncertain.

00:00:27.860 --> 00:00:29.120 align:middle line:90%
They have distributions.

00:00:29.120 --> 00:00:31.910 align:middle line:84%
We don't really know,
we can't predict

00:00:31.910 --> 00:00:33.980 align:middle line:84%
exactly how the
system will perform,

00:00:33.980 --> 00:00:36.332 align:middle line:84%
because we don't exactly
know what the costs are.

00:00:36.332 --> 00:00:37.790 align:middle line:84%
We don't know what
the demands are.

00:00:37.790 --> 00:00:39.870 align:middle line:84%
We don't know other
things about it.

00:00:39.870 --> 00:00:42.890 align:middle line:84%
So we won't be able
to predict it exactly.

00:00:42.890 --> 00:00:46.010 align:middle line:84%
So the performance of a
system, when we calculate it,

00:00:46.010 --> 00:00:48.380 align:middle line:84%
has to be some kind
of distribution,

00:00:48.380 --> 00:00:51.020 align:middle line:84%
some kind of range
of possibility

00:00:51.020 --> 00:00:55.100 align:middle line:84%
from the lowest to the best,
with different possibilities

00:00:55.100 --> 00:00:58.780 align:middle line:84%
of the different points
along the distribution.

00:00:58.780 --> 00:01:02.410 align:middle line:84%
And it's important to realize
that the outcomes of a system

00:01:02.410 --> 00:01:05.620 align:middle line:84%
are not the same as the
inputs to the system.

00:01:05.620 --> 00:01:09.490 align:middle line:84%
That is, if you put in
a normal distribution,

00:01:09.490 --> 00:01:11.890 align:middle line:84%
you don't necessarily get
a normal distribution out.

00:01:11.890 --> 00:01:14.830 align:middle line:84%
You're almost certain not to
get a normal distribution out.

00:01:14.830 --> 00:01:16.120 align:middle line:90%
Why is this?

00:01:16.120 --> 00:01:18.520 align:middle line:84%
It's because we have
complex systems.

00:01:18.520 --> 00:01:20.230 align:middle line:90%
They perform nonlinearly.

00:01:20.230 --> 00:01:22.510 align:middle line:84%
There'll be capacity
constraints so

00:01:22.510 --> 00:01:25.900 align:middle line:84%
that you can get the
downside, but not the upside.

00:01:25.900 --> 00:01:28.990 align:middle line:84%
There's all kinds of
reasons why what goes in

00:01:28.990 --> 00:01:31.990 align:middle line:84%
gets mixed up in
terms of the system.

00:01:31.990 --> 00:01:34.480 align:middle line:84%
And you get a different
kind of distribution

00:01:34.480 --> 00:01:35.960 align:middle line:90%
outside afterwards.

00:01:35.960 --> 00:01:38.200 align:middle line:90%
So keep that in mind.

00:01:38.200 --> 00:01:41.020 align:middle line:84%
Now, let me give you a
generic view of the kind

00:01:41.020 --> 00:01:42.760 align:middle line:90%
of distribution that can occur.

00:01:42.760 --> 00:01:45.440 align:middle line:84%
And let's see what we're
going to be doing about it.

00:01:45.440 --> 00:01:50.470 align:middle line:84%
So in general, we have a
distribution for the design

00:01:50.470 --> 00:01:52.870 align:middle line:84%
outcomes, represented
here in terms

00:01:52.870 --> 00:01:55.750 align:middle line:84%
of a probability
distribution function, not

00:01:55.750 --> 00:01:57.290 align:middle line:90%
a portable document.

00:01:57.290 --> 00:02:00.760 align:middle line:84%
So we have the possibility
that things don't go so well,

00:02:00.760 --> 00:02:03.160 align:middle line:84%
that there are losses, there's
poor performance some way.

00:02:03.160 --> 00:02:06.880 align:middle line:84%
That's on the left-hand
side of the graph.

00:02:06.880 --> 00:02:09.370 align:middle line:84%
And we have some
medium, perhaps.

00:02:09.370 --> 00:02:11.440 align:middle line:84%
It doesn't have to be
unimodal, but there's

00:02:11.440 --> 00:02:14.050 align:middle line:84%
various high points, and
more likely situations.

00:02:14.050 --> 00:02:17.245 align:middle line:84%
And then, we have
the possible upsides

00:02:17.245 --> 00:02:19.683 align:middle line:84%
that may occur if
everything goes really well

00:02:19.683 --> 00:02:21.850 align:middle line:84%
and if people really like
the product that you might

00:02:21.850 --> 00:02:23.420 align:middle line:90%
be managing, and so forth.

00:02:23.420 --> 00:02:25.760 align:middle line:84%
So there's some kind
of distribution.

00:02:25.760 --> 00:02:28.310 align:middle line:90%
So what do we do about it?

00:02:28.310 --> 00:02:32.080 align:middle line:84%
Well, what we want to do is
to have a proactive design.

00:02:32.080 --> 00:02:34.720 align:middle line:84%
That is, the
objective is that we

00:02:34.720 --> 00:02:38.920 align:middle line:84%
don't have to accept our
distribution that may

00:02:38.920 --> 00:02:41.650 align:middle line:90%
come from the base case design.

00:02:41.650 --> 00:02:44.740 align:middle line:84%
We can say, right, if
this is what's happening,

00:02:44.740 --> 00:02:48.620 align:middle line:84%
what can we do to
improve the design?

00:02:48.620 --> 00:02:52.840 align:middle line:84%
So what can we do
proactively is first of all,

00:02:52.840 --> 00:02:54.940 align:middle line:90%
to reduce the downside risks.

00:02:54.940 --> 00:02:58.000 align:middle line:84%
How could you make it
so that those downside

00:02:58.000 --> 00:03:03.110 align:middle line:84%
risks that you anticipated in
your first design don't occur?

00:03:03.110 --> 00:03:06.460 align:middle line:84%
Well, one way to do
it-- not the only way--

00:03:06.460 --> 00:03:09.070 align:middle line:84%
is to say, we're going to
build the system smaller.

00:03:09.070 --> 00:03:11.830 align:middle line:84%
So if it's built smaller
and it's not succeeding,

00:03:11.830 --> 00:03:14.300 align:middle line:90%
then you can lose less.

00:03:14.300 --> 00:03:15.738 align:middle line:90%
So that's one way to do it.

00:03:15.738 --> 00:03:16.780 align:middle line:90%
But there are other ways.

00:03:16.780 --> 00:03:19.570 align:middle line:84%
But in any case, we
can think about ways

00:03:19.570 --> 00:03:21.670 align:middle line:90%
to reduce the downside.

00:03:21.670 --> 00:03:25.550 align:middle line:84%
Complementarily, we can take
advantage of opportunities.

00:03:25.550 --> 00:03:29.180 align:middle line:84%
So for example, if the
product does really well

00:03:29.180 --> 00:03:30.980 align:middle line:84%
and if there's a
high demand for it,

00:03:30.980 --> 00:03:35.110 align:middle line:84%
we may be able to expand
our facilities, our system

00:03:35.110 --> 00:03:38.030 align:middle line:84%
so that we can rapidly
take advantage of it.

00:03:38.030 --> 00:03:42.940 align:middle line:84%
We've developed a new gizmo or
toy in advance of the Christmas

00:03:42.940 --> 00:03:43.450 align:middle line:90%
season.

00:03:43.450 --> 00:03:44.920 align:middle line:90%
It's very popular.

00:03:44.920 --> 00:03:50.050 align:middle line:84%
How do we go through and ramp
up the production and sales

00:03:50.050 --> 00:03:52.580 align:middle line:84%
to take advantage
of it, for example?

00:03:52.580 --> 00:03:56.890 align:middle line:84%
And at the same time, what we'd
like to do is certainly reduce

00:03:56.890 --> 00:03:59.590 align:middle line:84%
the overall cost-- at
least, the beginning costs--

00:03:59.590 --> 00:04:02.770 align:middle line:84%
so that we have
the win of avoiding

00:04:02.770 --> 00:04:05.860 align:middle line:84%
the downside, the win
of taking advantage

00:04:05.860 --> 00:04:08.990 align:middle line:90%
of the upside at a lesser cost.

00:04:08.990 --> 00:04:11.690 align:middle line:90%
So what do we then do?

00:04:11.690 --> 00:04:14.140 align:middle line:84%
So here's what I've
been suggesting--

00:04:14.140 --> 00:04:18.190 align:middle line:84%
the proactive design shifts
the distribution right.

00:04:18.190 --> 00:04:20.230 align:middle line:84%
So if you had your
base case design,

00:04:20.230 --> 00:04:23.290 align:middle line:84%
which was like this,
that somehow you

00:04:23.290 --> 00:04:25.450 align:middle line:84%
have gone through
things, which shift

00:04:25.450 --> 00:04:28.060 align:middle line:84%
the distribution towards
the positive side,

00:04:28.060 --> 00:04:30.280 align:middle line:84%
both by cutting
the downside tail

00:04:30.280 --> 00:04:33.220 align:middle line:84%
and by extending
the upside tail.

00:04:33.220 --> 00:04:36.358 align:middle line:84%
And you do that by
exploring the various ranges

00:04:36.358 --> 00:04:37.900 align:middle line:84%
of possible things
that you could do,

00:04:37.900 --> 00:04:41.900 align:middle line:84%
the flexibilities that
might help you do that.

00:04:41.900 --> 00:04:46.810 align:middle line:84%
So we want to end up, then, with
the win-win solution at lower

00:04:46.810 --> 00:04:51.370 align:middle line:84%
cost, less loss, more gain,
less investment compared

00:04:51.370 --> 00:04:54.660 align:middle line:84%
the alternatives, at
least at the start.

00:04:54.660 --> 00:04:58.365 align:middle line:84%
In a graph form, what we have
here is the curve that we have.

00:04:58.365 --> 00:05:01.410 align:middle line:84%
We try to chop the
downside, do things

00:05:01.410 --> 00:05:03.930 align:middle line:84%
that reduce the possibilities
that we have really

00:05:03.930 --> 00:05:06.390 align:middle line:84%
big losses, while
at the same time

00:05:06.390 --> 00:05:10.210 align:middle line:84%
being able to extend the upside
if things go really well.

00:05:10.210 --> 00:05:15.150 align:middle line:84%
So that is the essence of what
flexibility design attempts

00:05:15.150 --> 00:05:15.820 align:middle line:90%
to do.

00:05:15.820 --> 00:05:19.550 align:middle line:84%
And it very often
succeeds, thank goodness.

00:05:19.550 --> 00:05:25.000 align:middle line:90%