WEBVTT

00:00:00.000 --> 00:00:00.976 align:middle line:90%


00:00:00.976 --> 00:00:02.440 align:middle line:90%
[SQUEAKING]

00:00:02.440 --> 00:00:03.416 align:middle line:90%
[RUSTLING]

00:00:03.416 --> 00:00:05.368 align:middle line:90%
[CLICKING]

00:00:05.368 --> 00:00:18.580 align:middle line:90%


00:00:18.580 --> 00:00:20.200 align:middle line:84%
RICHARD DE NEUFVILLE:
Now, when we

00:00:20.200 --> 00:00:22.480 align:middle line:84%
had the move, the
possibility to say, OK,

00:00:22.480 --> 00:00:25.690 align:middle line:84%
we can relocate some of these
things to a distant place

00:00:25.690 --> 00:00:31.300 align:middle line:84%
and avoid some of the
dispatch and transport costs,

00:00:31.300 --> 00:00:33.700 align:middle line:84%
the solution there
was, OK, maybe we

00:00:33.700 --> 00:00:35.590 align:middle line:84%
want to move it to the
most distant place,

00:00:35.590 --> 00:00:37.510 align:middle line:90%
and we're going to do it--

00:00:37.510 --> 00:00:41.740 align:middle line:84%
the gas, we're going to
transfer it by pipeline and not

00:00:41.740 --> 00:00:46.210 align:middle line:84%
by truck, by tanker truck
of the finished product

00:00:46.210 --> 00:00:48.700 align:middle line:84%
so that that could
save us some money.

00:00:48.700 --> 00:00:55.330 align:middle line:84%
And indeed, overall, having
this solution for the numbers

00:00:55.330 --> 00:00:57.880 align:middle line:84%
we had increased the expected
value-- in this case,

00:00:57.880 --> 00:01:00.850 align:middle line:90%
by about another 10%.

00:01:00.850 --> 00:01:02.920 align:middle line:84%
But here's the
kind of thing that

00:01:02.920 --> 00:01:05.740 align:middle line:84%
happened is that,
in this case, to be

00:01:05.740 --> 00:01:08.350 align:middle line:84%
able to enable this
flexibility, you

00:01:08.350 --> 00:01:11.230 align:middle line:90%
had to invest in a pipeline.

00:01:11.230 --> 00:01:14.020 align:middle line:90%
And that would cost some money.

00:01:14.020 --> 00:01:16.810 align:middle line:84%
So under many cases,
if you invested in it

00:01:16.810 --> 00:01:19.520 align:middle line:84%
but didn't use it, you'd
have this extra cost

00:01:19.520 --> 00:01:25.540 align:middle line:84%
so that this
pipeline solution was

00:01:25.540 --> 00:01:27.830 align:middle line:90%
a disadvantage for all these.

00:01:27.830 --> 00:01:32.500 align:middle line:84%
They would reduce the expected
value, say, from about 11

00:01:32.500 --> 00:01:35.140 align:middle line:84%
here for both of the
alternatives to something

00:01:35.140 --> 00:01:37.180 align:middle line:90%
around, let's call it, 8 or 9.

00:01:37.180 --> 00:01:40.100 align:middle line:84%
So it was significant
over a range here.

00:01:40.100 --> 00:01:43.720 align:middle line:84%
But on the other hand, it
was really great demand.

00:01:43.720 --> 00:01:46.060 align:middle line:90%
It was a very profitable one.

00:01:46.060 --> 00:01:48.560 align:middle line:90%
So you had a tradeoff to happen.

00:01:48.560 --> 00:01:51.070 align:middle line:84%
So that very often,
as in this case,

00:01:51.070 --> 00:01:52.930 align:middle line:84%
if you would simply
comparing the red line

00:01:52.930 --> 00:01:57.250 align:middle line:84%
and the green line, it's not so
obvious which is the better one

00:01:57.250 --> 00:01:59.680 align:middle line:90%
or how you ought to organize it.

00:01:59.680 --> 00:02:02.440 align:middle line:84%
So it does frequently
happen, and you

00:02:02.440 --> 00:02:04.660 align:middle line:84%
should expect it
that it's not always,

00:02:04.660 --> 00:02:06.820 align:middle line:90%
well, one is uniformly better.

00:02:06.820 --> 00:02:13.720 align:middle line:84%
This kind of effect where
investing in the flexibility

00:02:13.720 --> 00:02:15.730 align:middle line:90%
has a cost.

00:02:15.730 --> 00:02:19.480 align:middle line:84%
And just like I
buy life insurance

00:02:19.480 --> 00:02:20.800 align:middle line:90%
not to be morbid about it.

00:02:20.800 --> 00:02:24.130 align:middle line:84%
I buy life insurance,
and every year goes by,

00:02:24.130 --> 00:02:27.620 align:middle line:84%
I've lost money on my bet,
but I'm very-- in this case,

00:02:27.620 --> 00:02:32.510 align:middle line:84%
I'm very happy with not cashing
in on my life insurance.

00:02:32.510 --> 00:02:38.020 align:middle line:84%
So this has some
comparisons, and it

00:02:38.020 --> 00:02:40.960 align:middle line:84%
shows that the
flexible with no move

00:02:40.960 --> 00:02:45.610 align:middle line:84%
can increase not just the
expected value but also

00:02:45.610 --> 00:02:48.280 align:middle line:84%
other kinds of measures which
I suggest that we should

00:02:48.280 --> 00:02:51.020 align:middle line:84%
be taking into
account regularly,

00:02:51.020 --> 00:02:53.660 align:middle line:84%
which is, for example,
the value at risk.

00:02:53.660 --> 00:02:58.330 align:middle line:84%
So both the flexible
with and without the move

00:02:58.330 --> 00:03:00.700 align:middle line:90%
increased the--

00:03:00.700 --> 00:03:05.770 align:middle line:84%
improved the value at
risk by 2 or 3 times

00:03:05.770 --> 00:03:12.970 align:middle line:84%
here in comparison with what
was happening with the fixed

00:03:12.970 --> 00:03:13.480 align:middle line:90%
design.

00:03:13.480 --> 00:03:19.060 align:middle line:84%
So the fixed design, you
had a 10% value risk of 1.8.

00:03:19.060 --> 00:03:23.620 align:middle line:84%
And here, you increased
it to 5 or 4, basically.

00:03:23.620 --> 00:03:25.570 align:middle line:84%
And the same thing
with the upside.

00:03:25.570 --> 00:03:27.920 align:middle line:84%
You had considerable
possible improvement.

00:03:27.920 --> 00:03:30.790 align:middle line:84%
So it's not just a
unidimensional thing.

00:03:30.790 --> 00:03:34.060 align:middle line:84%
It changes not just the
averages but the distribution

00:03:34.060 --> 00:03:37.720 align:middle line:84%
of averages, on the one
hand reducing the risk

00:03:37.720 --> 00:03:39.550 align:middle line:90%
and increasing the opportunity.

00:03:39.550 --> 00:03:42.800 align:middle line:84%
Those are the kinds of
things that you would expect.

00:03:42.800 --> 00:03:45.280 align:middle line:84%
Now let's think
about learning rates.

00:03:45.280 --> 00:03:50.140 align:middle line:84%
And so being a parametric
study, not knowing

00:03:50.140 --> 00:03:52.870 align:middle line:84%
what the learning rate might be
on this kind of construction,

00:03:52.870 --> 00:03:57.580 align:middle line:84%
we assumed it was either nothing
or it had different rates.

00:03:57.580 --> 00:04:00.880 align:middle line:84%
And what you can
see is, again, this

00:04:00.880 --> 00:04:03.570 align:middle line:84%
is part of the tradeoff
between economy

00:04:03.570 --> 00:04:05.120 align:middle line:90%
of the scale and learning.

00:04:05.120 --> 00:04:07.990 align:middle line:84%
If you're learning
as you build modules,

00:04:07.990 --> 00:04:10.870 align:middle line:84%
the later ones will be
cheaper than the earlier ones.

00:04:10.870 --> 00:04:15.400 align:middle line:84%
And so that it will decrease
the overall cost of expansion,

00:04:15.400 --> 00:04:23.290 align:middle line:84%
and this counterbalances the
effect of economies of scale

00:04:23.290 --> 00:04:27.460 align:middle line:84%
and increases the effect
of discount rates,

00:04:27.460 --> 00:04:30.260 align:middle line:84%
meaning the further you
build it out in the future,

00:04:30.260 --> 00:04:32.180 align:middle line:84%
the less expensive
it's going to be.

00:04:32.180 --> 00:04:37.190 align:middle line:90%
And as you can see here--

00:04:37.190 --> 00:04:41.720 align:middle line:84%
I was trying to illustrate
here that, for the no move

00:04:41.720 --> 00:04:44.720 align:middle line:84%
analysis simply looking
at changing the learning

00:04:44.720 --> 00:04:46.910 align:middle line:84%
rate on the construction
of the modules,

00:04:46.910 --> 00:04:50.840 align:middle line:84%
it pushed out the maximum
value you can have,

00:04:50.840 --> 00:04:53.330 align:middle line:84%
the far end of
the curve, meaning

00:04:53.330 --> 00:04:56.960 align:middle line:84%
that if demand really grew
and you build a lot but you

00:04:56.960 --> 00:05:00.500 align:middle line:84%
build it in the future and it
basically came to you at a very

00:05:00.500 --> 00:05:04.880 align:middle line:84%
cheap price, you had
tremendous advantages in terms

00:05:04.880 --> 00:05:09.170 align:middle line:84%
of the expected value going
from about 20 to about 50

00:05:09.170 --> 00:05:11.240 align:middle line:84%
in this particular case
with these numbers.

00:05:11.240 --> 00:05:14.180 align:middle line:84%
And I wouldn't put much
attention to these numbers

00:05:14.180 --> 00:05:17.150 align:middle line:84%
because they're not
backed by a specific case.

00:05:17.150 --> 00:05:18.470 align:middle line:90%
But it in shows--

00:05:18.470 --> 00:05:20.450 align:middle line:84%
and the intent of
this is to show

00:05:20.450 --> 00:05:23.990 align:middle line:84%
that there is that
distribution of change,

00:05:23.990 --> 00:05:27.240 align:middle line:84%
and the learning rate can
be really very powerful.

00:05:27.240 --> 00:05:30.110 align:middle line:84%
So even if it goes
from zero, none,

00:05:30.110 --> 00:05:32.940 align:middle line:84%
to some, 5%, which
is fairly modest,

00:05:32.940 --> 00:05:35.900 align:middle line:90%
it has a change from 20 to 28.

00:05:35.900 --> 00:05:39.350 align:middle line:84%
So it has a lot of
leverage in there

00:05:39.350 --> 00:05:42.140 align:middle line:90%
if there is a learning rate.

00:05:42.140 --> 00:05:44.210 align:middle line:84%
And this-- if
you're making cars,

00:05:44.210 --> 00:05:47.510 align:middle line:84%
you're going from having
produced 100,000 to 110,000.

00:05:47.510 --> 00:05:48.780 align:middle line:90%
It's not very much.

00:05:48.780 --> 00:05:52.580 align:middle line:84%
But if you're producing big
units such as in the Victoria

00:05:52.580 --> 00:05:55.970 align:middle line:84%
desalting and you go from
building one, your first time,

00:05:55.970 --> 00:05:59.120 align:middle line:84%
you didn't know how to do it, to
building a second and a third,

00:05:59.120 --> 00:06:02.360 align:middle line:84%
the effect can be
really quite significant

00:06:02.360 --> 00:06:04.760 align:middle line:84%
when you're dealing in
low numbers of units

00:06:04.760 --> 00:06:06.800 align:middle line:84%
when the modules
you're talking about

00:06:06.800 --> 00:06:11.270 align:middle line:90%
are in the units, single units--

00:06:11.270 --> 00:06:13.980 align:middle line:84%
1, 2, 3, 4, 5-- or
even in the teens.

00:06:13.980 --> 00:06:17.090 align:middle line:84%
So this can have a
huge effect, which

00:06:17.090 --> 00:06:20.030 align:middle line:90%
is the point of this graph.

00:06:20.030 --> 00:06:26.380 align:middle line:84%
So the general lesson
3 is the repetition

00:06:26.380 --> 00:06:29.590 align:middle line:84%
of modules promotes
learning, which

00:06:29.590 --> 00:06:33.560 align:middle line:84%
is most impactful for
initial replications.

00:06:33.560 --> 00:06:37.720 align:middle line:84%
So learning can increase
value significantly.

00:06:37.720 --> 00:06:40.300 align:middle line:84%
Let me repeat the
example that I think

00:06:40.300 --> 00:06:43.240 align:middle line:84%
I cited earlier in the class
but I want to emphasize

00:06:43.240 --> 00:06:52.930 align:middle line:84%
in this context, is when BP was
building platforms in the ROC

00:06:52.930 --> 00:06:57.040 align:middle line:84%
near Azerbaijan
their first they had

00:06:57.040 --> 00:07:00.910 align:middle line:84%
for a variety of circumstances
having to do with contract

00:07:00.910 --> 00:07:03.480 align:middle line:84%
disputes with designers
and what have you,

00:07:03.480 --> 00:07:08.350 align:middle line:84%
they ended up building three
identical units of platforms,

00:07:08.350 --> 00:07:11.680 align:middle line:84%
which was very different from
their traditional approach

00:07:11.680 --> 00:07:16.660 align:middle line:84%
was to optimize, to have
bespoke, tailor-made platforms

00:07:16.660 --> 00:07:20.140 align:middle line:84%
for different sites where they
had different expectations

00:07:20.140 --> 00:07:23.590 align:middle line:84%
of the fuel coming through
or the product coming

00:07:23.590 --> 00:07:26.260 align:middle line:84%
through at different times,
different depths of water,

00:07:26.260 --> 00:07:27.310 align:middle line:90%
and so forth.

00:07:27.310 --> 00:07:31.030 align:middle line:84%
And in that case is
the savings and labor

00:07:31.030 --> 00:07:33.130 align:middle line:90%
costs were reported to be--

00:07:33.130 --> 00:07:35.650 align:middle line:84%
I reported-- I don't have
the actual data for me,

00:07:35.650 --> 00:07:38.680 align:middle line:84%
but it was reported
to be about 40%.

00:07:38.680 --> 00:07:40.750 align:middle line:84%
People knew what was
going to happen next.

00:07:40.750 --> 00:07:43.130 align:middle line:90%
They can schedule things better.

00:07:43.130 --> 00:07:48.130 align:middle line:84%
They changed little bits of
the way-- the order in which

00:07:48.130 --> 00:07:48.940 align:middle line:90%
they did assembly.

00:07:48.940 --> 00:07:53.320 align:middle line:84%
Anyway, it was like
you and me, if we ever

00:07:53.320 --> 00:07:55.520 align:middle line:84%
did this-- which I
have, and many of you

00:07:55.520 --> 00:07:58.270 align:middle line:84%
have-- put out a
tent to go camping.

00:07:58.270 --> 00:08:00.790 align:middle line:84%
The first time you do
it, you make mistakes

00:08:00.790 --> 00:08:02.305 align:middle line:84%
and don't know
what you're doing.

00:08:02.305 --> 00:08:04.180 align:middle line:84%
And by the time you've
done it several times,

00:08:04.180 --> 00:08:06.970 align:middle line:90%
you're a lot more efficient.

00:08:06.970 --> 00:08:10.840 align:middle line:84%
And maybe you can think a
lot of other examples of when

00:08:10.840 --> 00:08:13.420 align:middle line:84%
you go from the first time
to the second or third times,

00:08:13.420 --> 00:08:16.480 align:middle line:84%
you get to know the
ropes, and it happens.

00:08:16.480 --> 00:08:20.530 align:middle line:84%
So that's the general
lesson is that this can have

00:08:20.530 --> 00:08:22.180 align:middle line:90%
quite significant impacts--

00:08:22.180 --> 00:08:24.220 align:middle line:90%
can have, may have.

00:08:24.220 --> 00:08:28.440 align:middle line:90%
It is not, of course, necessary.

00:08:28.440 --> 00:08:39.384 align:middle line:90%