10.07.2015 Views

FLEXIBILITY IN DESIGN - Title Page - MIT

FLEXIBILITY IN DESIGN - Title Page - MIT

FLEXIBILITY IN DESIGN - Title Page - MIT

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

de Neufville + Scholtes D R A F T September 30, 2009scenario. We can only correctly evaluate the real results of a system by looking at the range ofthe scenarios.An example illustrates the phenomenon. Imagine a game in which you roll a die and willreceive $3 if the resulting number is 3 or more -- and nothing otherwise. Assuming that all thepossibilities from 1 to 6 are equally likely, the average number showing will be 3.5. If youcalculate your payoff based on this average number, you obtain a payoff of $3. However, theaverage payoff is only $2! This is because there is a 1 in 3 chance of rolling a 1 or a 2, whichleads to a payoff of $0 instead of the $3 calculated based on the average number on the dice.This loss in the “downside scenario” is not counterbalanced by additional gains in the equallylikely “upside scenario” of rolling a 5 or a 6, in which case you still only get $3. You lose in thedownside but do not gain in the upside scenario.This example is widely replicated in industry whenever it has to plan production capacity.Auto manufacturers, for instance, typically have designed plant sizes based on their most likelyforecasts of sales of new models. If the model is a success and demand is greater than expected,they can expand capacity to a limited degree (and substantial cost) by working overtime or extrashifts. However, if demand is poor, then they definitely have big losses. (See Box 2.1 for practicalapplication to the design of a xxxxx new example, Stefan to develop.)[Box 2.1 about here]The flaw of averages effect is easily understood – and remarkably often neglected inplanning. Upsides and downsides of input uncertainties (numbers on the dice, demand for cars,use of parking spaces) will nicely balance out around their average, but these upsides anddownsides will often have asymmetric effects on the measure of value for a system performance(such as its net present value). We simply cannot expect system performance to balance outaround the performance calculated based on average scenarios.The need for forecastsForecasts are fundamental to design. We plan and implement systems around anticipated futuredemands and opportunities. Our estimates of what we will need shape what we build, where welocate it, and when we implement it. These projections focus our thinking and determine themajor characteristics of any design.Meanwhile, the value of any design derives from how well it matches the actualrealizations of future demands and opportunities. A system has value to the extent that it matchesongoing needs, at the right time and place. A “bridge to nowhere” may be technically brilliant,structurally elegant, and constructed efficiently and safely. However, if it leads nowhere, serveslittle traffic, and thus does not fulfill a meaningful function – it has little value. The development ofthe Iridium system to provide worldwide telephony illustrates the point. The deployment of thisfleet of satellites that could talk both to each other and to any point on the planet was anPart 1: Chapters 1 to 3 <strong>Page</strong> 23 of 69

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!