Seeking Wisdom

by Peter Bevelin

  We optimize one component at a time instead of optimizing the whole (what we finally want to accomplish). TransCorp forgot to consider how a change influences the whole system. Cost cutting doesn't automatically translate into higher value. TransCorp's decision to foe people caused manufacturing and delivery problems, which in turn caused delays in customer production. This created a loss of customers and reputation. The end result was lower profits.

  Why reduce prices? What is the purpose? What does TransCorp ultimately want to achieve?

  Systems adjust in response to feedback. A positive feedback amplifies an effect, while a negative dampens it. Take the stock market as an example of a positive feedback. The stock market falls causing a sell-off. This creates a ripple effect of further sell-off and price declines. The opposite occurs in a stock market bubble. A thermostat is an example of negative feedback.

  Try to optimize the whole and not asystem's individual parts. Think through what other variables may change when we alter a factor in a system. Trace out the short and long-term consequences in numbers and effects of a proposed action to see if the net result agrees with our ultimate goal.

  Ask: What key factors influence the outcome of the system and how do these factors interact? What other things may change as a consequence of some action? Given these conditions, what likely consequences (wanted and unwanted) will the

  122

  proposed action have on the system, considering all the relevant factors that influence or are part of the system? Will the net result be what we want? A manager can for example ask: How is the value of the business likely to change considering important factors that influence business value?‌

  Considering the whole includes anticipating the reactions of others.

  The reaction of others

  Game theory is a study of conflict between thoughtful and potential deceitful opponents.

  William Poundstone (from Prisoner's Dilemma)

  TransCorp cut the price and lost volume.

  What happened? TransCorp's competitors matched the price cut. Competitors can match price cuts or even go below it to regain, keep or increase market share. When thinking through consequences, consider what other people are likely to do. Since our interests may conflict with others, the final outcome of our decision often depends on what others will do. What other people do may

  depend on what they think we will do, their available choices, interests, and how they are thinking- including their misjudgments. As we have learned, humans don't always act rationally.

  Game theory deals with what happens when individuals or groups of people interact with one another to achieve their goals. We saw an example of game theory in Part One (Prisoner's Dilemma). It also applies to negotiations. Factors that decide the final outcome of a negotiation are: 1) the number of participants, 2) if we meet the participants again in the future, 3) the time lapse in between, 4) the degree of anonymity and communication, and 5) our relative position of strength which includes our other options, back-up alternatives and need to reach an agreement.

  The winner's curse

  I sent the club a wire stating, "Please accept my resignation.

  I don't care to belong to any club that will have me as a member. "

  Groucho Marx

  Several mining firms including MineCorp, one of TransCorps subsidiaries, are bidding on the right to mine silver.

  No firm knows for certain how much silver there is and hence what the true value is.They each hire an expert to make an educated guess. By definition, these expert guesses will range from too low to too high. Some firm's expert will probably be

  123

  dose. But they won't win. The winning firm was MineCorp since their expert had the most optimistic estimate of the value (the seller accepts the highest bid). But there was less silver in the mine than their expert guessed and less value than what MineCorp paid for the rights. This means that the winning bidder was cursed since the bid was higher than the value. Later it was shown that MineCorp also underestimated its costs of production.

  Three Atlantic Richfield engineers, Capen, Clapp and Campbell, introduced the idea behind the Winner's Curse when they did a study of companies bidding for oil fields. Their basic idea was Uournal of Petroleum Technology, June 1971), that "a lease winner tends to be the bidder who most overestimates reserve potential."

  Let's say TransCorp has 10 projects from 10 divisions to choose from. They only have time and money to invest in one project. Which one are they most likely to pick? Of course, the one that looks most attractive. But all division managers have an incentive to make their own project the most attractive one. The risk is therefore that TransCorp chooses the project with the most optimistic forecast and therefore more likely to disappoint.

  "Hurray, I won the auction!,"said John. "What you 'won' was the right to pay more for something than everyone else thought it was worth,"said Mary.

  Winning is an informative event, telling us whose estimate was most optimistic. When we place a bid on a house, company, project, or negotiate to buy something, we don't realize what is implied by an acceptance of our offer. That we may have overestimated its value and therefore paid too much.

  Research shows that the more bidders there are competing for a limited object, each having the same information, and the more uncertain its value is, the more likely we are to overpay. Instead, if our objective is to create value, the more bidders there are, the more conservative our bidding should be. This also implies that the less information we have compared to other bidders or the more uncertain we are about the underlying value, the lower we should bid. If we participate in auctions, we must ascertain the true value of what's being sold or its value to us.

  When we negotiate with one party and want an acceptance for an offer, the other party may have an informational advantage. The other party is most likely to accept our offer when it is least favorable to us, especially if it is a one-time relationship or if the other party is anonymous.

  Consider the seller's perspective. Ask: Why are they selling? How would I reason ifl think it through from the viewpoint of the other person? Why would I make a better decision than someone who has all the information?

  124

  Predictions

  Do not, therefore, expect any prophecy from me: had I known what one will discover tomorrow, I would have published it long ago, to secure priority.

  - Henri Poincare (French mathematician and scientist, 1854-1912)

  When asked about corporate responsibility for social problems, Charles Munger answered:

  I'm all for fixing social problems. I'm all for being generous to the less fortunate. And I'm all for doing things where, based on a slight preponderance of the evidence, you guess that it's likely to do more good than harm...

  What I'm against is being very confident and feeling that you know, for sure, that your particular intervention will do more good than harm given that you're dealing with highly complex systems wherein everything is interacting with everything else.

  Greek philosopher Heraclitus wrote: "Nothing endures but change." The world is too complicated to predict all the effects of some action. Maybe a business can predict scenarios like reduction in demand and intensified competition, but some events, their timing, magnitude or consequence, are impossible to anticipate.

  Mark Twain said: "The art of prophecy is very difficult, especially with respect to the future." It is hard to predict something when we don't (or can't) foresee or understand how an entire system works, what key variables are involved, their attributes, how they influence one another and their impact. Even if we know the key variables, their values may be impossible to estimate. They may also change over time and be dependent on context. It may also be impossible to estimate how they will interact as a whole.

  The more parts involved and the more they interact, the more can happen, and the harder it is to determine consequences of individual actions.

  According to Dr. Gerald Edelman, the brain is a
n example of a complex system:

  A complex system is one in which smaller parts form a heterogeneous set of components which are more or less independent. But as these parts connect with each other in larger and larger aggregates, their functions tend to become integrated, yielding new functions that depend on such high order integration. This is, in fact, just what happens in the brain.

  As the number of variables grows, the number of possible interactions grows even faster. Assume that two subsystems, A and B, cause the behavior of a system.

  125

  Each subsystem consists of 5 parts. If we only consider two-way interactions among parts, there are 10 interactions between A-parts, 10 between B-parts, and 25 interactions between A and B parts. This means that the behavior of the system is composed of 55 determinants (SA-parts + 5B-parts + 10 interactions between A parts + 10 interactions between B-parts + 25 interactions between A and B parts). 18% (10 of 55) of all determinants derive from individual effects of parts in A and B while about 82% (45 of 55) derive from interactions. Now imagine a system where A and B each consists of 100 parts. There are now 20,100 determinants (100+100+4,950+4,950+10,000) and 19,900 interactions meaning that 99% (19,900 of20,100) of system determinants derive from interactions.

  We often take too little notice of how variables interact. Take the economy as an example. There are many factors to consider. They include interest rates, currency exchange rates, balance of trade figures, unemployment rates, consumer confidence, political factors, the stock market, business cycles, biases, etc. These factors are interconnected, and it is hard to tell which is most important. Add to this that people's behavior isn't fixed. We are emotional creatures, our preferences change, and we react to each other's actual or expected decisions. A prediction may also make us change our expectations and behavior, making the prediction more or less likely to come true.

  Charles Munger says: "We try and predict what individual investments will swim well in relation to the tide. And then we tend to accept the effects of the tide as those effects fall."

  "Ifsomeone could forecast the stock market, why are they selling advice through $100 newsletters?"

  Fidelity's former manager Peter Lynch said in One Up on Wall Street: "There are 60,000 economists in the U.S., many of them employed full-time trying to forecast recessions and interest rates, and if they could do it successfully twice in a row, they'd all be millionaires by now... As far as I know, most of them are still gainfully employed, which ought to tell us something."

  Predictions about the future are often just projections of past curves and present trends. This is natural since our predictions about the future are made in the present. We therefore assume the future will be much like the present. But the future can't be known until it arrives. It is contingent on events we can't see. For example, who in the year 1900, could foresee events like World War I and II, the 1929 stock market crash, Chernobyl or technologies like the television, laser, computer, internet or the DVD? Many key inventions happened by accident and sagacity. For example, in 1867 Alfred Nobel accidentally discovered that when

  126

  nitroglycerin dripped on kieselguhr (a mineral made of small, fossilized remains of sea animals), it formed a stable paste that was safer to use than liquid nitroglycerin alone. He called it dynamite.

  Don't believe people that say they can forecast unforeseeable variables. Nobody can forecast interest or currency rates, the GDP, turning points in the economy, the stock market, etc. Massive amounts of information, advanced computers or fancy mathematical formulas don't help. Warren Buffett says that we tend to put too much comfort in computer models and the precision they project: "We believe the precision they project is a chimera. In fact, such models can lull decision-makers into a false sense of security and thereby increase their chances of making a really huge mistake."

  Economics isn't like physics. There are no reliable or precise formulas where we easily can fill in the values of various economic factors, and then have the work done. Charles Munger says: "Economics involves too complex a system... economics should emulate physics' basic ethos, but its search for precision in physics-like formulas is almost always wrong in economics." J.M. Keynes adds: "To convert a model into a quantitative formula is to destroy its usefulness as an instrument of thought."

  Financial writer Roger Lowenstein writes in When Genius Failed: "The next time a Merton [Robert Merton, 1997 Nobel Laureate for developing mathematical risk management formulas] proposes an elegant model to manage risks and foretell odds, the next time a computer with a perfect memory of the past is said to quantify risks in the future, investors should run - and quickly- the other way."

  That an event has happened many times before, doesn't mean that it will continue to happen. And just because an event has never happened before, doesn't mean it can't happen in the future. Take catastrophic events as an example. Who could have predicted the September 11, 2001 terrorist attack on the World Trade Center? Hijacking four planes simultaneously and using them to attack the

  U.S. was improbable. Yet it happened.

  Modern History Professor Richard Evans wrote in In Defence of History: "Time and again, history has proved a very bad predictor of future events. This is because history never repeats itself; nothing in human society... ever happens twice under exactly the same conditions or in exactly the same way."

  Sometimes we can guess that certain things are bound to happen, but we can't predict when they will happen.

  Will it rain two weeks from now?

  Some things are possible to predict short-term but impossible to predict long-

  127

  term. Small changes make a big difference over time. Long-term weather forecasts are an example. Many factors determine the weather. Factors that can't be reliably measured ahead of time. Small changes in the temperature and pressure over the ocean can lead to large variations in the future development of storm systems. Weather predictions become more inexact the further out they are.

  The difficulty lies in initial condition uncertainties and model error. For example, small errors in the initial values ofvariables can grow and produce errors in the forecast. There may also be gaps in the initial data. But even if we know the initial conditions perfectly, the models are not perfect. Small model errors either in physics or numerics can also grow and produce different states. For example, not all atmospheric processes are understood. Furthermore, all weather models operate on a finite grid or limited area, typically in the order of 10 to 100 kilometers, depending on the study area. This means that the numerical resolution and representation is finite. But many physical processes and features that influence the weather occur on smaller scales than those resolved by the grid. For example, energy transfers at the surface, small atmospheric processes (such as isolated thunderstorms), topography, lakes, and vegetation. The model must treat or "parameterize" the effects of these "sub-grid" features' effects on the resolved scale. These parameterizations are simplifications and approximations and may also account for many of the model errors. So even if we know all the principles behind the weather and what governs the atmosphere, fundamental limitations make it hard to make accurate predictions.

  Many meteorologists know they can't make a perfect forecast and have

  therefore given up on predicting whether it will or will not rain beyond a few days into the future. Instead, they have altered their approach and try to quantify the uncertainty in the prediction ("The probability of precipitation is 20% this Saturday''). This uncertainty is small at short lead times and greater at longer lead times. It also varies with the weather situation, the location and the size of the area the prediction covers. When meteorologists make forecasts two weeks out, they look at the climatological frequency of precipitation, determined from a history of what has happened in the past.

  All prediction is inherently uncertain, and we have a duty to tell people about the uncertainties of our predictions and our past error rates. Albert Einstein wrote in a Mar
ch 14, 1954, letter: "The right to search for truth... implies also a duty; one must not conceal any part of what one has recognized to be true."

  128

  - Two -‌

  SCALE AND LIMITS

  Scale of size and time

  Changes in size or time influences form, function and behavior. If something of a certain size is made bigger or smaller, it may not work the same way. Some things get better and others get worse. For example, changes in the size of an organism affect its strength, surface area, complexity, metabolism, longevity, and speed of movement.

  How do weight, strength and surface area change when we change size?

  If we double the length of a similarly shaped object, surface area increases 4 times and volume 8 times. Surface area increases at the square of length and volume at the cube oflength (to get areas we multiply two lengths together, and to get volumes we multiply three lengths).

  Volume always grows faster than surface area as we increase size, independent of an object's shape. This places limitations on the size of things.

  Does it make any difference in melting time if we use one ice cube (volume 8) or 8 smaller ones (total volume 8}?

  If we increase size, volume grows faster than surface area. What does this mean? The relationship between surface area and volume decreases when we increase the size. It also means that the relationship increases when we decrease the size. Take ice cubes as an example. Let's assume the larger one has a side length of 2 and the smaller ones a side length of 1.