http://brontecapital.blogspot.com/2009/12/kodak-bill-gates-and-efficient-markets.html
"Warren Buffett has a group of his best investing friends get together once a year...One regular attendee was Bill Gates of Microsoft fame. From here I will quote Alice Schroder:
After a while Buffett asked everyone to pick their favourite stock.
What about Kodak? asked Bill Ruane. He looked back at Gates to see what he would say.
"Kodak is toast,” said Gates.
Nobody else in the Buffett Group knew that the internet and digital technology would make film cameras toast. In 1991, even Kodak didn’t know it was toast."
Kodak did eventually go into a long decline that continues until today - but not before tripling in value. As John Hempton points out, even though Gates was right, he looked stupid for years before he looked smart.
This story illustrates something I've been thinking about for awhile. It's not always enough to be right - sometimes you have to also be right about the right time, and that's very, very hard.
There's a great set of books by Isaac Asimov called the Foundation Series, set many thousand years in the future, with the plot revolving around a mathematician named Hari Seldon who invents the discipline of psychohistory. Psychohistory is a branch of mathematical sociology that can predict the probabilistic future on a large scale based on mass psychology, but is error-prone on a small scale. Seldon foresees the fall of the current galactic empire and a subsequent 30,000-year period of dark-age barbarism. The series revolves around his attempt to reduce the term of this dark age to a single millenium.
Like Seldon, we seek to forecast the future as well. (Being a futurist, apparently, has become an actual paying job these days, with business cards and everything: http://en.wikipedia.org/wiki/List_of_futurologists)
Anyone that has ever tried to forecast sales, earnings or construction timelines knows, however, that most forecasts are "magic" and not the happy, sparkly kind. It's nice to put the numbers down on paper but it doesn't mean they're right. The other word often used for these types of forecasts is "bullshit."
In truth, the areas in which we are reasonably good at forecasting temporally accurate real-world outcomes tend to be very narrow in scope (e.g. tomorrow's weather, next week's election result). We aren't especially good at broader scopes (e.g. the weather next year on a certain day, next year's election outcomes, when a financial bubble will pop). This is despite economists' and statisticians' well-practiced ability to perform interesting correlations and regressions until our head spins and we all want to vomit. You start to wonder whether the word "significant" means something different for them than for the rest of us.
Predicting the future is hard, and predicting when that future will happen is even harder - even for a very smart guy like Bill Gates casting a knowledgeable eye on a fellow tech firm. A number of people, including some good friends (you know who you are and this is my begrudging nod to you), knew that the recent financial crisis was coming - but hardly anyone knew when exactly. It's nice to know but it doesn't help much unless there's a forward-tested predictive model behind it (which is hard to do for a variety of reasons, small recessions sample, unique conditions, yada yada). Even if you were right about what and when, how would we have known whether to believe you?
I wonder whether it's time we take Asimov's idea of psychohistory seriously and turn it into a real discipline. I don't agree with his fatalistic determinism - it's difficult to predict individual action and chaotic systems can be sensitive to small changes - and his time scale seems quite challenging as well, but there's no history without psychology and mass psychology is probably not completely random. If we can ascertain patterns, such as, for instance, how and how long it takes for groups to absorb new information, mass reactions and reaction times to new-paradigm change, levels of trust/corruption in extreme social-inequality situations, etc - then perhaps we can use this information to change our future. Or at least make some money on the stock market.
"Warren Buffett has a group of his best investing friends get together once a year...One regular attendee was Bill Gates of Microsoft fame. From here I will quote Alice Schroder:
After a while Buffett asked everyone to pick their favourite stock.
What about Kodak? asked Bill Ruane. He looked back at Gates to see what he would say.
"Kodak is toast,” said Gates.
Nobody else in the Buffett Group knew that the internet and digital technology would make film cameras toast. In 1991, even Kodak didn’t know it was toast."
Kodak did eventually go into a long decline that continues until today - but not before tripling in value. As John Hempton points out, even though Gates was right, he looked stupid for years before he looked smart.
This story illustrates something I've been thinking about for awhile. It's not always enough to be right - sometimes you have to also be right about the right time, and that's very, very hard.
There's a great set of books by Isaac Asimov called the Foundation Series, set many thousand years in the future, with the plot revolving around a mathematician named Hari Seldon who invents the discipline of psychohistory. Psychohistory is a branch of mathematical sociology that can predict the probabilistic future on a large scale based on mass psychology, but is error-prone on a small scale. Seldon foresees the fall of the current galactic empire and a subsequent 30,000-year period of dark-age barbarism. The series revolves around his attempt to reduce the term of this dark age to a single millenium.
Like Seldon, we seek to forecast the future as well. (Being a futurist, apparently, has become an actual paying job these days, with business cards and everything: http://en.wikipedia.org/wiki/List_of_futurologists)
Anyone that has ever tried to forecast sales, earnings or construction timelines knows, however, that most forecasts are "magic" and not the happy, sparkly kind. It's nice to put the numbers down on paper but it doesn't mean they're right. The other word often used for these types of forecasts is "bullshit."
In truth, the areas in which we are reasonably good at forecasting temporally accurate real-world outcomes tend to be very narrow in scope (e.g. tomorrow's weather, next week's election result). We aren't especially good at broader scopes (e.g. the weather next year on a certain day, next year's election outcomes, when a financial bubble will pop). This is despite economists' and statisticians' well-practiced ability to perform interesting correlations and regressions until our head spins and we all want to vomit. You start to wonder whether the word "significant" means something different for them than for the rest of us.
Predicting the future is hard, and predicting when that future will happen is even harder - even for a very smart guy like Bill Gates casting a knowledgeable eye on a fellow tech firm. A number of people, including some good friends (you know who you are and this is my begrudging nod to you), knew that the recent financial crisis was coming - but hardly anyone knew when exactly. It's nice to know but it doesn't help much unless there's a forward-tested predictive model behind it (which is hard to do for a variety of reasons, small recessions sample, unique conditions, yada yada). Even if you were right about what and when, how would we have known whether to believe you?
I wonder whether it's time we take Asimov's idea of psychohistory seriously and turn it into a real discipline. I don't agree with his fatalistic determinism - it's difficult to predict individual action and chaotic systems can be sensitive to small changes - and his time scale seems quite challenging as well, but there's no history without psychology and mass psychology is probably not completely random. If we can ascertain patterns, such as, for instance, how and how long it takes for groups to absorb new information, mass reactions and reaction times to new-paradigm change, levels of trust/corruption in extreme social-inequality situations, etc - then perhaps we can use this information to change our future. Or at least make some money on the stock market.
RSS Feed