Re: [asa] Trees don't lie

From: George Murphy <>
Date: Mon Mar 23 2009 - 22:02:49 EDT

Yes, & that's why I said "pretty good statistically."


----- Original Message -----
From: "Glenn Morton" <>
To: "George Murphy" <>
Cc: "asa" <>
Sent: Monday, March 23, 2009 8:57 PM
Subject: Re: [asa] Trees don't lie

>I think one of the things that makes expert guesses, even with physics so
>ineffectual is the observation (of someone whose quotation I can't find
>right now), that one can't predict the trajectory of a ball in a
>gravitational field unless you can predict what life will do. A batter
>hits the ball, and you can predict the trajectory up until the point at
>which a hand in a glove attached to a living being catches it and changes
>the trajectory.
> Any system which has life in it, is hugely unpredictable.
> ----- Original Message -----
> From: "George Murphy" <>
> To: "Glenn Morton" <>
> Cc: "asa" <>
> Sent: Monday, March 23, 2009 8:51 AM
> Subject: Re: [asa] Trees don't lie
>> Glenn -
>> Very quickly, & as far as I have anything to contribute winding this up
>> for now -
>> It's necessary distinguish between two things. (a) The basic dynamics
>> of a complex system which, to the extent we can model it mathematically,
>> require nonlinear equations, and (b) the analysis of some important data
>> about the system which may use various math approximations. Think of the
>> difference between Newton's laws and Kepler's. Many (though not all)
>> nonlinear systems display sensitivity to initial conditions so that we
>> can't predict the future state (i.e. the precise values of positions &
>> velocities for mechanical systems) far into the future. But we can
>> analyze data using various approximations and assumptions in order to
>> estimate certain important facts about the future bahavior of the system.
>> The latter seems to me to be what Hubbert did quite successfully.
>> (BTW, it's easy from the math at the site I gave to see why the peak
>> occurs at the point where half the initial reserve has been exhausted -
>> in this approximation of course.)
>> About the other systems (elections, MLB) I didn't say I "wanted"
>> equations. I said earlier that I don't think we can have a complete math
>> description of such things. But to the extent we do, or can use a math
>> metaphor, they're nonlinear. & there are knowledgeable people who can
>> make predictions that are not perfect but are pretty good statistically.
>> They don't do that by solving equations though. I agree with Plato that
>> God is a mathematician but he has other interests as well!
>> Shalom
>> George
>> ----- Original Message -----
>> From: "Glenn Morton" <>
>> To: "George Murphy" <>
>> Cc: "asa" <>
>> Sent: Sunday, March 22, 2009 10:56 PM
>> Subject: Re: [asa] Trees don't lie
>>> Hi George,
>>>> Again you of course know a lot more about oil production than I. But
>>>> when I take a quick look at a discussion of Hubbert's math at
>>>> I see a slightly wavy curve of P/Q vs Q and the phrase "Let us fit a
>>>> straight line to this set of dots from 1958 on ..." . Now I understand
>>>> that Hubbert's original analysis was more complex but what's being done
>>>> here looks like making the simplest - linear - mathematical description
>>>> of what is really a pretty complex phenomenon in order to predict, not
>>>> the precise future state of the system but one crucial feature of
>>>> particular interest. I would be very surprised if the real dynamics of
>>>> oil production - if we could have a precise mathematical description -
>>>> is really describable with strictly linear equations. If it is, please
>>>> enlighten my ignorance by showing me that it is or referring me to an
>>>> appropriate source.
>>> .
>>> I will absolutely grant that you know more math than I. But, Hubbert's
>>> equation works and it is linear. The situation in oil production is
>>> complex only because of political decisions made by country leaders to
>>> restrict production or open the taps. Hubbert predicted a world peak
>>> oil in 1995. The only reason that that prediction didn't happen was
>>> because in the late 1970s and early 1980s, the world became incredibly
>>> more energy efficient due to the high prices of oil. But all that did
>>> was push the peak back about 10-15 years. I think we have peaked
>>> because of the current down turn. We will never catch up again with the
>>> decline because we are not drilling as much now. we were barely keeping
>>> up with production when we were drilling all out. Now that we aren't,
>>> world oil supply, defined as the amount of oil coming to market per day
>>> will decline rather quickly. Buckle your belts.
>>>> Similarly for the other systems I mentioned. What is the justification
>>>> for saying that the dynamics of a presidential election or a MLB season
>>>> are linear? To the extent that the term is meaningful, I suspect that
>>>> the dynamics of all complex systems involving human beings are
>>>> "nonlinear." A group of 50 Indians confronting 50 Pakistanis isn't
>>>> just 50 times a 1 on 1 meeting between an Indian and a Pakistani.
>>> I didn't say that the presidential election was linear.
>>>> In discussing elections it's significant that you put "equation" in
>>>> quotes - and never actually write down any. That's because you don't
>>>> have any. You have relationships that to some extent are
>>>> semi-quantitative but you don't have A = B relationships. & unless you
>>>> do, the terms "linear" & "nonlinear" can be used only in the loose (but
>>>> I hope meaningful) way I did in the previous paragraph. & since you
>>>> don't have real equations, the concepts of chaos theory can also be
>>>> applied only in a loose way - which is not to say that they might not
>>>> be of some value if you're careful.
>>> George, those are the kinds of systems that Tetlock studied--those
>>> lacking the kinds of equations you seem to want. That is why experts
>>> aren't any good.

To unsubscribe, send a message to with
"unsubscribe asa" (no quotes) as the body of the message.
Received on Mon Mar 23 22:04:02 2009

This archive was generated by hypermail 2.1.8 : Mon Mar 23 2009 - 22:04:02 EDT