Talk:Malthusian catastrophe

I have completed the edits I planned to make to this page. I would be interested to see any comments.

Buzz Bloom

Some of this article's information has been moved to Neanderthals Bandits and Farmers or Cannibals and Kings articles where it more rightfully belongs. The remainder contained some pretty basic errors (e.g. supply and demand) and has been mostly rewritten. I am pretty confident about this, but if you think it was correct we can discuss it here. User:H7asan

H7asan

We obviously have a disagreement regarding the relevance of "Beyond the Limits". I found the entire book exactly on the point. It deals with the exhaustion of food (and other resources) as a result of unconstrained population growth (as well as the unconstrained growth of consumption). I definitely think this book should be referenced from a discussion of neo-Malthusean theory. Why do you think otherwise? Also, what is the proper mechanism for getting a disagreement of this kind resolved?

By the way, I thought your moving of the discussion about the Harris and Tudge books to their own pages was a good idea.

Buzz Bloom

I have nothing against the Beyond the Limits book. (Actually I know nothing about it.) My problem was with the article which was empty. User:H7asan

---

I plan to remove the two paragraphs beginning with "Another problem is that there is no strong evidence ... " including the two graphs. This discussion is irrelevant to the topic of the Mathusuan catastrophe. Malthus never described population growth as being expoential. He said the growth would be expoential in unchecked, and then only until a subsistance level was reached. Growth of a population until a subsistance level would correspond to what Securiger describes in the current text I plan to remove as a Logistic curve. All that the curves show is that the current trend of world population from 1950-2000 may be begining to reach a new limit of a kind that Malthus discusses: use of contraception, which Malthus called a vice.

I put this notice of intent here to elicit comments or alternative suggestions before doing it.

I also plan to edit the remaining material in the "Non-occurrence of the catastrophe" section cbecajuse I think it un fairly represents the state of the world at the end of the 19th century, which the anthropoligist Marvin Harris describes as one of approaching catastrophe as predicted by Malthus. The section should discuss the innovations of the twentieth century that offer opportunities to avoid the catastrophe, or only postpose it. From this perspective, I would change he title of the section to "Postponement or non-occurrence of the catastrophe".

I also elicit comments or alternative suggestions regarding these intentions.

User:BuzzB Feb 28, 2004

I disagree with both proposed edits, quite strongly. Firstly, the paragraphs beginning "Another problem is..." are highly relevant. Malthus proposed a particular theory, which was essentially premised on three claims, one of them being the idea that a human population undergoes geometric growth if unchecked. Malthus' Essay has been disputed by many, and one major point of disputation - indeed one of the few points, pro- or anti-, that bothers to look at empirical facts - is that there is absolutely no evidence in support of this basic premise. It was pointed out as soon as the Essay was published, and continues to be pointed out today; if you like you can propose hypotheses to explain that fact away, but simply removing all evidence of it would severely bias the article.

Equally, we could point out that there is no evidence that food supply increases arithmetically, and that in fact it patently does not. One of the nicer summaries is this, written by Hazlitt in 1822:

All that is true of Mr Malthus's doctrine then, is this, that the tendency of population to increase remains after the power of the earth to produce more food is gone; that the one is limited, the other unlimited. This is enough for the morality of the question: his mathematics are altogether spurious.

Secondly, you propose to edit the remaining material in that section, because you claim that it "un fairly represents the state of the world at the end of the 19th century, which the anthropoligist Marvin Harris describes as one of approaching catastrophe as predicted by Malthus". Huh? That section doesn't even discuss the end of the nineteenth century! If you meant "end of the 18th century", which is mentioned, then of that time it says "At the time Malthus wrote, most societies had populations at or near their agricultural limits" - which is not contradicted by your point!? (Although there is plenty of evidence to believe that that statement is also somewhat exaggerated).

I should point out that when I get time to do it justice, I plan to make extensive additions to this article, which in my opinion is currently very shallow and unencyclopedic. It currently represents the shallow, ill-defined, handwaving version of the Malthusian theory that is frequently dragged out in the pub or at dinner parties in support of some political argument or another. But in fact Malthus had a much more complete theory than is represented here, which was one of the seminal theories that gave rise to economics. (Although there is very little of the detail that is still widely accepted.) We need to work in its rôle in the development of economics. Additionally the current article needs to mention Wallace, who had the idea first. Oh, and it also doesn't even mention the basic Malthusian idea that increased food supply automatically generated increased population until everyone was starving again, which segues into the rôle the theory in had in justifying the oppression of the poor in nineteenth century politics - again from Hazlitt:

The instant, however, any increase in population, with or without an increase in the means of subsistence, is hinted, the disciples of Mr Malthus are struck with horror at the vice and misery which must ensue to keep this double population down; nay, mention any improvement, any reform, any addition to the comforts or necessaries of life, any diminution of vice and misery, and the infallible result in their apprehensive imagination is only an incalculable increase of vice and misery, from the increased means of subsistence, and increased population that would follow. They have but this one idea in their heads; it comes in at every turn, and nothing can drive it out.

Securiger 11:28, 1 Mar 2004 (UTC)

Graph of World Population

Hmm. I just wanted to object to a few things:

• The World Population graph is described as "clearly... close to linear". Really? To me it looks "clearly curved". (In fact, I think I see evidence of the logistic curve, but that could well be spurious.) As alluded to in the article, 50 years is an awfully short time to get a good idea of how human population levels are changing. In fact, graphing the data from 1804-1999 given in the first external link (http://www.geohive.com/global/linkg.php?xml=hist2&xsl=hist2) at that point in the article, would give a strong impression of exponential growth. Yes, maybe we're starting to see the beginning of the "slowing down" in growth that's predicted by the logistic model, but it's relatively early in that process, IMO, so I would be very hesitant to claim that the growth is no longer exponential — certainly not based on the data given here. (dcljr, continued below)
  • Yes, graphing the 6 points 1804-1999 does look closer to exponential (see below) - but the data prior to 1950 are extrapolations or approximations, based partly on the assumption of exponential growth prior to 1950 (and of course data after 2004 is purely extrapolated with some unstated model). Only the data highlighted in blue are based largely on actual census counts. (In any case, it looks closer still to two linear segments with an critical point near 1960). So the reason for concentrating on the last 50 years is because that is the sole period for which we have reasonably reliable data. And when you graph that real data, you get something that offers little support for the common assumption that "it's obviously exponential". Also it was not stated that it's "no longer exponential", but rather that there is no strong evidence that it ever has been. In fact it could be a very slow exponential, or maybe a very slow logistical, or perhaps linear, or quadratic - the point is we really don't know, and at any rate it certainly isn't a simple function. But at least the reason for choosing this period should be clarified. (Securiger)
    Missing image
    Extrapolated_world_population_history.png
    Image:Extrapolated world population history.png
    
    
    • Hmm... Part of the reason it might look closer to two linear segments is because the interpolating curve is (I assume) a cubic spline (and there's no point for it to go through between c. 1805 and 1925). Anyway, I agree with the rest of your paragraph. - dcljr 22:53, 7 Sep 2004 (UTC)

• Note that the plot of World Population Increase suggests that the rate of increase may actually still be going up, perhaps even (approximately) linearly (you always have to expect short-term fluctuations from the overall trend), which would imply quadratic growth. (dcljr)
  • How do you figure that? Apart from two years, it has been going down every year since 1987 - which is a third of the period for which we actually have reliable data! (Overall, there has been downturn in the growth rate for 26 of the 54 years considered.) (Securiger)
• The sentence that begins "Also the rate of increase should increase, whereas, of the increase between 1960 and today, five-sixths occurred in the early 1960s", aside from being confusing, is completely misleading, since a mere glance of the Increase graph shows something highly unusual happening in the years 1957-1962, resulting in a lcoal minimum in 1960! That dip in the graph is the only reason the statement above is true (to the extent that it is). (dcljr)
  • I'll try to rephrase the sentence you find confusing. The point is that in a positive exponential, the first difference (and second difference, and all other differences) is also a positive, upward trending exponential. Thus when you get a true exponential growth curve and plot the differences between years, that rate-of-growth curve is itself an upward curving exponential. The rate-of-growth curve for human population clearly does not look like that at all. This is seen even more so in the 2nd difference curve (below), which however I would not include on the main page because second differences are heavily affected by noise. If population was exponential, the second difference curve should also be exponential, in fact it has a lot of noise oscillating around zero but with an overall downward trend. As for the statement which you claim is "completely misleading", umm, your "objection" agrees almost exactly with the point and meaning of that sentence: if we were looking at exponential growth, most of the growth would be recent, but in fact most of the growth is due to "something highly unusual" happening back then - the big dip from '57 to '60, and also the huge surge from '60 to '63. And even if we interpolate the years '57 to '63 to remove this curious feature, 75% of the growth increase between 1950 and the peak year, 1990, occurred in the first half of that period. This is just not at all consistent with a positive exponential growth. It seems I need to make some clarifications on why this chart, and the data it represents, are wholly inconsistent with the implications of exponential growth. (Securiger)
    Missing image
    Population_2nd_derivative.png
    Image:Population 2nd derivative.png
    
    
    • Maybe I misunderstood your purpose of pointing out the circa-1960 thing. I don't know. In any case, I wouldn't read much into the data of around that time. The "hiccup" might just be "noise" or might be due to a completely administrative cause (a change, say, in how censuses were taken or recorded in one or more large countries at the time — who knows?). I think most of our "differences" can be summed up by the following statement from the article: "...short-term trends, even on the scale of decades or centuries, do not necessarily disprove the underlying mechanisms...". I've been taking a much more long-term perspective, figuring that things like the 1960-ish "hiccup" and the "decrease in the increase" since 1986 are likely short-term deviations from the overall pattern over centuries (which is essentially unknowable anyway, but at least an exponential [and logistical] model has some theoretical basis). Anyway, I think both of us can agree that in the last 50 years or so the trend has not appeared to be exponential. On a completely different note, it would be interesting to consider (not in the article itself — or even here, necessarily) what role (tele)communications and transportation plays in all of this. Might population growth be "stabilizing" (2nd derivative graph above) as a result of the increased interconnectedness of human populations? Perhaps that's the reason behind the "critical point" of around 1960? - dcljr 22:53, 7 Sep 2004 (UTC)

• Finally, I think the correlation coefficient is a pretty useless measure of anything in this context; someone should do an appropriate statistical test on the yearly data instead (I suggest an F-test to see whether an exponential term is needed over linear [intercept and slope] terms, and possibly an approximate lack of fit test). - dcljr 08:00, 6 Aug 2004 (UTC)
  • Why do you think it is useless here? <math>r^2<math> is supposed to measure the fraction of the variability in y explained by the function of x - in this case, a linear model and exponential one explain the variability about equally well. I don't understand what you mean by "F-test to see whether an exponential term is needed over linear", but an F-test finds no significant difference in the residuals from linear and exponential models. Securiger 16:58, 7 Aug 2004 (UTC)
    • Why is it useless? Because an exponential with slow growth can look linear and have a correlation close to 1! As for r², the article mentions the correlation coefficient not the coefficient of determination. Although obviously they're computationally the same in this case, the author was using it specifically to indicate linearity. In any case, even if you grant that "practically speaking" the correlation is close to 1, consider what "practice" we're putting this information to: we're using these models to predict population levels far into the future (sometimes as far into the future as we have "reliable" data in the past, in fact — see above graph) and there can be a big difference between extrapolation using a linear model and one using an exponential. (Of course. That's why we're discussing this in the first place.) Oh, and I meant an ANOVA "F-test" for testing whether a coefficient in a regression model is zero (as opposed to an "F-test" for testing the equality of two population variances). I should have been more specific. I'm not sure what "F-test" you did. - dcljr 22:53, 7 Sep 2004 (UTC)