Thursday, February 5, 2015

Why Did The First Farmers Farm?

Numbers Make Smart People Stupid

An open access paper in PNAS by Samuel Bowles (of the interdisciplinary Sante Fe Institute) and Jung-Kyoo Choi (a Korean academic economist), "Coevolution of farming and private property during the early Holocene," is a classic example of smart people doing lousy analysis with quantitative methods.  But, the focus on private property institutions in the context of the development of farming technology seems driven more by the political agenda of the investigators, than by the data.

If you read academic journal articles long enough, you will notice the trend.  Someone outside a field uses a simple mathematical model to solve problems that have confounded experts in the field for decades or centuries and publish it in a general interest journal. This provides a clear, precise answer to the problem.  This creates a huge controversy as experts in the field attempt to explain the huge conceptual or methodological errors made by the outsiders, usually with objections that are well founded.  The wider public, however, remembers the flawed study, but not the apt criticisms of it made in response to it, long after it is published.

One of the authors at the Language Log blog rants about some similar cases involving historical linguistics at Science magazine where a study with weak methodology used mathematical "entropy analysis" to argue that the Indus River Valley civilization's symbols were part of a written language.  Science wasn't interested in publishing an attempted replication using better data and methods that contradicted this famous result, concluding that the Harappan symbols, like the symbols used by the Vinca civilization of the Balkans and symbols used by the Picts were a mere proto-linguistic symbol system and not a full fledged written language.

Another PNAS computational linguistic paper by repeat offender and New Zealand academic Quentin D. Atkinson and others was also recently ripped apart by the linguistics experts at Language Log in a takedown that I discussed in a recent post.  Atkinson released another notoriously flawed computational linguistics paper on serial founder effects and phonemic diversity in 2011 which I blogged at the time.  He co-authored a flawed computational linguistics paper on the age of the Indo-European language family in Nature in 2003 and reproduced by others with improved by still flawed methods in 2011 (blogged here).

And, there was also the flawed foray into computational linguistics analyzing the case for an Altaic language family, also sponsored by the Sante Fe Institute, with retired eminent physicist Murray Gell-Mann as a co-author.  This study was actually better than some of Atkinson's efforts, but is was bad enough that Gell-Mann was ridiculed for it, and included some glaring overstatements about what had been shown.  For example, deceptively the study prominently discussed the Korean language as part of the Altaic language family in its analysis, despite the fact that the study didn't include any Korean words in the data sets that were used for the paper's computational analysis.

The Bowles and Choi Paper On The Neolithic Revolution

Bowles and Choi are trying to answer the question of why people started farming when they did (an event called the Neolithic Revolution), despite the fact that the first farmers were generally less well fed than contemporaneous foragers.  They pose the question, somewhat misleadingly, as follows (emphasis added):
[A]s a number of archaeologists have pointed out (1113), farming was probably not economically advantageous in many places where it was first introduced. Indeed, recent estimates suggest that the productivity of the first farmers (calories per hour of labor including processing and storage) was probably less than that of the foragers they eventually replaced, perhaps by a considerable amount (14) (SI Appendix). In many parts of the world, stature and health status appear to have declined with cultivation (15). Farming did raise the productivity of land and animals, and this, we will see, was critical to its success. However, why an erstwhile hunter–gatherer would adopt a new technology that increased the labor necessary to obtain a livelihood remains a puzzle.  
They are correct that calories per hour of labor including processing and storage was probably less for early farmers than for foragers, and that stature and health status declined among early adopters of farming.  My quarrel is with their use of the term "economically advantageous", when what they really mean is "nutritionally advantageous."  Their failure to distinguish these two concepts is a critical flaw in the study.

There is no good reason at all to believe that farmers were acting in anything but their best economic interests taken as a whole, even though this may have involved nutritional deficits.  The question instead, is what benefits did early adopters of farming receive that made up for the nutritional deficits that they experienced.

When People Seem Irrational, You Are Ignorant Of One Or More Key Facts 

One of the most powerful lesson of economics is that people, even ill educated people whom their betters don't give much credit for acting rationally do indeed very consistently, at least on average, respond to incentives and take actions that enhance their well being.  Almost always, when people do something that looks like it is irrational, this is because the observer isn't aware of the factors that are driving people to take the action in question.

For example, while renting rather than owning a home when interest rates are low and small down payment can be secured, making it cheaper to make mortgage payments than rent payments, may seem like an irrational choice for a working class individual, when one looks closer, it frequently isn't irrational.  If that working class individual has bad credit, the interest rates he would have to pay may be much higher than those paid by his middle class peers.  And, if that working class individual is likely to be unemployed for a few months at a time over the course of the next decade and has little savings, owning a home exposes him to a loss of a down payment in a foreclosure when he is unemployed and can no longer make his monthly payment, while his down side loss in the event of a few months of unemployment may be less (particularly if he keeps his limited assets in a liquid savings account rather than a down payment to buffer him at these times) if he rents.  A choice that looked irrational when only a simplistic analysis is conducted may make more sense when more factors are considered.

Key Factual Observations About the Neolithic Revolution

This Bowles and Choi paper is better than most at presenting some of the key historical facts from archaeology and paleoclimate studies, even though its model is too simple to capture many of these insights and adds little to what can be completed more reliably without a mathematical simulation which is really just a gimmick for presenting the foregone conclusion associated with the assumptions that go into the model. 


The key paleoclimate data are presented in the chart below.

Ignoring for the moment their simulated data in light gray bar graph form:
Estimated dates of some well-studied cases of the initial emergence of cultivation are on the horizontal axis (8, 54, 55). Climate variability (Left) is an indicator of the 100-y maximum difference in surface temperature measured by levels of δ18O from Greenland ice cores (SI Appendix). A value of 4 on the vertical axis indicates a difference in average temperature over a 100-y period equal to about 5 °C.
As the chart indicates, intermittent periods of wild temperature variation over the span of just a few generations was the norm for the entire Upper Paleolithic era (about 40,000-50,000 years ago), after which temperatures became much more stable starting at the beginning of the Holocene era about 10,000 years ago when farming first emerged in the Fertile Crescent and China and the middle latitudes of the Americas (farming arose independently at later times in the New Guinea Highlands, Sub-Saharan Africa and the Eastern United States).

Thus, a key part of the answer to the question of why farming emerged when it did is that the climate was too unpredictable for farming to be a viable means of food production during almost the entire Upper Paleolithic era.  Farming didn't emerge before the Holocene because it couldn't in the climate conditions at the time.

It is very plausible to think that property rights were an effect driven by the fundamental economics of farming and herding, rather than an important cause of this shift, when reduced climate variability provides a much more plausible proximate cause of the Neolithic revolution.  The authors certainly don't suggest any plausible way that the Neolithic revolution could have been possible earlier if Upper Paleolithic humans had adopted different property regimes, something that ought to have been possible if property rights were as important as they suggest in this transition.  And, Coase's theorem, a jewel of modern economics, which basically says that economics will generally drive people to find a work around to secure economically efficient arrangements in the face of bad legal rules, tends to favor that direction of causation.

Prior to the Upper Paleolithic era, modern human presence was largely restricted to Africa and mainland Asia (probably excluding North Asia).  Modern humans prior to that era were absent from the Americas, Oceania, Australia, the islands east of the Wallace line, Japan, Taiwan, Tibet, the Andaman Islands, and Crete. 

There is no solid positive evidence for any modern human presence in Asia beyond South Asia until the Toba eruption ca. 75,000 years ago, although there is evidence of a modern human presence in Southwest Asia more than 100,000 years ago, some of which was part of the same archaeological culture as a contemporaneous group of modern humans in Upper Egypt and the Sudan.  Before then, modern humans were confined to Africa where they evolved in the first place as foragers.

The Archaeological Record in the Fertile Crescent Neolithic

Bowles and Choi do nicely summarize some key elements of the archaeological record describing the Neolithic transition:
Kuijt and Finlayson very plausibly write that a “transition from extramural to intramural storage system may reflect evolving systems of ownership and property … with later food storage systems becoming part of household or individual based systems” (2). . .                          
Southwestern Asia provides the best-documented cases providing evidence of the gradual adoption of food production along with evidence suggesting the emergence of private property in stores in the Levant between 14,500 and 8,700 B.P. (SI Appendix). At the beginning of this period, Natufians hunted and collected wild species and possibly practiced limited wild-species cultivation along with limited storage (41, 42). Somewhat before 11,000 B.P. there is direct evidence of storage of limited amounts of wild plants outside of dwellings, consistent with the hypothesis that access to stored goods was not limited to the members of a residential unit (2, 43). A millennium later, goats and sheep had been domesticated (constituting a substantial investment), and we find large-scale dedicated storage located inside dwellings, suggesting more restricted access (4, 44).
Over none of this period could one describe these communities as either simply foragers or farmers. Their livelihoods were mixed; in many cases, their residential patterns varied over time between sedentary and mobile, and it seems their property rights, too, varied among the types of objects concerned, with elements of both private and common property in evidence. Bogaard (4) and her coauthors found that at Catalhoyuk in central Anatolia (10,500–10,100 B.P.), “families stored their own produce of grain, fruit, nuts and condiments in special bins deep inside the house.” This restricted-access storage coexisted with the prominent display of the horns and heads of hunted wild cattle. The authors concluded that “plant storage and animal sharing” was a common juxtaposition for “the negotiation of domestic [the authors elsewhere call it “private”] storage and interhouse sharing.” The process of change was neither simple, nor monotonic, nor rapid. However, in both its institutions and its technology, Levantine people were living in a very different world in 8,700 B.P. from the world of the early Natufians almost six millennia earlier.
Thus, a second key fact is that people didn't just "adopt farming."  They became proto-farmers of wild crops at a time when hunting was good enough for them to remain at one location.

Another Economics Consideration

Yet another key consideration that doesn't really properly get posed is the fact that in economics, you need to look at the margins, and not at averages.  Those who adopted farming first, no doubt, had the most comparative advantage.  This is to say, they were best at farming relative to their success at hunting.  Perhaps they had mobility problems due to injured family members, or pregnant women in the family.  Perhaps they were in places that had been over-hunted and storage of hunted and gathered products reflected this scarcity (people hoard things that have value), but had especially fertile soils.

Anyway, the key point is that even if the first farmers were less well fed than the average hunter-gatherer, that doesn't imply that this was true for this particular subset of individuals.

Trade Offs

Why else would someone sacrifice calories to farm?

Farmers can live sedentary lives, which means that they can invest resources to build homes, rather than rebuilding camps every few days, providing better quality shelter with less of an investment in labor, that shelters them from the elements, hungry animals or unfriendly tribes of fellow humans.  This saved time may have also freed up time to make other things, like clothing, that can also protect people from the elements and thereby increase their ability to survive.  And, it can allow them to accumulate goods, like pottery, baskets and food stores, that can buffer short term shortfalls in food production.

A little more hunger may be worth fewer deaths from hypothermia, wild animals, brief periods of unsuccessful hunting and gathering, and attacks by hostile foreign tribes.  These benefits may be particularly beneficial for the survival of children, the temporarily injured, and pregnant women, whose well being enhances selective fitness, and to the elderly whose knowledge and capacity to free up the time of community members who are able to hunt and gather and farm and herd while caring for children and doing sedentary tasks like making flour, ceramics, weapons and clothing.

Proto-farmers may also have traded lower average calories in order to have a steadier food supply not so strongly subject to feast and famine cycles, that culled weaker members of the community leaving those who survived better fed, but meant that many died of hunger when the hunting and gathering wasn't as abundant, wresting control from Nature to mankind.


Marnie said...

"Science magazine where a study with weak methodology used mathematical "entropy analysis" to argue that the Indus River Valley civilization's symbols were part of a written language. Science wasn't interested in publishing an attempted replication using better data and methods that contradicted this famous result, concluding that the Harappan symbols, like the symbols used by the Vinca civilization of the Balkans and symbols used by the Picts were a mere proto-linguistic symbol system and not a full fledged written language."

I ran across this paper. What the authors didn't account for is that all "written" languages require some of the information content to be conveyed culturally. So, for instance, Europeans today are all taught their ABCs at a young age. Mandarin speakers are taught to read and draw their character sets from a young age. Similarly, Vinca script writers, or Blackfoot Winter Count keepers, existed within a cultural contexts where the meaning of their symbols was culturally conveyed.

The amount of information conveyed culturally, vs in the written symbols, does vary in these different systems, but that does not mean that a system that more heavily weights cultural transmission (than standardized writing), is not a writing sytem.

Anyway, I was quite surprised that anyone took the entropy argument, used to disqualify systems like Vinca script, at all seriously.

Surprisingly, I've heard several well published archaeologists repeat the entropy argument in that paper.

Makes me wonder what you guys are smoking.

Marnie said...

This post is a very good critique of some the oft thrown around ill constructed assumptions about the transition to farming.

One further comment I would make is that although the definition of property probably became more specific with the transition to farming, it is improbably that "hunters-gatherers" did not have a good idea of territory.

For instance, historical accounts of American and Canadian native hunter-gatherers had very specific ideas about the boundaries of their territories. These boundaries were usually marked by rivers or mountain ridges.

Hunters also did a lot of gathering, and the areas where they gathered must have been very important, since the native grains, berries and tubers gathered would have been critical to survival.

Sorry for the rant, but the idea that hunter-gatherers didn't have an idea of property really bugs me.

andrew said...

"the idea that hunter-gatherers didn't have an idea of property really bugs me."

Agreed. They probably had different rules and institutions, but they almost certainly had something resembling property.

Marnie said...

I just wanted to add that this article about the transition to farming is excellent.

Thanks for writing it.