Both Nature and New Scientist recently covered the work of Peter Turchin, who suggests, based on prior trends, that the US is in for a new period of political instability peaking around 2020. He finds that historically US instability has peaked about every fifty years:
This is somewhat compelling but also somewhat lacking. What's the scoop on the nearly hundred years quiet from 1780 to 1870? Also none of these incidences were widespread. In 1870 it was farmers. In 1920 it was bank rushes and in 1970 it was the civil rights movement. The most interesting part to me is how lynchings (sadly) directly jumped with riots in the 19th and 20th centuries. Now they are all but gone. Are we going to see a new violent expression of unrest? Also terrorism is a relative newcomer and is currently the biggest mover. However, our current government is far more vigilant on this front than any of the other subjects on the graph. My conclusion is that this study mostly shows us the direct relation between economic trouble and social unrest, which is good. I'm not seeing it as too useful a predictive model. I would definitely say that with our current economic troubles and increasingly polarized social climate the US may see some strife.
Maybe the (gated) paper is more impressive than the charts. One of the biases that humans exhibit is to see patterns where there are none.
On the first one, I don't see a 50-year cycle. I see internal political turmoil caused by the civil war, then world war I, then the baby boom. Then I see a surge in foreign terrorism.
On the second one, I deny that anyone can prove a 150 cycle in a 220 year time frame. If you need two points to make a line, you need two repetitions to make a cycle.
Apart from that, I again see specifics, not a cycle. I don't know what the six variables are, but I can certainly see some likely explanations. The plunge around 1820 strikes me as a combination of revolutionary success, the end of the Federalist party, and peace followed by post-Napoleonic exhaustion in Europe and westward expansion in the US. The 1860's saw a spike from civil war and reconstruction. The slump from about 1915 looks like a herd reaction to war, depression, war, and cold war, followed by a gradual return to normalcy.
Finally, the Malthusian explanations immediately strike the economist in me completely dumbfounded. I thought this kind of thing had finally -- and thankfully -- gone out of fashion.
I'm wary of these types of models: first, they don't really posit a mechanism except vague qualitative speculation, second, they are based on too little and too much noisy data.
I haven't read the paper, but the abstract does mention a mechanism: "A possible explanation of this pattern, discussed in the article, is offered by the structural-demographic theory, which postulates that labor oversupply leads to falling living standards and elite overproduction, and those, in turn, cause a wave of prolonged and intense sociopolitical instability."
Stuart Armstrong posted on Less Wrong about tests for "counterfactual resilience" for hypothetical objective trends ( http://tinyurl.com/bvdcm2a ); it seems it could be applied here.
The plausible mechanism Turchin's theory is built upon is Jack Goldstone's neo-malthusian structural-demographic theory. Here is one outline, from Turchin: http://www.eeb.uconn.edu/pe...
I think at least one of those articles mentions that, but I remember thinking they should have emphasized it more.
US demographics are changing very rapidly. Populations originating from less stable parts of the world are making up more and more of the overall US population. A less homogeneous population leads to lower trust levels and higher likelihood of involvement of the legal system or the criminal justice system.
There are a lot more reasons to be concerned than Turchinoid cycles.
Well, my first thought is generational cycles, but the existence of a pattern is more important than the cause in this case. Something psychohistorical, supernatural, climatological, or conspiratorial could cause this without changing the prediction.
There's no plausible mechanism presented for what would cause periodic crescendos of instability. Isn't it much more likely that instability is a random walk with slowly changing linear trends?
That seems to gibe much more with common sense than instability being a periodic function with a stationary random component fluttering around the long term deterministic path.
There's not even a suggestion about why this might be. Turchin just throws out a handful of somewhat similar datapoints and says "Look at this!" without even bothering to computer some measure of statistical significance.
Well, Turchin got this one bang on the money.
This is somewhat compelling but also somewhat lacking. What's the scoop on the nearly hundred years quiet from 1780 to 1870? Also none of these incidences were widespread. In 1870 it was farmers. In 1920 it was bank rushes and in 1970 it was the civil rights movement. The most interesting part to me is how lynchings (sadly) directly jumped with riots in the 19th and 20th centuries. Now they are all but gone. Are we going to see a new violent expression of unrest? Also terrorism is a relative newcomer and is currently the biggest mover. However, our current government is far more vigilant on this front than any of the other subjects on the graph. My conclusion is that this study mostly shows us the direct relation between economic trouble and social unrest, which is good. I'm not seeing it as too useful a predictive model. I would definitely say that with our current economic troubles and increasingly polarized social climate the US may see some strife.
The 50-year cycle does appear to be real. Sunspots?
Maybe the (gated) paper is more impressive than the charts. One of the biases that humans exhibit is to see patterns where there are none.
On the first one, I don't see a 50-year cycle. I see internal political turmoil caused by the civil war, then world war I, then the baby boom. Then I see a surge in foreign terrorism.
On the second one, I deny that anyone can prove a 150 cycle in a 220 year time frame. If you need two points to make a line, you need two repetitions to make a cycle.
Apart from that, I again see specifics, not a cycle. I don't know what the six variables are, but I can certainly see some likely explanations. The plunge around 1820 strikes me as a combination of revolutionary success, the end of the Federalist party, and peace followed by post-Napoleonic exhaustion in Europe and westward expansion in the US. The 1860's saw a spike from civil war and reconstruction. The slump from about 1915 looks like a herd reaction to war, depression, war, and cold war, followed by a gradual return to normalcy.
Finally, the Malthusian explanations immediately strike the economist in me completely dumbfounded. I thought this kind of thing had finally -- and thankfully -- gone out of fashion.
Max
I'm wary of these types of models: first, they don't really posit a mechanism except vague qualitative speculation, second, they are based on too little and too much noisy data.
I haven't read the paper, but the abstract does mention a mechanism: "A possible explanation of this pattern, discussed in the article, is offered by the structural-demographic theory, which postulates that labor oversupply leads to falling living standards and elite overproduction, and those, in turn, cause a wave of prolonged and intense sociopolitical instability."
Stuart Armstrong posted on Less Wrong about tests for "counterfactual resilience" for hypothetical objective trends ( http://tinyurl.com/bvdcm2a ); it seems it could be applied here.
The plausible mechanism Turchin's theory is built upon is Jack Goldstone's neo-malthusian structural-demographic theory. Here is one outline, from Turchin: http://www.eeb.uconn.edu/pe...
I think at least one of those articles mentions that, but I remember thinking they should have emphasized it more.
US demographics are changing very rapidly. Populations originating from less stable parts of the world are making up more and more of the overall US population. A less homogeneous population leads to lower trust levels and higher likelihood of involvement of the legal system or the criminal justice system.
There are a lot more reasons to be concerned than Turchinoid cycles.
Well, my first thought is generational cycles, but the existence of a pattern is more important than the cause in this case. Something psychohistorical, supernatural, climatological, or conspiratorial could cause this without changing the prediction.
There's no plausible mechanism presented for what would cause periodic crescendos of instability. Isn't it much more likely that instability is a random walk with slowly changing linear trends?
That seems to gibe much more with common sense than instability being a periodic function with a stationary random component fluttering around the long term deterministic path.
There's not even a suggestion about why this might be. Turchin just throws out a handful of somewhat similar datapoints and says "Look at this!" without even bothering to computer some measure of statistical significance.