Book Review: Secular Cycles

Link post


There is a tide in the af­fairs of men. It cy­cles with a pe­riod of about three hun­dred years. Dur­ing its flood, farms and busi­nesses pros­per, and great em­pires en­joy golden ages. Dur­ing its ebb, war and famine stalk the land, and states col­lapse into bar­barism.

Chi­nese pop­u­la­tion over time

At least this is the the­sis of Peter Turchin and Sergey Nefe­dov, au­thors of Sec­u­lar Cy­cles. They start off Malthu­sian: due to nat­u­ral re­pro­duc­tion, pop­u­la­tion will keep in­creas­ing un­til it reaches the limits of what the land can sup­port. At that point, ev­ery­one will be stuck at sub­sis­tence level. If any group ever en­joys a stan­dard of liv­ing above sub­sis­tence level, they will keep re­pro­duc­ing un­til they are back down at sub­sis­tence.

Stan­dard Malthu­sian the­ory evokes images of a pop­u­la­tion sta­ble at sub­sis­tence level for­ever. But Turchin and Nefe­dov ar­gues this isn’t how it works. A pop­u­la­tion at sub­sis­tence will always be one meal away from starv­ing. When a famine hits, many of them will starve. When a plague hits, they will already be too sickly to fight it off. When con­flict ar­rives, they will be des­per­ate enough to en­list in the armies of whichever war­lord can offer them a warm meal.

Th­ese are not piece­meal events, pick­ing off just enough of the pop­u­la­tion to bring it back to sub­sis­tence. They are great cat­a­clysms. The Black Plague kil­led 30% – 60% of Euro­peans; the An­to­nine Plague of Rome was al­most as deadly. The Thirty Years War kil­led 25% – 40% of Ger­mans; the Time of Trou­bles may have kil­led half the pop­u­la­tion of me­dieval Rus­sia.

Thus the sec­u­lar cy­cle. When pop­u­la­tion is low, ev­ery­one has more than enough land. Peo­ple grow rich and re­pro­duce. As time goes on, the same amount of farm­land gets split among more and more peo­ple. Wages are driven down to sub­sis­tence. War, Famine, and Pestilence rav­age the land, with Death not far be­hind. The kil­lings con­tinue un­til pop­u­la­tion is low again, at which point the cy­cle starts over.

This ap­plies mostly to peas­ants, who are most at risk of starv­ing. But no­bles go through a re­lated pro­cess. As a cy­cle be­gins, their num­bers are low. As time goes on, their pop­u­la­tion ex­pands, both through nat­u­ral re­pro­duc­tion and through up­ward mo­bil­ity. Even­tu­ally, there are more no­bles than there are good po­si­tions…

(this part con­fused me a lit­tle. Shouldn’t num­ber of good po­si­tions scale with pop­u­la­tion? IE if one baron rules 1,000 peas­ants, the num­ber of ba­ro­nial po­si­tions should scale with the size of a so­ciety. I think T&N hint at a few an­swers. First, some po­si­tions are ab­solute rather than rel­a­tive, eg “King” or “Minister of the Econ­omy”. Se­cond, no­ble num­bers may some­times in­crease faster than peas­ant num­bers, since no­bles have more food and bet­ter chances to re­pro­duce. Third, dur­ing boom times, the ranks of no­bles are swelled through up­ward mo­bil­ity. Fourth, con­spicu­ous con­sump­tion is a ratchet effect: dur­ing boom times, the ex­pec­ta­tions of no­bil­ity should grad­u­ally rise. Fifth, some­times the rele­vant de­nom­i­na­tor is not peas­ants but land: if a no­ble only has one acre of land, it doesn’t mat­ter how many peas­ants he con­trols. Sixth no­bles usu­ally sur­vive famines and plagues pretty well, so af­ter those have done their work, there are far fewer peas­ants but ba­si­cally the same num­ber of no­bles. All of these fac­tors con­tribute to ex­cess no­ble pop­u­la­tion – or as T&N call it, “elite over­pro­duc­tion”)

…and the no­bles form “ri­val pa­tron­age net­works” to fight for the few re­main­ing good spots. The state goes from united (or at least all no­bles united against the peas­ants) to di­vided, with coal­i­tions of no­bles duk­ing it out (no pun in­tended). This can lead ei­ther to suc­cess­ful peas­ant re­bel­lion, as some no­bles sup­port the peas­ants as part of in­ter-no­ble power plays, or just to civil war. Although famine and plague barely af­fect no­bles, war af­fects them dis­pro­por­tionately – both be­cause they are of­ten knights or other front-line sol­diers, and be­cause kil­ling the other side’s no­bles was a ma­jor strate­gic goal (think Game of Thrones). So a civil war usu­ally fur­ther de­pletes the already-de­pleted peas­ant pop­u­la­tion, and fi­nally de­pletes no­ble pop­u­la­tions, lead­ing to a gen­eral un­der­pop­u­la­tion and the be­gin­ning of the next cy­cle.

Com­bine these two pro­cesses, and you get the ba­sic struc­ture of a sec­u­lar cy­cle. There are about a hun­dred years of un­al­loyed growth, as peas­ant and no­ble pop­u­la­tions re­bound from the last dis­aster. Dur­ing this pe­riod, the econ­omy is strong, the peo­ple are op­ti­mistic and pa­tri­otic, and the state is strong and united.

After this come about fifty years of “stagfla­tion”. There is no more room for easy growth, but the sys­tem is able to ab­sorb the sur­plus pop­u­la­tion with­out crack­ing. Peas­ants may not have enough land, but they go to the city in search of jobs. Nobles may not have enough of the po­si­tions they want, but they go to col­lege in or­der to be­come bu­reau­crats, or join the ret­inues of stronger no­bles. The price of la­bor reaches its low­est point, and the haves are able to ex­ploit the des­per­a­tion of the have-nots to reach the zenith of their power. From the out­side, this pe­riod can look like a golden age: huge cities buzzing with peo­ple, uni­ver­si­ties crammed with stu­dents, ul­tra-rich no­bles throw­ing money at the arts and sci­ences. From the in­side, for most peo­ple it will look like a nar­row­ing of op­por­tu­nity and a hard-to-ex­plain but grow­ing sense that some­thing is wrong.

After this comes a crisis. The mechanisms that have pre­vi­ously ab­sorbed sur­plus pop­u­la­tion fail. Famine and dis­ease rav­age the peas­antry. State fi­nances fall apart. So­cial trust and pa­tri­o­tism dis­ap­pear as it be­comes in­creas­ingly ob­vi­ous that it’s ev­ery man for him­self and that peo­ple with scru­ples will be defeated or ex­ploited by peo­ple with­out.

After this comes the de­pres­sion pe­riod (marked “in­ter­cy­cle” on this graph, but I’m go­ing to stick with the book’s term). This graph makes it look puny, but it can last 100 to 150 years. Dur­ing this pe­riod, the peas­ant pop­u­la­tion is low, but the no­ble pop­u­la­tion is still high. This is most likely a pe­riod of very weak or even ab­sent state power, fre­quent bar­bar­ian in­va­sions, and fre­quent civil war. The peas­ant pop­u­la­tion is in a good po­si­tion to ex­pand, but can­not do so be­cause wars keep kil­ling peo­ple off or forc­ing them into walled towns where they can’t do any farm­ing. Usu­ally it takes a cou­ple more wars and dis­asters be­fore the no­ble pop­u­la­tion has de­creased enough to re­verse elite over­pro­duc­tion. At this point the re­main­ing no­bles look around, de­cide that there is more than enough for all of them, and feel in­cen­tivized to co­op­er­ate with the for­ma­tion of a strong cen­tral­ized state.

This cy­cle is in­ter­wo­ven with a sec­ond 40-60 year pro­cess that T&N call the “fathers-and-sons cy­cle” or “bi­gen­er­a­tional cy­cle”. The data tend to show waves of di­s­or­der about ev­ery 40-60 years. Dur­ing the “in­te­gra­tive trend” (T&N’s term for the op­ti­mistic growth and stagfla­tion phases), these can just be minor protests or a small re­bel­lion that is eas­ily crushed. Dur­ing the “dis­in­te­gra­tive trend” (crisis + de­pres­sion), they usu­ally rep­re­sent in­di­vi­d­ual out­breaks of civil war. For ex­am­ple, dur­ing the Ro­man Repub­lic, the vi­o­lence around the death of Tiberius Grac­chus in 133 BC was rel­a­tively limited, be­cause Rome had not yet en­tered its crisis phase. 40 years later, in the depths of the crisis phase, there was a sec­ond out­break of vi­o­lence (91 – 82 BC) in­clud­ing the So­cial War and Sulla’s wars, which es­ca­lated to full-scale (though limited) civil war. 40 years later there was a third out­break (49 – 27 BC) in­clud­ing Cae­sar and Au­gus­tus’s very large civil wars. After that the new in­te­gra­tive trend started and fur­ther vi­o­lence was sup­pressed.

I’m not sure this even makes sense in the Sec­u­lar Cy­cle sys­tem, since it should ap­ply only to in­di­vi­d­ual coun­tries and not to the en­tire world, but these sure are some in­ter­est­ing data.

In Sec­u­lar Cy­cles, T&N mostly just iden­ti­fies this pat­tern from the data and doesn’t talk a lot about what causes it. But in some of Turchin’s other work, he ap­plies some of the math used to model epi­demics in pub­lic health. His model imag­ines three kinds of peo­ple: naives, rad­i­cals, and mod­er­ates. At the start of a cy­cle, most peo­ple are naive, with a few rad­i­cals. Rad­i­cals grad­u­ally spread rad­i­cal­ism, ei­ther by con­vert­ing their friends or pro­vok­ing their en­e­mies (eg a ter­ror­ist at­tack by one side con­vinces pre­vi­ously dis­en­gaged peo­ple to join the other side). This spreads like any other epi­demic. But as vi­o­lence gets worse, some peo­ple con­vert to “mod­er­ates”, here mean­ing not “wishy-washy peo­ple who don’t care” but some­thing more like “peo­ple dis­en­chanted with the cy­cle of vi­o­lence, de­ter­mined to get peace at any price”. Moder­ates sup­press rad­i­cals, but as they die off most peo­ple are naive and the cy­cle be­gins again. Us­ing var­i­ous pa­ram­e­ters for his model Turchin claims this pre­dicts the forty-to-sixty year cy­cle of vi­o­lence ob­served in the data.

So this is the ba­sic the­sis of Sec­u­lar Cy­cles. Pre-in­dus­trial his­tory op­er­ates on two cy­cles: first, a three-hun­dred year cy­cle of the rise-and-fall of civ­i­liza­tions. And sec­ond, a 40-60 year cy­cle of vi­o­lent di­s­or­der that only be­comes rele­vant dur­ing the low­est parts of the first cy­cle.


This is all in the first chap­ter of the book! The next eight chap­ters are case stud­ies of eight differ­ent his­tor­i­cal pe­ri­ods and how they fol­lowed the sec­u­lar cy­cle model.

For ex­am­ple, Chap­ter 7 is on the Ro­man Em­pire. It starts with Au­gus­tus in 27 BC. The Ro­man Repub­lic has just un­der­gone a hun­dred years of civil war, from the Grac­chi to Mar­ius to Sulla to Pom­pey to Cae­sar to Antony. All of this de­creased its pop­u­la­tion by 30% from its sec­ond-cen­tury peak. That means things are set to get a lot bet­ter very quickly.

The ex­pan­sion phase of the Em­pire lasted from Au­gus­tus (27 BC) to Nerva (96 AD), fol­lowed by a stagfla­tion phase from Nerva to An­to­nius Pius (165 AD). Through­out both phases, the pop­u­la­tion grew – from about 40 mil­lion in Au­gus­tus’ day to 65 mil­lion in An­to­nius’. Wheat prices stayed sta­ble un­til Nerva, then dou­bled from the be­gin­ning of the sec­ond cen­tury to its end. Le­gionary pay fol­lowed the in­verse pat­tern, stay­ing sta­ble un­til Nerva and then de­creas­ing by a third be­fore 200. The fi­nances of the state were the same – pretty good un­til the late sec­ond cen­tury (de­spite oc­ca­sional crazy peo­ple be­com­ing Em­peror and spend­ing the en­tire trea­sury build­ing stat­ues of them­selves), but cra­ter­ing dur­ing the time of Mar­cus Aure­lius and Com­modus (who de­based the denar­ius down to only 2 g silver).

Through­out ex­pan­sion and stagfla­tion, the Em­pire was rel­a­tively peace­ful (the “Pax Ro­mana”). Sure, oc­ca­sion­ally a crazy per­son would be­come Em­peror and they would have to kill him. There was even one small civil war which lasted all of a year (69 AD). But in gen­eral, these were iso­lated in­ci­dents.

Through­out the ex­pan­sion phase, up­ward mo­bil­ity was high and in­come in­equal­ity rel­a­tively low. T&N mea­sure this as how many con­suls (the high­est po­si­tion in the Ro­man gov­ern­men­tal hi­er­ar­chy) had fathers who were also con­suls. This de­creased through­out the first cen­tury – from 46% to 18% – then started creep­ing back up dur­ing the stagfla­tion phase to reach 32% at the end of the sec­ond cen­tury.

The crisis phase be­gan in 165 AD at the peak of Rome’s pop­u­la­tion and wealth. The An­to­nine Plague rav­aged the Em­pire, kil­ling up to 30% of the pop­u­la­tion. Fif­teen years later, the cen­tury-long dom­i­nance of the Good Em­per­ors ended, and Com­modus took the throne. Then he was mur­dered and Perti­nax took the throne. Then he was mur­dered and Didius Ju­li­anus took the throne. Then he was mur­dered and Sep­ti­mius Severus took the throne.

Now we are well into the dis­in­te­gra­tive trend, and the shorter 40-60 year cy­cle comes into play. Sep­ti­mus Sev­erius founds a dy­nasty that lasts 41 years, un­til Sep­ti­mius Alexan­der (the cousin of the grand­son of Sep­ti­mius Severus’ sister-in-law; it’s com­pli­cated) was as­sas­si­nated by his own sol­diers in Ger­many. This be­gins the Cri­sis Of The Third Cen­tury, a time of con­stant civil war, mass de­pop­u­la­tion, and eco­nomic col­lapse. The Five Good Em­per­ors of the sec­ond cen­tury ruled 84 years be­tween them (av­er­age of 17 years per em­peror). The fifty year Cri­sis in­cluded 27 em­per­ors, for an av­er­age of less than 2 years per em­peror.

Fi­nally, in 284, Em­peror Dio­cle­tian ended the civil wars, re-es­tab­lished cen­tral­ized au­thor­ity, and es­sen­tially re­founded the Ro­man Em­pire – a nice round 310 years af­ter Au­gus­tus did the same. T&N mark this as the end of a sec­u­lar cy­cle and the be­gin­ning of a new in­te­gra­tive trend.

T&N are able to tell this story. But they don’t just tell the story. They are able to cite var­i­ous statis­tics to back them­selves up. The Ro­man pop­u­la­tion statis­tics. The price of wheat and other food­stuffs. The av­er­age wages for la­bor­ers. They es­pe­cially like coin hoards – the amount of buried trea­sure from a given pe­riod dis­cov­ered by trea­sure-hunters – be­cause they ar­gue you only bury your money dur­ing times of in­sta­bil­ity, so this forms a semi-ob­jec­tive way of mea­sur­ing how un­sta­ble things are.

They are at their best when pre­sent­ing very broad sum­mary statis­tics. For ex­am­ple, Ro­man in­dus­try pro­duced vast amounts of lead, which en­tered the at­mo­sphere and set­tled into the Green­land ice sheet. Here is Ro­man lead out­put per year as mea­sured in ice cores:

This shows four peaks for the four cy­cles T&N iden­tify in Rome: the King­dom, the Repub­lic, the Early Em­pire of Au­gus­tus (Prin­ci­pate, the one de­scribed above), and the Late Em­pire of Dio­cle­tian (Dom­i­nate). It even shows a saw­tooth-y pat­tern cor­re­spond­ing to the shorter bi­gen­er­a­tional cy­cles.

Or here is build­ing ac­tivity in Rome, mea­sured by how many build­ings ar­chae­ol­o­gists have found from a given time:

This is a lit­tle less perfect (why is there a big gap in the mid­dle of the Prin­ci­pate? I guess Au­gus­tus is a hard act to fol­low, build­ing-wise) but it still looks good for the cy­cle the­ory.

And here is an In­dex Of Poli­ti­cal In­sta­bil­ity, which “com­bines mea­sures of du­ra­tion, in­ten­sity, and scale of poli­ti­cal in­sta­bil­ity events, coded by a team of pro­fes­sional his­to­ri­ans”:

Rome is the one on top. In­sta­bil­ity clearly peaks dur­ing the crisis-de­pres­sion phases be­tween T&N’s sec­u­lar cy­cles – again with a saw­tooth pat­tern rep­re­sent­ing the bi­gen­er­a­tional cy­cles.


See­ing pat­terns in ran­dom noise is one of the ba­sic hu­man failure modes. Sec­u­lar Cy­cles is so prima fa­cie crack­pot­tish that it should re­quire moun­tains of data be­fore we even start won­der­ing if it might be true. I want to make it clear that the book – plus Turchin in­di­vi­d­u­ally in some of his other books and pa­pers – pro­vides these moun­tains. I can’t show ev­ery sin­gle case study, graph, and table in this book re­view. But the chap­ter above on the Ro­man Prin­ci­pate in­cluded 25 named figures and graphs, plus countless more in­for­mal pre­sen­ta­tions of data se­ries, from “real wages of agri­cul­tural la­bor­ers in Ro­man Egypt dur­ing the sec­ond cen­tury” to “mean an­nual real land rents for wheat fields in artabas per aroura, 27 BC to 268 CE” to “im­pe­rial hand­outs per reign-year” to “im­por­ta­tion of Afri­can red slip ware into the Albegna Valley of Etruria, 100 – 600”. And this is just one chap­ter, ran­domly cho­sen. There are seven oth­ers just like this. This book un­der­stands the bur­den of proof it is un­der, and does ev­ery­thing it can to meet it.

Still, we should be skep­ti­cal. How many de­grees of free­dom do T&N have, and is it enough to un­der­mine their case?

First, they get some free­dom in the civ­i­liza­tions they use as case stud­ies. They could have searched through ev­ery re­gion and pe­riod and cherry-picked eight civ­i­liza­tions that rose and fell over a pe­ri­ods of three hun­dred years. Did they? I don’t think so. The case stud­ies are England, France, Rome, and Rus­sia. Th­ese are some of the civ­i­liza­tions of great­est in­ter­est to the English-speak­ing world (ex­cept Rus­sia, which makes sense in con­text be­cause the au­thors are both Rus­sian). They’re also some of the civ­i­liza­tions best-stud­ied by An­glo­phone his­to­ri­ans and with the most data available (the au­thors’ method­ol­ogy re­quires hav­ing good time-se­ries of pop­u­la­tions, bud­gets, food pro­duc­tion, etc).

Also, it’s not too hard to look at the civ­i­liza­tions they didn’t study and fill in the gaps. The book barely men­tions China, but it seems to fit the model pretty well (“the em­pire united longs to di­vide; di­vided longs to unite”). In fact, tak­ing the quo­ta­tion com­pletely se­ri­ously – the em­pire was first united dur­ing the Qin Dy­nasty start­ing in 221 BC, which lasted only 20 years be­fore segu­ing into the Han Dy­nasty in 202 BC. The Han ex­panded and pros­pered for about a cen­tury, had an­other cen­tury of com­pli­cated in­trigue and fre­quently re­volt, and then ended in dis­aster in the first part of the first cen­tury, with a set of failed re­forms, civil war, the sack of the cap­i­tal, some more civil war, peas­ant re­volt, and even more civil war. The sep­a­rate pe­riod of the Eastern Han Dy­nasty be­gan in 25 AD, about 240 years af­ter the be­gin­ning of the Qin-Han cy­cle. The Eastern Han also grew and pros­pered for about a hun­dred years, then had an­other fifty years of sim­mer­ing dis­con­tent, then fell apart in about 184 AD, with an­other se­ries of civil wars, peas­ant re­bel­lions, etc. This was the Three King­doms Pe­riod dur­ing which “the em­pire united longs to di­vide, di­vided longs to unite” was writ­ten to de­scribe. It lasted an­other eighty years un­til 266 AD, af­ter which the Jin Dy­nasty be­gan. The Jin Dy­nasty was kind of crap, but it lasted an­other 180 years un­til 420, fol­lowed by 160 years of di­vi­sion, fol­lowed by the Sui and Tang dy­nas­ties, which were not crap. So I don’t think it takes too much pat­tern-match­ing to iden­tify a Western-Han-to-Eastern-Han Cy­cle of 240 years, fol­lowed by an Eastern-Han-to-Jin Cy­cle of 241 years, fol­lowed by a Jin-to-Sui/​Tang-Cy­cle of 324 years.

One could make a more hos­tile anal­y­sis. Is it re­ally fair to lump the Western Jin and Eastern Jin con­ve­niently to­gether, but sep­a­rate the Western Han and Eastern Han con­ve­niently apart? Is it re­ally fair to call the crappy and re­volt-prone Jin Dy­nasty an “in­te­gra­tive trend” rather than a dis­in­te­gra­tive trend that lasted much longer than the the­ory should pre­dict? Is it re­ally fair to round off cy­cles of 240 and 320 years to “ba­si­cally 300 years”?

I think the an­swer to all of these is “T&N aren’t mak­ing pre­dic­tions about the length of Chi­nese dy­nas­ties, they’re mak­ing pre­dic­tions about the na­ture of sec­u­lar cy­cles, which are cor­re­lated with dy­nas­ties but not iden­ti­cal to them”. If I had the equiv­a­lent to lead core read­ings for China, or an “in­sta­bil­ity in­dex”, or time se­ries data for wages or health or pot­tery im­por­ta­tion or so on, maybe it would be perfectly ob­vi­ous that the Eastern and Western Han defined two differ­ent pe­ri­ods, but the Eastern and Western Jin were part of the same pe­riod – the same way one look at the lead core data for Rome shows that the Julio-Clau­dian dy­nasty vs. the Fla­vian Dy­nasty is not an in­ter­est­ing tran­si­tion.

A sec­ondary an­swer might be that T&N ad­mit all sorts of things can al­ter the length of sec­u­lar cy­cles. They trag­i­cally de­vote only a few pages to “Ibn Khal­dun cy­cles”, the the­ory of 14th cen­tury Ara­bic his­to­rian Ibn Khal­dun that civ­i­liza­tions in the Maghreb rise and fall on a one hun­dred year pe­riod. But they dis­cuss it just enough to say their data con­firm Ibn Khal­dun’s ob­ser­va­tions. The ac­cel­er­ated timescale (100 vs. 300 years) is be­cause the Maghreb is mas­sively polyg­y­nous, with suc­cess­ful lead­ers hav­ing harems of hun­dreds of con­cu­bines. This speeds up the elite over­pro­duc­tion pro­cess and makes ev­ery­thing hap­pen in fast-for­ward. T&N also ad­mit that their the­ory only de­scribes civ­i­liza­tions in­so­far as they are self-con­tained. This ap­prox­i­mately holds for hege­mons like Rome at its height, but fails for eg Poland, whose his­tory is go­ing to be much more in­fluenced by when Rus­sia or Ger­many de­cides to in­vade than by the in­ter­nal mechanisms of Pol­ish so­ciety. In­so­far as ex­ter­nal shocks – whether cli­matic, for­eign mil­i­tary, or what­ever else – af­fect a civ­i­liza­tion, sec­u­lar cy­cles will be stretched out, com­pressed, or just to­tally ab­sent.

This sort of thing must ob­vi­ously be true, and it’s good T&N say it, but it’s also a free pass to add as many epicy­cles as you need to ex­plain failure to match data. All I can say look­ing at China is that, if you give it some wig­gle room, it seems to fit T&N’s the­o­ries okay. The same is true of a bunch of other civ­i­liza­tions I plugged in to see if they would work.

Se­cond, T&N get some de­grees of free­dom based on what statis­tics they use. In ev­ery case, they pre­sent statis­tics that sup­port the pres­ence of sec­u­lar cy­cles, but they’re not the same statis­tics in ev­ery case. On the one hand, this is un­avoid­able; we may not have good wage data for ev­ery civ­i­liza­tion, and lev­els of pot­tery im­por­ta­tion might be more rele­vant to an­cient Rome than to 19th-cen­tury Rus­sia. On the other hand, I’m not sure what pre­vents them from just never men­tion­ing the In­sta­bil­ity In­dex if the In­sta­bil­ity In­dex doesn’t show what they want it to show.

Here are some ran­dom Rome-re­lated in­di­ca­tors I found on­line:

None of them show the same four-peaked King­dom-Repub­lic-Prin­ci­pate-Dom­i­nate pat­tern as the ones Sec­u­lar Cy­cles cites, or the ones Turchin has on­line.

Third, a lot of the statis­tics them­selves have some de­grees of free­dom. A lot of them are things like “In­sta­bil­ity In­dex” or “In­dex of So­cial Well-Be­ing” or “Gen­eral Bad­ness In­dex”. Th­ese seem like the kind of scores you can fid­dle with to get the re­sults you want. Turchin claims he hasn’t fid­dled with them – his in­sta­bil­ity in­dex is taken from a 1937 pa­per I haven’t been able to find. But how many pa­pers like that are there? Am I get­ting too con­spir­a­to­rial now?

Like­wise, we don’t have di­rect ac­cess to the bud­get of the Ro­man Em­pire (or Plan­ta­genet England, or…). His­to­ri­ans have tried to re­con­struct it based on ar­chae­ol­ogy and the few records that have sur­vived. T&N cite these peo­ple, and the peo­ple they cite are at the top of their fields and say what T&N say they say. But how much flex­i­bil­ity did they have in de­cid­ing which es­ti­mate of the Ro­man bud­get to cite? Is there enough dis­agree­ment that they could cite the high es­ti­mate for one pe­riod and the low es­ti­mate for an­other, then prove it had gone down? I don’t know (though a few hours’ work ought to be enough to es­tab­lish this).

I wish I could find com­men­tary by other aca­demics and his­to­ri­ans on Sec­u­lar Cy­cles, or Turchin’s work more gen­er­ally. I feel like some­body should ei­ther be vi­o­lently de­bunk­ing this, or else throw­ing the au­thors a ticker-tape pa­rade for hav­ing solved his­tory. Nei­ther is hap­pen­ing. The few com­ments I can find are mostly limited to naval gaz­ing about whether his­tory should be quan­ti­ta­tive or qual­i­ta­tive. The few ex­cep­tions I can find are blog posts by peo­ple I know and re­spect urg­ing me to read Turchin five years ago, ad­vice I am sorry for not tak­ing. If you know of any good crit­i­cism, please tell me where to find it.

Un­til then, my very quick dou­ble-check­ing sug­gests T&N are pretty much on the level. But there could still be sub­tler forms of overfit­ting go­ing on that I don’t know enough about his­tory to de­tect.


If this is true, does it have any im­pli­ca­tions for peo­ple to­day?

First, a very weak im­pli­ca­tion: it makes his­tory eas­ier to learn. I was shocked how much more I re­mem­bered about the Plan­ta­genets, Tu­dors, Capeti­ans af­ter read­ing this book than af­ter read­ing any nor­mal his­tory book about them. I think the se­cret in­gre­di­ent is struc­ture. If his­tory is just “one damn thing af­ter an­other”, there’s no frame­work for figur­ing out what mat­ters, what’s worth learn­ing, what fol­lows what else. The sec­u­lar cy­cle idea cre­ates a struc­ture that ev­ery­thing fits into neatly. I know that the Plan­ta­genet Dy­nasty lasted from 1154 – 1485, be­cause it had to, be­cause that’s a 331 year sec­u­lar cy­cle. I know that the im­por­tant events to re­mem­ber in­clude the Anar­chy of 1135 – 1153 and the War of the Roses from 1455 – 1487, be­cause those are the two crisis-de­pres­sion pe­ri­ods that frame the cy­cle. I know that af­ter 1485 Henry Tu­dor took the throne and be­gan a new age of English his­tory, be­cause that’s the be­gin­ning of the in­te­gra­tive phase of the next cy­cle. All of this is a lot eas­ier than try­ing to re­mem­ber these names and dates ab­sent any con­text. I would recom­mend this book for that rea­son alone.

Se­cond, I think this might give new con­text to Piketty on in­equal­ity. T&N de­scribe in­equal­ity as start­ing out very low dur­ing the growth phase of a sec­u­lar cy­cle, ris­ing to a peak dur­ing the stagfla­tion phase, then drop­ping pre­cip­i­tously dur­ing the crisis. Piketty de­scribes the same: in­equal­ity ris­ing through the peace­ful pe­riod of 1800 to 1900, drop­ping pre­cip­i­tously dur­ing the two World Wars, then grad­u­ally ris­ing again since then. This doesn’t make a huge amount of sense, since I’m not sure you can fit the post in­dus­trial world into sec­u­lar cy­cles. But I no­tice Piketty seems to think of this as a once-off event – in­equal­ity has been ris­ing for­ever, bro­ken only by the freak crisis of the two World Wars – and it’s in­ter­est­ing to read T&N talk about the ex­act same pro­cess re­cur­ring again and again through­out his­tory.

Fi­nally, and most im­por­tant: is there any sense in which this is still go­ing on?

The eas­iest an­swer would be no, there isn’t. The sec­u­lar cy­cles are based around Malthu­sian pop­u­la­tion growth, but we are now in a post-Malthu­sian regime where land is no longer the limit­ing re­source. And the cy­cles seem to as­sume huge crises kil­ling off 30% to 50% of the pop­u­la­tion, but those don’t hap­pen any­more in First World coun­tries; the Civil War was the blood­iest pe­riod of US his­tory, and even it only kil­led 2% of Amer­i­cans. Even Ger­many only lost about 15% of its pop­u­la­tion in World Wars I + II.

But Turchin has an­other book, Ages Of Dis­cord, ar­gu­ing that they do. I have bought it and started it and will re­port back when I’m done.

Even with­out a frame­work, this is just in­ter­est­ing to think about. In pop­u­lar un­der­stand­ing of Amer­i­can his­tory, you can trace out op­ti­mistic and pes­simistic pe­ri­ods. The na­tional nar­ra­tive seems to in­clude a story of the 1950s as a golden age of op­ti­mism. Then ev­ery­one got an­gry and vi­o­lent in the early 1970s (the Sta­tus 451 re­view of Days Of Rage is pretty great here, and re­minds us that “peo­ple have com­pletely for­got­ten that in 1972 we had over nine­teen hun­dred do­mes­tic bomb­ings in the United States”). Then ev­ery­thing sud­denly got bet­ter once Rea­gan de­clared “morn­ing in Amer­ica” in the 1980s, with an era of op­ti­mism and good feel­ings last­ing through the Clin­ton ad­minis­tra­tion. Then things start­ing to turn bad some­time around Bush II. And now ev­ery­body hates each other, and fas­cists and an­tifa are fight­ing in the streets, and peo­ple are talk­ing about how “ci­vil­ity” and “bi­par­ti­san­ship” are evil tools of op­pres­sion, and Pre­dic­tIt says an avowed so­cial­ist has a 10% chance of be­com­ing pres­i­dent of the US. To what ex­tent is this nar­ra­tive true? I don’t know, but it’s definitely the nar­ra­tive.

One thing that strikes me about T&N’s cy­cles is the ide­olog­i­cal com­po­nent. They de­scribe how, dur­ing a growth phase, ev­ery­one is op­ti­mistic and pa­tri­otic, se­cure in the knowl­edge that there is enough for ev­ery­body. Dur­ing the stagfla­tion phase, in­equal­ity in­creases, but con­cern about in­equal­ity in­creases even more, zero-sum think­ing pre­dom­i­nates, and so­cial trust craters (both be­cause peo­ple are ac­tu­ally defect­ing, and be­cause it’s in lots of peo­ple’s in­ter­est to play up the de­gree to which peo­ple are defect­ing). By the crisis phase, par­ti­san­ship is much stronger than pa­tri­o­tism and rad­i­cals are talk­ing openly about how vi­o­lence is eth­i­cally obli­ga­tory.

And then, even­tu­ally, things get bet­ter. There is a new Au­gus­tan Age of virtue and the reestab­lish­ment of all good things. This is a re­ally in­ter­est­ing claim. Western philos­o­phy tends to think in terms of trends, not cy­cles. We see ev­ery­thing go­ing on around us, and we think this is some end­less trend to­wards more par­ti­san­ship, more in­equal­ity, more ha­tred, and more state dys­func­tion.

Sec­u­lar Cy­cles offers a nar­ra­tive where end­less trends can end, and things can get bet­ter af­ter all. Of course, it also offers a nar­ra­tive where some­times this pro­cess in­volves the death of 30% – 50% of the pop­u­la­tion. Maybe I should read Turchin’s other books be­fore spec­u­lat­ing any fur­ther.