How democracy ends: a review and reevaluation

Last month I at­tended a talk by David Runci­man, the au­thor of a re­cent book called How Democ­racy Ends. I was pre­pared for out­rage-stir­ring and pearl-clutch­ing, but was pleas­antly sur­prised by the qual­ity of his ar­gu­ments, which I’ve sum­marised be­low, along with my own thoughts on these is­sues. Note, how­ever, that I haven’t read the book it­self, and so can’t be con­fi­dent that I’ve por­trayed his ideas faith­fully.

Which les­sons from his­tory?

Many peo­ple have com­pared re­cent pop­ulist move­ments with the stir­rings of fas­cism a cen­tury ago. And in­deed it’s true that a lot of similar rhetoric be­ing thrown around. But Runci­man ar­gues that this is one of the least in­ter­est­ing com­par­i­sons to be made be­tween these two times. Some things that would be much more sur­pris­ing to a denizen of the early 20th cen­tury:

  • Sig­nifi­cant ad­vances in technology

  • Mas­sive trans­for­ma­tions in so­cietal demographics

  • Very few changes in our institutions

The last of those is par­tic­u­larly sur­pris­ing in light of the first two. Par­li­a­men­tary democ­ra­cies in the An­glo­phone world have been gov­erned by the same in­sti­tu­tions—and in some cases, even the same par­ties—for cen­turies. Con­ti­nen­tal Euro­pean democ­ra­cies were more dis­rupted by World War 2, but have been very sta­ble since then, de­spite the world chang­ing in many ways over­all. That’s true even for in­sti­tu­tions that are prob­a­bly harm­ful—con­sider the per­sis­tence of the elec­toral col­lege in the US, the House of Lords in the UK, the monar­chy in Aus­tralia (de­spite their ar­dent re­pub­li­can­ism move­ment), first-past-the-post vot­ing sys­tems in many coun­tries, and so on. (In fact, Runci­man spec­u­lates that Amer­i­cans voted for Trump partly be­cause of how much con­fi­dence they had in the dura­bil­ity of their in­sti­tu­tions—a con­fi­dence which so far seems to have been well-founded.)

So his­tory gives us pretty good ev­i­dence for the ro­bust­ness of demo­cratic in­sti­tu­tions to the nor­mal flow of time—but not to ex­cep­tional cir­cum­stances. In fact, an in­abil­ity to make nec­es­sary changes may well ren­der them more frag­ile in the face of sharp op­po­si­tion. If and when pres­sure mounts, are they go­ing to snap like the demo­cratic in­sti­tu­tions of 1930s Europe did? Runci­man ar­gues that they won’t, be­cause of the na­ture of the de­mo­graphic changes the West has seen. There are three par­tic­u­larly im­por­tant axes of vari­a­tion:

  • Wealth: the av­er­age per­son is many times wealthier than they were a cen­tury ago, and the mid­dle class is much larger.

  • Ed­u­ca­tion: we’ve gone from only a few per­cent of peo­ple get­ting ter­tiary ed­u­ca­tion (and many of the re­main­der not finish­ing high school) to nearly 50% of young peo­ple be­ing in uni­ver­sity in many west­ern coun­tries.

  • Age: in the last cen­tury, the me­dian age has risen by over ten years in most west­ern coun­tries.

Th­ese three fac­tors are some of the most pow­er­ful pre­dic­tors of be­havi­our that we have, and so we should take them into ac­count when judg­ing the like­li­hood of demo­cratic failure. For in­stance, wealthier and more ed­u­cated peo­ple are much less likely to sup­port pop­ulist or ex­trem­ist groups. But Runci­man fo­cuses the most on age, which I think is the cor­rect ap­proach. Wealth is rel­a­tive—even if peo­ple are ac­tu­ally much richer, they can feel poor and an­gry af­ter a re­ces­sion (as they did in the 1930s, de­spite still be­ing many times wealthier than al­most all their an­ces­tors). Ed­u­ca­tion may just be cor­re­lated with other fac­tors, rather than the ac­tual cause of last­ing differ­ences in mind­set. But there are pretty clear biolog­i­cal and so­cial rea­sons to think that the be­havi­our and pri­ori­ties of older peo­ple are ro­bustly and sig­nifi­cantly differ­ent from those of younger peo­ple. You need only look at the age dis­tri­bu­tion of vi­o­lent crime, for ex­am­ple, to see how strong this effect is (al­though it may have less­ened some­what over re­cent years, since mar­riage rates are de­clin­ing and sin­gle men cause more trou­ble).

In short: the failures of democ­racy in the 30’s were based on large pop­u­la­tions of young men who could be mo­bil­ised in anger by mil­i­taris­tic lead­ers—see for in­stance the brown­shirts in Ger­many and black­shirts in Italy. But that’s not what the failure of democ­racy in our time would look like, be­cause that group of peo­ple is much smaller now. For bet­ter or worse, older pop­u­la­tions are less dis­rup­tive and more com­pla­cent. To see where that might lead, con­sider Ja­pan: an age­ing pop­u­la­tion which can’t drag it­self out of eco­nomic tor­por, re­sis­tant to im­mi­gra­tion, dom­i­nated for decades by one poli­ti­cal party, bet­ting the coun­try’s fu­ture on us­ing robots to re­place the miss­ing work­force.

Changes ahead

Dur­ing a Q&A af­ter the talk, I pointed out that Ja­pan is very differ­ent to west­ern coun­tries in its par­tic­u­larly strong cul­ture of so­cial con­for­mity and sta­bil­ity. Age trends notwith­stand­ing, I have much more difficulty imag­ing the same quiet tol­er­ance of slow de­cline oc­cur­ring in the US or UK. So, given that gov­ern­ment in­sti­tu­tions are very difficult to change, where will peo­ple di­rect their frus­tra­tion if lack­lus­tre growth con­tinues in the com­ing decades?

In re­sponse, Runci­man raised two pos­si­bil­ities. Firstly, that peo­ple will “go around their gov­ern­ments”, find­ing new do­mains in which poli­tics is less rele­vant. We could call this the “Wild West” pos­si­bil­ity. Of course, there’s no longer an un­colon­ised West to ex­plore—but there is the in­ter­net, which isn’t demo­crat­i­cally run and prob­a­bly never will be. We already see fewer young men work­ing full-time, be­cause the al­ter­na­tive of spend­ing most of their time gam­ing has be­come more ap­peal­ing. As vir­tual wor­lds be­come even more im­mer­sive, it seems plau­si­ble that peo­ple will be­gin to care much less about poli­ti­cal is­sues.

One prob­lem with the idea of “go­ing around gov­ern­ments”, though, is that gov­ern­ments are just much big­ger now than they used to be. And as tech­nol­ogy com­pa­nies profit from the grow­ing role of the in­ter­net, there’ll be pres­sure for gov­ern­ments to in­ter­vene even more to fight in­equal­ity. So a sec­ond op­tion is a more Chi­nese ap­proach, with in­creas­ingly au­to­cratic Western gov­ern­ments ex­ert­ing heavy pres­sure on (and per­haps even­tu­ally con­trol over) tech com­pa­nies.

A more op­ti­mistic pos­si­bil­ity is for the in­ter­net to make democ­racy more ac­countable. Runci­man in­vites us to con­sider Plato’s origi­nal ar­gu­ment against di­rect democ­racy (in which peo­ple vote on in­di­vi­d­ual is­sues) - that it would lead to rule by the poor, the ig­no­rant, and the young, all of whom nec­es­sar­ily out­num­ber the wealthy, wise and old. This ar­gu­ment turned out not to ap­ply for rep­re­sen­ta­tive democ­racy, since elected rep­re­sen­ta­tives tend to be wealthy, ed­u­cated and old de­spite their con­stituents be­ing the op­po­site. But now it’s in­ap­pli­ca­ble for a differ­ent rea­son—that al­though our rep­re­sen­ta­tives haven’t changed much, the rest of us are start­ing to look much more like them. So maybe it’ll be­come fea­si­ble to im­ple­ment a more di­rect democ­racy, fa­cil­i­tated by the in­ter­net and mod­ern com­mu­ni­ca­tion tech­nol­ogy. (This still seems like a bad idea to me, though.)

Base rates and complacency

The last sec­tion was a lit­tle spec­u­la­tive, so let’s take a step back and think about how to make pre­dic­tions about these sorts of events in gen­eral. Runci­man’s anal­y­sis above pro­vides good rea­sons not to draw a spe­cific par­allel be­tween the rise of fas­cism last cen­tury and re­cent poli­ti­cal events. But it would take ex­traor­di­nary ev­i­dence to ex­empt us from ex­trap­o­lat­ing broader his­tor­i­cal trends, in par­tic­u­lar the fact that states always col­lapse even­tu­ally, and that the base rate for coups and other forms of in­ter­nal strife is fairly high. Are the ex­traor­di­nary changes we’ve seen since the in­dus­trial rev­olu­tion suffi­cient to jus­tify be­lief in our ex­cep­tion­al­ism?

It’s true that since World War 2, al­most no wealthy democ­ra­cies have de­scended into au­toc­racy or chaos (Turkey and Ire­land be­ing two edge cases). It’s also true that, de­spite wide­spread poli­ti­cal dis­illu­sion­ment, norms against vi­o­lence have held to a re­mark­ably large de­gree. But draw­ing judge­ments from the his­tor­i­cal pe­riod “since World War 2” is a clas­sic case of the Texan Sharp­shooter’s Fal­lacy (and pos­si­bly also an­thropic bias?). In fact, this re­cent lull should make us skep­ti­cal about our abil­ity to eval­u­ate the ques­tion ob­jec­tively, be­cause peo­ple are in gen­eral very bad at an­ti­ci­pat­ing ex­treme events that haven’t oc­curred in liv­ing mem­ory. I think this is true de­spite these pos­si­bil­ities be­ing dis­cussed in the me­dia. For ex­am­ple, while there’s a lot of talk about Trump be­ing a po­ten­tial au­to­crat, few Amer­i­cans are re­spond­ing by stock­piling food or in­vest­ing in for­eign cur­ren­cies or em­i­grat­ing. This sug­gests that hos­tility to­wards Trump is driven pri­mar­ily by par­ti­san poli­tics, rather than gen­uine con­cern about demo­cratic col­lapse. An ad­di­tional data point in favour of this hy­poth­e­sis is how eas­ily the Repub­li­can poli­ti­cal es­tab­lish­ment has fallen in line.

Another key ques­tion which isn’t of­ten dis­cussed is the na­ture of mod­ern mil­i­tary cul­ture. His­tor­i­cally, this has been a ma­jor fac­tor af­fect­ing gov­ern­men­tal sta­bil­ity. But, apart from vague in­tu­itions about mod­ern mil­i­taries be­ing fairly placid, I find my­self re­mark­ably ig­no­rant on this sub­ject, and sus­pect oth­ers are as well. What facts do you know about your coun­try’s mil­i­tary, about the char­ac­ter of its com­man­ders or the dis­tri­bu­tion of power within it, that make you con­fi­dent that it won’t launch a coup if, for ex­am­ple, one of its gen­er­als is nar­rowly defeated in a dis­puted pres­i­den­tial elec­tion (as in Gore vs Bush)? Note that mil­i­tary de­mo­graph­ics haven’t changed nearly as much as those of our so­cieties over­all. They’re still pri­mar­ily com­posed of young work­ing-class men with­out de­grees—a group that’s un­usu­ally an­gry about to­day’s poli­tics. So while I am pretty con­vinced by Runci­man’s ar­gu­ments, this is one way in which they may not ap­ply. Also con­sider that war­fare is much less hands-on than it used to be, and fire­power much more cen­trally con­cen­trated, both of which make coups eas­ier.

And what about ex­treme events?

So far I’ve looked at so­cietal col­lapse from a poli­ti­cal point of view. But many his­tor­i­cal tran­si­tions were pre­cip­i­tated by nat­u­ral dis­asters or dis­eases. See, for in­stance, the Mayan col­lapse, or the Lit­tle Ice Age, or the Black Death. To­day, we’re much safer from nat­u­ral dis­asters, both be­cause of our tech­nol­ogy and be­cause of the scale of our so­cieties—few peo­ple live in coun­tries in which the ma­jor­ity of a pop­u­la­tion can be struck by a sin­gle nat­u­ral dis­aster. Similarly, we’re also much safer from nat­u­ral dis­eases. But we’re much more vuln­er­a­ble to se­vere man-made dis­asters, which I think are very likely to oc­cur over the next cen­tury. Since this post is fo­cused on poli­ti­cal col­lapse as a dis­tinct phe­nomenon to tech­nolog­i­cal dis­aster, I won’t dis­cuss ex­treme risks from tech­nol­ogy here. How­ever, it’s worth­while to look at the ways in which smaller tech­nolog­i­cal harms might ex­ac­er­bate other trends. AI-caused un­em­ploy­ment and the more gen­eral trend to­wards bi­modal out­comes in west­ern coun­tries are likely to cause so­cial un­rest. Mean­while ter­ror­ism is go­ing to be­come much eas­ier—con­sider be­ing able to 3D-print as­sas­sin drones run­ning fa­cial recog­ni­tion soft­ware, for in­stance. And due to an­tibiotic overuse, it’s likely that our safety from dis­ease will de­cline over the com­ing years (even with­out the ad­di­tional dan­ger of bioter­ror­ism us­ing en­g­ineered dis­eases). Fi­nally, I think we’re much softer than we used to be—it won’t take nearly as much dan­ger to dis­rupt a coun­try. Runci­man is prob­a­bly cor­rect that we’re less sus­cep­ti­ble to a col­lapse into au­thor­i­tar­i­anism than we were in the past—but the same trends driv­ing that change are also push­ing us to­wards new rea­sons to worry.

In ad­di­tion to the talk by Runci­man, this post was in­spired by dis­cus­sions with my friends Todor and Julio, and benefited from their feed­back.