Cortés, Pizarro, and Afonso as Precedents for Takeover

Cross­posted from AI Im­pacts.

Epistemic sta­tus: I am not a his­to­rian, nor have I in­ves­ti­gated these case stud­ies in de­tail. I ad­mit I am still un­cer­tain about how the con­quis­ta­dors were able to colonize so much of the world so quickly. I think my ig­no­rance is ex­cus­able be­cause this is just a blog post; I wel­come cor­rec­tions from peo­ple who know more. If it gen­er­ates suffi­cient in­ter­est I might do a deeper in­ves­ti­ga­tion. Even if I’m right, this is just one set of his­tor­i­cal case-stud­ies; it doesn’t prove any­thing about AI, even if it is sug­ges­tive. Fi­nally, in de­scribing these con­quis­ta­dors as “suc­cess­ful,” I sim­ply mean that they achieved their goals, not that what they achieved was good.


In the span of a few years, some minor Euro­pean ex­plor­ers (later known as the con­quis­ta­dors) en­coun­tered, con­quered, and en­slaved sev­eral huge re­gions of the world. That they were able to do this is sur­pris­ing; their tech­nolog­i­cal ad­van­tage was not huge. (This was be­fore the sci­en­tific and in­dus­trial rev­olu­tions.) From these cases, I think we learn that it is oc­ca­sion­ally pos­si­ble for a small force to quickly con­quer large parts of the world, de­spite:

  1. Hav­ing only a minus­cule frac­tion of the world’s re­sources and power

  2. Hav­ing tech­nol­ogy + diplo­matic and strate­gic cun­ning that is bet­ter but not that much better

  3. Hav­ing very lit­tle data about the world when the con­quest begins

  4. Be­ing disunited

Which all sug­gests that it isn’t as im­plau­si­ble that a small AI takes over the world in mildly fa­vor­able cir­cum­stances as is some­times thought.

EDIT: In light of good push­back from peo­ple (e.g. Lucy.ea8 and e.g. Matthew Bar­nett) about the im­por­tance of dis­ease, I think one should prob­a­bly add a caveat to the above: “In times of chaos & dis­rup­tion, at least.”

NEW EDIT: After read­ing three gi­ant his­tory books on the sub­ject, I take back my pre­vi­ous edit. My origi­nal claims were cor­rect.

Three shock­ing true stories

I highly recom­mend you read the wiki pages your­self; oth­er­wise, here are my sum­maries:

Cortés: [wiki] [wiki]

  • April 1519: Hernán Cortés lands in Yu­catan with ~500 men, 13 horses, and a few can­nons. He de­stroys his ships so his men won’t be able to re­treat. His goal is to con­quer the Aztec em­pire of sev­eral mil­lion peo­ple.

  • He makes his way to­wards the im­pe­rial cap­i­tal, Tenochti­tlán. Along the way he en­coun­ters var­i­ous lo­cal groups, fight­ing some and al­ly­ing with some. He is con­stantly out­num­bered but his tech­nol­ogy gives him an ad­van­tage in fights. His force grows in size, be­cause even though he loses Spa­niards he gains lo­cal al­lies who re­sent Aztec rule.

  • Tenochti­tlán is an is­land fortress (like Venice) with a pop­u­la­tion of over 200,000, mak­ing it one of the largest and rich­est cities in the world at the time. Cortés ar­rives in the city ask­ing for an au­di­ence with the Em­peror, who re­ceives him war­ily.

  • Cortés takes the em­peror hostage within his own palace, in­di­rectly rul­ing Tenochti­tlán through him.

  • Cortés learns that the Span­ish gov­er­nor has landed in Mex­ico with a force twice his size, in­tent on ar­rest­ing him. (Cortés’ ex­pe­di­tion was ille­gal!) Cortés leaves 200 men guard­ing the Em­peror, marches to the coast with the rest, sur­prises and defeats the new Spa­niards in bat­tle, and in­cor­po­rates the sur­vivors into his army.

  • July 1520: Back at the cap­i­tal, the lo­cals are start­ing to rebel against his men. Cortés marches back to the cap­i­tal, unit­ing his forces just in time to be be­sieged in the im­pe­rial palace. They mur­der the em­peror and fight their way out of the city overnight, tak­ing heavy losses.

  • They shelter in an­other city (Tlax­cala) that was think­ing about re­bel­ling against the Aztecs. Cortés al­lies with the Tlax­calans and launches a gen­eral up­ris­ing against the Aztecs. Not ev­ery­one sides with him; many city-states re­main loyal to Tenochti­t­lan. Some try to stay neu­tral. Some join him at first, and then aban­don him later. Smal­lpox sweeps through the land, kil­ling many on all sides and caus­ing gen­eral chaos.

  • May 1521: The fi­nal as­sault on Tenochti­tlán. By this point, Cortés has about 1,000 Span­ish troops and 80,000 − 200,000 al­lied na­tive war­riors. He had 16 can­nons and 13 boats. The Aztecs have 80,000 − 300,000 war­riors and 400 boats. Cortés and his al­lies win.

  • Later, the Span­ish would be­tray their na­tive al­lies and as­sert hege­mony over the en­tire re­gion, in vi­o­la­tion of the treaties they had signed.

Pizarro [wiki] [wiki]

  • 1532: Fran­cisco Pizarro ar­rives in Inca ter­ri­tory with 168 Span­ish sol­diers. His goal is to con­quer the Inca em­pire, which was much big­ger than the Aztec em­pire.

  • The Inca em­pire is in the mid­dle of a civil war and a dev­as­tat­ing plague.

  • Pizarro makes it to the Em­peror right af­ter the Em­peror defeats his brother. Pizarro is al­lowed to ap­proach be­cause he promises that he comes in peace and will be able to provide use­ful in­for­ma­tion and gifts.

  • At the meet­ing, Pizarro am­bushes the Em­peror, kil­ling his ret­inue with a vol­ley of gun­fire and tak­ing him hostage. The re­main­der of the Em­peror’s forces in the area back away, prob­a­bly con­fused and scared by the novel weapons and hes­i­tant to keep fight­ing for fear of risk­ing the Em­peror’s life.

  • Over the next months, Pizarro is able to lev­er­age his con­trol over the Em­peror to stay al­ive and or­der the In­cans around; even­tu­ally he mur­ders the Em­peror and makes an al­li­ance with lo­cal forces (some of the Inca gen­er­als) to take over the cap­i­tal city of Cuzco.

  • The Span­ish con­tinue to rule via pup­pets, pri­mar­ily Manco Inca, who is their pup­pet ruler while they crush var­i­ous re­bel­lions and con­soli­date their con­trol over the em­pire. Manco Inca es­capes and launches a re­bel­lion of his own, which is partly suc­cess­ful: He ut­terly wipes out four columns of Span­ish re­in­force­ments, but is un­able to re­take the cap­i­tal. With the morale and loy­alty of his fol­low­ers dwindling, Manco Inca even­tu­ally gives up and re­treats, leav­ing the Span­ish still in con­trol.

  • Then the Span­ish ended up fight­ing each other for a while, while also putting down more lo­cal re­bel­lions. After a few decades Span­ish dom­i­nance of the re­gion is com­plete. (1572).

Afonso [wiki] [wiki] [wiki]

  • 1506: Afonso helps the Por­tuguese king come up with a shock­ingly am­bi­tious plan. Eight years prior, the first Euro­peans had rounded the coast of Africa and made it to the In­dian Ocean. The In­dian Ocean con­tained most of the world’s trade at the time, since it linked up the world’s biggest and wealthiest re­gions. See this map of world pop­u­la­tion (times­tamp 3:45). Re­mem­ber, this is prior to the In­dus­trial and Scien­tific Revolu­tions; Europe is just com­ing out of the Mid­dle Ages and does not have an ob­vi­ous tech­nolog­i­cal ad­van­tage over In­dia or China or the Mid­dle East, and has an ob­vi­ous eco­nomic dis­ad­van­tage. And Por­tu­gal is a just tiny state on the edge of the Ibe­rian pen­in­sula.

  • The plan is: Not only will we go into the In­dian Ocean and par­ti­ci­pate in the trad­ing there—cut­ting out all the mid­dle­men who are cur­rently in­volved in the trade be­tween that re­gion and Europe—we will con­quer strate­gic ports around the re­gion so that no one else can trade there!

  • Long story short, Afonso goes on to com­plete this plan by 1513. (!!!)

Some com­par­i­sons and con­trasts:

  • Afonso had more Euro­pean sol­diers at his dis­posal than Cortes or Pizarro, but not many more—usu­ally he had about a thou­sand or so. He did have more re­in­force­ments and sup­port from home.

  • Like them, he was usu­ally sig­nifi­cantly out­num­bered in bat­tles. Like them, the em­pires he warred against were vastly wealthier and more pop­u­lous than his forces.

  • Like them, Afonso was of­ten able to ex­ploit lo­cal con­flicts to gain lo­cal al­lies, which were cru­cial to his suc­cess.

  • Un­like them, his goal wasn’t to con­quer the em­pires en­tirely, just to get and hold strate­gic ports.

  • Un­like them, he was fight­ing em­pires that were tech­nolog­i­cally ad­vanced; for ex­am­ple, in sev­eral bat­tles his en­e­mies had more can­nons and gun­pow­der than he did.

  • That said, it does seem that Por­tuguese tech­nol­ogy was qual­i­ta­tively bet­ter in some re­spects (ships, ar­mor, and can­nons, I’d say.) Not dra­mat­i­cally bet­ter, though.

  • While Afonso’s was a naval cam­paign, he did fight many land bat­tles, usu­ally marine as­saults on port cities, or defenses of said cities against coun­ter­at­tacks. So su­pe­rior Euro­pean naval tech­nol­ogy is not by it­self enough to ex­plain his vic­tory, though it cer­tainly was im­por­tant.

  • Plague and civil war were not in­volved in Afonso’s suc­cess.

What ex­plains these dev­as­tat­ing con­quests?

Wrong an­swer: I cherry-picked my case stud­ies.

His­tory is full of in­cred­ibly suc­cess­ful con­querors: Alexan­der the Great, Ghenghis Khan, etc. Per­haps some peo­ple are just re­ally good at it, or re­ally lucky, or both.

How­ever: Three in­cred­ibly suc­cess­ful con­querors from the same tiny re­gion and time pe­riod, con­quer­ing three sep­a­rate em­pires? Fol­lowed up by dozens of less suc­cess­ful but still very suc­cess­ful con­querors from the same re­gion and time pe­riod? Surely this is not a co­in­ci­dence. More­over, it’s not like the con­quis­ta­dors had many failed at­tempts and a few suc­cesses. The Aztec and Inca em­pires were the two biggest em­pires in the Amer­i­cas, and there weren’t any other In­dian Oceans for the Por­tuguese to fail at con­quer­ing.

Fun fact: I had not heard of Afonso be­fore I started writ­ing this post this morn­ing. Fol­low­ing the Rule of Three, I needed a third ex­am­ple and I pre­dicted on the ba­sis of Cortes and Pizarro that there would be other, similar sto­ries hap­pen­ing in the world at around that time. That’s how I found Afonso.

Right an­swer: Technology

How­ever, I don’t think this is the whole ex­pla­na­tion. The tech­nolog­i­cal ad­van­tage of the con­quis­ta­dors was not over­whelming.

What­ever tech­nolog­i­cal ad­van­tage the con­quis­ta­dors had over the ex­ist­ing em­pires, it was the sort of tech­nolog­i­cal ad­van­tage that one could ac­quire be­fore the Scien­tific and In­dus­trial rev­olu­tions. Tech­nol­ogy didn’t change very fast back then, yet Por­tu­gal man­aged to get a lead over the Ot­tomans, Egyp­ti­ans, Mughals, etc. that was suffi­cient to bring them vic­tory. On pa­per, the Aztecs and Span­ish were pretty similar: Both were me­dieval, feu­dal civ­i­liza­tions. I don’t know for sure, but I’d bet there were at least a few tech­niques and tech­nolo­gies the Aztecs had that the Span­ish didn’t. And of course the tech­nolog­i­cal similar­i­ties be­tween the Por­tuguese and their en­e­mies were much stronger; the Ot­tomans even had ac­cess to Euro­pean mer­ce­nar­ies! Even in cases in which the con­quis­ta­dors had tech­nol­ogy that was com­pletely novel—like steel ar­mor, horses, and gun­pow­der were to the Aztecs and In­cas—it wasn’t god-like. The ar­mored sol­diers were still kil­lable; the gun­pow­der was more effec­tive than ar­rows but limited in sup­ply, etc.

(Con­trary to pop­u­lar leg­end, nei­ther Cortés nor Pizarro were re­garded as gods by the peo­ple they con­quered. The In­cas con­cluded pretty early on that the Span­ish were mere men, and while the idea did float around the Aztecs for a bit the mod­ern his­tor­i­cal con­sen­sus is that most of them didn’t take it se­ri­ously.)

Ask your­self: Sup­pose Cortés had found 500 lo­cal war­riors, gave them all his equip­ment, trained them to use it ex­pertly, and left. Would those lo­cal men have taken over all of Mex­ico? I doubt it. And this is de­spite the fact that they would have had much bet­ter lo­cal knowl­edge than Cortés did! Same goes for Pizarro and Afonso. Per­haps if he had found 500 lo­cal war­riors led by an ex­cep­tional com­man­der it would work. But the ex­pla­na­tion for the con­quis­ta­dor’s suc­cess can’t just be that they were all ex­cep­tional com­man­ders; that would be posit­ing too much in­nate tal­ent to oc­cur in one small re­gion of the globe at one time.

Right an­swer: Strate­gic and diplo­matic cunning

This is my non-ex­pert guess about the miss­ing fac­tor that joins with tech­nol­ogy to ex­plain this pat­tern of con­quis­ta­dor suc­cess.

They didn’t just have tech­nol­ogy; they had effec­tive strat­egy and they had effec­tive diplo­macy. They made long-term plans that worked de­spite be­ing breath­tak­ingly am­bi­tious. (And their short-term plans were usu­ally pretty effec­tive too, read the sto­ries in de­tail to see this.) De­spite not know­ing the lo­cal cul­ture or his­tory, these con­quis­ta­dors made sur­pris­ingly savvy diplo­matic de­ci­sions. They knew when they could get away with break­ing their word and when they couldn’t; they knew which out­rages the lo­cals would tol­er­ate and which they wouldn’t; they knew how to con­vince lo­cals to ally with them; they knew how to use words to es­cape mil­i­tar­ily im­pos­si­ble situ­a­tions… The lo­cals, by con­trast, of­ten badly mis­judged the con­quis­ta­dors, e.g. not think­ing Pizarro had the will (or the abil­ity?) to kid­nap the em­peror, and think­ing the em­peror would be safe as long as they played along.

This raises the ques­tion, how did they get that ad­van­tage? My an­swer: they had ex­pe­rience with this sort of thing, whereas lo­cals didn’t. Pre­sum­ably Pizarro learned from Cortés’ ex­pe­rience; his strat­egy was pretty similar. (See also: the prior con­quest of the Ca­nary Is­lands by the Span­ish). In Afonso’s case, well, the Por­tuguese had been sailing around Africa, con­quer­ing ports and build­ing forts for more than a hun­dred years.

Les­sons I think we learn

I think we learn that:

It is oc­ca­sion­ally pos­si­ble for a small force to quickly con­quer large parts of the world, de­spite:

  1. Hav­ing only a minus­cule frac­tion of the world’s re­sources and power

  2. Hav­ing tech­nol­ogy + diplo­matic and strate­gic cun­ning that is bet­ter but not that much better

  3. Hav­ing very lit­tle data about the world when the con­quest begins

  4. Be­ing disunited

Which all sug­gests that it isn’t as im­plau­si­ble that a small AI takes over the world in mildly fa­vor­able cir­cum­stances as is some­times thought.

EDIT: In light of good push­back from peo­ple (e.g. Lucy.ea8 and e.g. Matthew Bar­nett) about the im­por­tance of dis­ease, I think one should prob­a­bly add a caveat to the above: “In times of chaos & dis­rup­tion, at least.”

Hav­ing only a minus­cule frac­tion of the world’s re­sources and power

In all three ex­am­ples, the con­quest was more or less com­pleted with­out sup­port from home; while Spain/​Por­tu­gal did send re­in­force­ments, it wasn’t even close to the en­tire na­tion of Spain/​Por­tu­gal fight­ing the war. So these con­quests are ex­am­ples of non-state en­tities con­quer­ing states, so to speak. (That said, their claim to rep­re­sent a large state may have been cru­cial for Cortes and Pizarro get­ting au­di­ences and re­spect ini­tially.) Cortés landed with about a thou­sandth the troops of Tenochti­t­lan, which con­trol­led a still larger em­pire of vas­sal states. Of course, his troops were bet­ter equipped, but on the other hand they were also cut off from re­sup­ply, whereas the Aztecs were in their home ter­ri­tory, able to draw on a large civilian pop­u­la­tion for new re­cruits and re­sup­ply.

The con­quests suc­ceeded in large part due to diplo­macy. This has im­pli­ca­tions for AI takeover sce­nar­ios; rather than imag­in­ing a con­flict of hu­mans vs. robots, we could imag­ine hu­mans vs. hu­mans-with-AI-ad­visers, with the lat­ter fac­tion win­ning and some­how by the end of the con­flict the AI ad­visers have man­aged to be­come de facto rulers, us­ing the hu­mans who obey them to put down re­bel­lions by the hu­mans who don’t.

Hav­ing tech­nol­ogy + diplo­matic and strate­gic skill that is bet­ter but not that much better

As pre­vi­ously men­tioned, the con­quis­ta­dors didn’t en­joy god-like tech­nolog­i­cal su­pe­ri­or­ity. In the case of Afonso the tech­nol­ogy was pretty similar. Tech­nol­ogy played an im­por­tant role in their suc­cess, but it wasn’t enough on its own. Mean­while, the con­quis­ta­dors may have had more diplo­matic and strate­gic cun­ning (or ex­pe­rience) than the en­e­mies they con­quered. But not that much more—they are only hu­man, af­ter all. And their en­e­mies were pretty smart.

In the AI con­text, we don’t need to imag­ine god-like tech­nol­ogy (e.g. swarms of self-repli­cat­ing nanobots) to get an AI takeover. It might even be pos­si­ble with­out any new phys­i­cal tech­nolo­gies at all! Just su­pe­rior soft­ware, e.g. pi­lot­ing soft­ware for mil­i­tary drones, tar­get­ing soft­ware for anti-mis­sile defenses, cy­ber­war­fare ca­pa­bil­ities, data anal­y­sis for mil­i­tary in­tel­li­gence, and of course ex­cel­lent pro­pa­ganda and per­sua­sion.

Nor do we need to imag­ine an AI so savvy and per­sua­sive that it can per­suade any­one of any­thing. We just need to imag­ine it about as cun­ning and ex­pe­rienced rel­a­tive to its en­e­mies as Cortés, Pizarro, and Afonso were rel­a­tive to theirs. (Pre­sum­ably no AI would be ex­pe­rienced with world takeover, but per­haps an in­tel­li­gence ad­van­tage would give it the same benefits as an ex­pe­rience ad­van­tage.) And if I’m wrong about this ex­pla­na­tion for the con­quis­ta­dor’s suc­cess—if they had no such ad­van­tage in cun­ning/​ex­pe­rience—then the con­clu­sion is even stronger.

Ad­di­tion­ally, in a rapidly-chang­ing world that is un­der­go­ing slow take­off, where there are lesser AIs and AI-cre­ated tech­nolo­gies all over the place, most of which are suc­cess­fully con­trol­led by hu­mans, AI takeover might still hap­pen if one AI is bet­ter, but not that much bet­ter, than the oth­ers.

Hav­ing very lit­tle data about the world when the con­quest begins

Cortés in­vaded Mex­ico know­ing very lit­tle about it. After all, the Span­ish had only re­al­ized the Amer­i­cas ex­isted two decades prior. He heard ru­mors of a big wealthy em­pire and he set out to con­quer it, know­ing lit­tle of the tech­nol­ogy and tac­tics he would face. Two years later, he ruled the place.

Pizarro and Afonzo were in bet­ter epistemic po­si­tions, but still, they had to learn a lot of im­por­tant de­tails (like what the lo­cal power cen­ters, norms, and con­flicts were, and ex­actly what tech­nol­ogy the lo­cals had) on the fly. But they were good at learn­ing these things and mak­ing it up as they went along, ap­par­ently.

We can ex­pect su­per­hu­man AI to be good at learn­ing. Even if it starts off know­ing very lit­tle about the world—say, it figured out it was in a train­ing en­vi­ron­ment and hacked its way out, hav­ing in­ferred a few gen­eral facts about its cre­ators but not much else—if it is good at learn­ing and rea­son­ing, it might still be pretty dan­ger­ous.

Be­ing disunited

Cortés in­vaded Mex­ico in defi­ance of his su­pe­ri­ors and had to defeat the army they sent to ar­rest him. Pizarro ended up fight­ing a civil war against his fel­low con­quis­ta­dors in the mid­dle of his con­quest of Peru. Afonzo fought Greek mer­ce­nar­ies and some traitor Por­tuguese, con­quered Malacca against the or­ders of a ri­val con­quis­ta­dor in the area, and was ul­ti­mately de­moted due to poli­ti­cal ma­neu­vers by ri­vals back home.

This as­ton­ishes me. Some­how these con­quests were com­pleted by peo­ple who were at the same time busy in­fight­ing and back­stab­bing each other!

Why was it that the con­quis­ta­dors were able to split the lo­cals into fac­tions, ally with some to defeat the oth­ers, and end up on top? Why didn’t it hap­pen the other way around: some am­bi­tious lo­cal ruler talks to the con­quis­ta­dors, ex­ploits their in­ter­nal di­vi­sions, al­lies with some to defeat the oth­ers, and ends up on top?

I think the an­swer is partly the “diplo­matic and strate­gic cun­ning” men­tioned ear­lier, but mostly other things. (The con­quis­ta­dors were di­s­united, but pre­sum­ably were united in the ways that mat­tered.) At any rate, I ex­pect AIs to be pretty good at co­or­di­nat­ing too; they should be able to con­quer the world just fine even while com­pet­ing fiercely with each other. For more on this idea, see this com­ment.

By Daniel Kokotajlo


Thanks to Katja Grace for feed­back on a draft. All mis­takes are my own, and should be pointed out in the com­ments. Edit: Also, when I wrote this post I had for­got­ten that the ba­sic idea for it prob­a­bly came from this com­ment by JoshuaFox.