Mexico's national ideology of mestizaje — the claim that racial mixing has produced a single, unified Mexican people — is a lie. It functions as ideological cover for a pigmentocracy: a racial caste system in which lighter-skinned, European-phenotype individuals dominate political and economic power, while darker-skinned, indigenous-phenotype individuals are systematically excluded.
This study tests that claim computationally. We collect public photographs of three populations:
We generate facial composites (averaged faces) for each group and compare them against standardized anthropological phenotype references using multi-metric similarity analysis.
"In Mexico, the lighter your skin, the more likely you are to hold a professional position, earn a higher income, and attain higher education. The darker your skin, the more likely you are to live in poverty."
— PERLA Project (Telles & Steele, 2012)
| Group | Total | Male (used) | Source |
|---|---|---|---|
| Iberian Politicians | 81 | 64 | Wikipedia/Wikimedia Commons |
| Mexican Governors/Presidents | 98 | 83 | Wikipedia/Wikimedia Commons |
| Cartel Leaders | 56 | 53 | Wikipedia/Wikimedia Commons |
All photographs are publicly available. Only male subjects are used for composite generation to ensure comparability against male phenotype references. Female subjects are included in collages but excluded from composites.
12 standardized male phenotype composites from humanphenotypes.net, split into two categories:
| European Types | Indigenous American Types | ||
|---|---|---|---|
| Alpinid | Central European | Californid | NW Mexico/Baja California |
| Armenoid | Near Eastern/Caucasus | Arizonid | SW United States/N Mexico |
| Dinarid | Southeastern European | Mexicid | Central Mexico |
| Mediterranid | Southern European | Pueblid | Pueblo peoples |
| Maya | Yucatan/Guatemala | ||
| Sonorid | Sonoran Desert | ||
| Silvid | South American | ||
| Chocomotilon | Colombian/Venezuelan | ||
Combined = 0.45 × ColorHist + 0.30 × EdgeSSIM + 0.25 × SSIMThe central finding. Three composites, left to right: colonial source → political elite → cartel leaders. The gradient is visible to the naked eye.
| Group | #1 Match | Score | Type |
|---|---|---|---|
| Iberian | Alpinid | 0.622 | European |
| Mexican Governor | Alpinid | 0.552 | European |
| Cartel Leader | Californid | 0.514 | Indigenous American |
The Iberian and Governor composites both match European phenotypes (Alpinid). The Cartel composite matches an Indigenous American phenotype (Californid — the type native to northwestern Mexico, where most major cartels are geographically based).
| Group | Avg European Similarity | Avg Indigenous Similarity | Ratio (I/E) |
|---|---|---|---|
| Iberian | 0.574 | 0.454 | 0.792 |
| Mexican Governor | 0.501 | 0.419 | 0.837 |
| Cartel Leader | 0.438 | 0.381 | 0.870 |
The Iberian composite is the most European-leaning (lowest ratio). The Governor composite sits between Iberian and Cartel. The Cartel composite is the least European-leaning. This is the phenotype gradient: colonial source → political elite → excluded population.
In 1989, David Hackett Fischer published Albion's Seed and changed how we understand America. He showed that four distinct British folkways — Puritan, Cavalier, Quaker, and Borderer — transplanted themselves into colonial America and never left. Not for four hundred years. Not through a revolution, a civil war, industrialization, or mass immigration. The folkway persists because it is deeper than ideology, deeper than economics, deeper than language. It is the inherited pattern of how a people organize kinship, authority, honor, death, and land. It is the deepest unit of cultural analysis. Everything else — politics, economics, even religion — is downstream.
This study applies Fischer's framework to Mexico. The argument is simple:
This is part of a larger thesis on the primacy of folkways in human organization, drawing on Fischer's work in the Anglo-American context and extending it to the Ibero-American one. The argument is that folkways are the most powerful and persistent force in human political organization — more durable than states, more explanatory than economics, more honest than ideology. If you want to understand why Mexico looks the way it does, you don't start with drug policy or NAFTA or corruption indices. You start with the folkways.
The Spanish who conquered and settled Mexico were not generic "Spaniards." They were overwhelmingly Castilians — from the central meseta, the kingdom that drove the Reconquista, the culture that produced the hidalgo (hijo de algo, "son of something"). The Castilian folkway is characterized by:
Now look at the phenotype data. The Iberian composite matches Alpinid, not Mediterranid. This demands explanation.
The Alpinid type is Central European. Southern Spain — Andalusia, the coastal regions — is Mediterranid. But the Spanish political class is not from southern Spain. It is from Castile, the Basque Country, Galicia, Catalonia — the northern and central regions where centuries of Visigothic, Frankish, Burgundian, and Habsburg dynastic marriage deposited a northern European aristocratic admixture into the ruling families. The Spanish political class is phenotypically more "Alpine" than "Mediterranean" because it descends from the same Central European aristocratic stock that ruled half of Europe.
The Mexican governor composite also matches Alpinid. Five centuries. An ocean. A revolution. An independence war. A second revolution. And the criollo elite still carries the phenotypic signature of Castilian aristocratic blood, maintained through endogamy so thorough that the face of Mexican power in 2026 still looks like the face of Spanish power in 2026.
The criollos didn't just inherit Castilian culture. They inherited Castilian blood, and they kept it. The mestizaje myth says everyone mixed. The faces say the people at the top didn't.
What mestizaje erases is that pre-Columbian Mexico was not one culture. It was dozens of distinct folkways, as different from each other as Fischer's Puritans were from his Borderers:
These are not historical curiosities. These are living folkways. They did not dissolve into mestizaje. They persisted — in land use patterns, in kinship structures, in the way communities organize authority, in the territories people fight and die for.
This is the core claim. Cartel territorial boundaries are not arbitrary lines drawn by criminals. They track pre-Columbian cultural and linguistic boundaries:
| Cartel Territory | Pre-Columbian Region | Indigenous Folkway |
|---|---|---|
| Sinaloa Cartel | Aztatlan / Cahita lands | Northwestern (Mayo, Yaqui, Cahita) |
| CJNG (Jalisco) | Western Mesoamerican frontier | Caxcan / Coca / Tecuexe |
| La Familia / Knights Templar | Purepecha Empire (Irechequa Tzintzuntzan) | Purepecha (Tarascan) |
| Los Zetas / Gulf (Tamaulipas) | Huasteca / Chichimec frontier | Huastec / semi-nomadic north |
| Tijuana Cartel | Kumeyaay / Pai Pai lands | Californid (Yuman peoples) |
| Juarez Cartel | Suma / Manso / Jumano lands | Chihuahuan Desert peoples |
The Sinaloa Cartel operates in exactly the territory of the Cahita-speaking peoples. La Familia Michoacana rose in the exact territory of the Purepecha Empire — the one Mesoamerican state that never submitted to the Aztecs. The Zetas controlled the Huasteca, the ancient frontier between Mesoamerica and the nomadic north. These are not coincidences. These are folkways.
If you opened Europa Universalis IV and placed the cartel map over the 1444 start date, the borders would align. The Sinaloa Cartel is the Cahita people. La Familia is the Purepecha. The CJNG is the Caxcan confederacy. They have goofy names and they traffic fentanyl, but the underlying political geography is pre-Columbian. It always was.
The conventional framing of Mexican cartels — as criminal enterprises motivated by profit from the drug trade — is not wrong, but it is shallow. It is like saying the American Revolution was about tea taxes. The drugs are the occasion, not the cause.
At their most pure expression, cartels are ethnic organizations that want sovereignty and autonomy. They are the political expression of indigenous folkways that were never absorbed into the Castilian state — that were pushed to the margins, excluded from legitimate power, and left with no institutional path to self-governance within the Mexican republic.
The Mexican state's refusal to recognize cartels as anything other than criminal organizations is itself a continuation of the colonial project. To acknowledge the indigenous folkway roots of cartel territorial organization would be to acknowledge that Mexico is not one nation but many — that mestizaje failed — that the Castilian folkway conquered the government but never conquered the land. The war on drugs is a culture war. It always was.
Return to the faces. The Iberian composite: Alpinid. The Governor composite: Alpinid. The Cartel composite: Californid.
The Alpinid signal is the Castilian aristocratic bloodline — Visigothic and Habsburg, maintained across five centuries and an ocean through endogamy. The Californid signal is the indigenous Northwestern peoples — the Cahita, Yaqui, Mayo. The people of Sinaloa.
The gradient is not random. It is the fossilized caste system of New Spain, still visible in the faces of the people who hold power and the people who are excluded from it. The faces tell the same story as the folkways, the land tenure patterns, the CONAPRED discrimination surveys, the cartel territory map, and the pre-Columbian linguistic map. It is all one story.
A Castilian colonial elite that never left. A patchwork of indigenous nations that never surrendered. A civic myth called mestizaje stretched over the gap like a tarp over a crater. And underneath it, the same war that started in 1519, fought by the same peoples, over the same land, with different names.
Mestizaje is the name the winners gave to their victory. The cartels are what the losers built when they stopped believing in it.
If faces reveal what the colonial system tried to hide, surnames reveal what the colonial system imposed. Both groups carry overwhelmingly Spanish surnames — but the patterns diverge dramatically.
| Metric | Cartel Leaders | Governors/Presidents |
|---|---|---|
| Spanish/Castilian surnames | 100.0% | 81.0% |
| Non-Spanish European surnames | 0.0% | 18.3% |
| Indigenous surnames | 0.0% | 0.7% (1 person) |
| People with non-Spanish Euro surnames | 0 / 53 | 28 / 97 (28.9%) |
Nearly one in three governors/presidents carries a non-Spanish European surname, revealing the cosmopolitan elite immigration streams that built Mexico's ruling class:
| Origin | Names |
|---|---|
| French | Ebrard, Casaubón, Massieu, Servién |
| English/Irish | Fox, Creel, Bartlett, Yarrington |
| Lebanese | Kuri, Fayad, Harfuch, Nahle |
| Ashkenazi Jewish | Sheinbaum |
| German/Dutch | Müller, Coppel |
| Basque aristocratic | de Gortari, de Ochoa |
| Castilian aristocratic | Ponce de León, de la Madrid, de Lamadrid |
| Catalan | Vila, Rovirosa, Sansores |
| Scandinavian | Borge |
| Chinese | Chong |
Among cartel leaders: zero. Every single surname — Guzmán, Zambada, Carrillo, Arellano, Cárdenas, Oseguera, Beltrán — is a Spanish colonial patronymic. No French. No English. No Lebanese. No German. No cosmopolitan elite immigration. The ruling class absorbed immigrants from every corner of Europe and the Levant. The cartel world did not.
The absence of indigenous surnames among cartel leaders does not mean they are Spanish. It means the colonial sistema de castas worked exactly as intended: indigenous peoples were baptized with Spanish names, erasing their linguistic identity. The only indigenous surname in the entire dataset is May (Maya origin) — belonging to a governor, not a cartel leader.
Three Nahuatl first names appear among politicians — Cuauhtémoc Blanco, Cuitláhuac García, Xóchitl Gálvez — but these are modern nationalist gestures, not inherited indigenous identity. The surnames remain Spanish.
This is the paradox: the names cannot distinguish indigenous from Spanish because the colonial system made sure of it. The faces tell the truth that the names cannot.
| Phenotype | Combined | Color | Edge | SSIM |
|---|---|---|---|---|
| Alpinid | 0.6223 | 0.8586 | 0.5215 | 0.3179 |
| Californid | 0.5793 | 0.7736 | 0.5096 | 0.3132 |
| Mediterranid | 0.5747 | 0.7263 | 0.5469 | 0.3351 |
| Armenoid | 0.5531 | 0.6925 | 0.5108 | 0.3530 |
| Dinarid | 0.5440 | 0.8010 | 0.3703 | 0.2898 |
| Arizonid | 0.4973 | 0.6171 | 0.4593 | 0.3271 |
| Chocomotilon | 0.4967 | 0.6965 | 0.3688 | 0.2906 |
| Pueblid | 0.4820 | 0.5871 | 0.4851 | 0.2892 |
| Maya | 0.4299 | 0.5442 | 0.3822 | 0.2814 |
| Mexicid | 0.4232 | 0.4628 | 0.4707 | 0.2950 |
| Silvid | 0.3903 | 0.3690 | 0.4945 | 0.3035 |
| Sonorid | 0.3335 | 0.3257 | 0.3698 | 0.3039 |
| Phenotype | Combined | Color | Edge | SSIM |
|---|---|---|---|---|
| Alpinid | 0.5521 | 0.6969 | 0.5289 | 0.3194 |
| Californid | 0.5268 | 0.6489 | 0.5236 | 0.3110 |
| Mediterranid | 0.4981 | 0.5694 | 0.5337 | 0.3269 |
| Arizonid | 0.4878 | 0.6013 | 0.4588 | 0.3181 |
| Armenoid | 0.4825 | 0.5427 | 0.5135 | 0.3367 |
| Dinarid | 0.4697 | 0.6382 | 0.3739 | 0.2815 |
| Pueblid | 0.4557 | 0.5256 | 0.4851 | 0.2947 |
| Chocomotilon | 0.4307 | 0.5452 | 0.3738 | 0.2931 |
| Mexicid | 0.3932 | 0.3927 | 0.4783 | 0.2921 |
| Maya | 0.3831 | 0.4350 | 0.3931 | 0.2777 |
| Silvid | 0.3735 | 0.3267 | 0.5019 | 0.3035 |
| Sonorid | 0.3012 | 0.2419 | 0.3867 | 0.3053 |
| Phenotype | Combined | Color | Edge | SSIM |
|---|---|---|---|---|
| Californid | 0.5135 | 0.6392 | 0.4652 | 0.3453 |
| Arizonid | 0.4853 | 0.6316 | 0.3989 | 0.3257 |
| Mediterranid | 0.4581 | 0.5009 | 0.4770 | 0.3585 |
| Alpinid | 0.4558 | 0.4992 | 0.4674 | 0.3639 |
| Dinarid | 0.4188 | 0.5175 | 0.3465 | 0.3276 |
| Armenoid | 0.4197 | 0.4086 | 0.4543 | 0.3981 |
| Pueblid | 0.4035 | 0.4410 | 0.4224 | 0.3133 |
| Silvid | 0.3635 | 0.3360 | 0.4501 | 0.3093 |
| Chocomotilon | 0.3539 | 0.3940 | 0.3246 | 0.3167 |
| Mexicid | 0.3366 | 0.2983 | 0.4120 | 0.3153 |
| Maya | 0.3353 | 0.3523 | 0.3240 | 0.3182 |
| Sonorid | 0.2559 | 0.1711 | 0.3289 | 0.3210 |
This is exploratory computational anthropology, not peer-reviewed research. Known limitations:
All source photographs are from Wikipedia/Wikimedia Commons (public domain or CC-licensed). All analysis code is available on request.