What the World Bans, and Why: Patterns in 35,000 Book-Censorship Records

Once book-censorship records are structured—who banned what, where, when, and on what stated grounds—you can stop asking whether a particular title is banned and start asking the harder question: what does the world ban, and why? The Verboten index holds 19,283 banned or restricted titles across 34,987 dated, source-cited events in 119 countries. Read in aggregate, it tells a story that no single banned-book headline does.

Politics, not morality, leads the world

Ranked by the number of distinct titles affected, the single most common stated reason for a book ban worldwide is political content—cited against 9,813 titles. Morality-based reasons follow but trail badly: immorality (4,797), sexual content (3,119), LGBTQ+ content (2,573), violence (2,089), race and colonialism (1,579), religious or blasphemous content (1,496). The popular image of censorship as mostly about sex and profanity is, globally and historically, wrong: more books are suppressed for what they say about power than for anything else.

LGBTQ+ bans are an American story

The sharpest concentration in the entire dataset is the LGBTQ+ category. Of the 2,573 titles banned or restricted on LGBTQ+ grounds, 2,431—about 95 percent—are in the United States, with no other country breaking 65. This is not a global pattern that happens to include the US; it is a national phenomenon that barely exists elsewhere in the record. Where the rest of the world reaches for “political” and “religious,” the recent American wave reaches for “LGBTQ+” and “sexual.”

The 2020s are off the chart

Bin the bans by the decade they took effect and one bar dwarfs the rest. The 2010s saw roughly 685 distinct titles newly banned or restricted; the 2020s, not yet finished, already hold 9,411—more than an order of magnitude higher, and more than every prior decade in the dataset combined. The surge is overwhelmingly the US school-board wave: restriction events scoped to individual school districts, filed under immorality, sexual content, and LGBTQ+ themes. It is the largest single feature in four centuries of records.

Two different machines

The data separates cleanly into two censorship regimes that work nothing alike. The first is the national outright ban: a government prohibits a title across the whole country, usually on political or religious grounds. Of 34,987 events, 23,924 are outright bans and 11,734 are scoped to a national government. This is the mode that put The Satanic Verses on prohibition lists in 22 countries, the most-banned title in the world, and that drives the records of Malaysia and Iran.

The second is the local restriction: not a national ban but a removal or constraint at the level of a school district or library, often reversible, and counted carefully apart from outright bans (10,854 restriction events, 23,089 scoped to schools). This is almost entirely the modern American mode. The United States accounts for 8,009 of the dataset's distinct banned titles—more than twice the next country—but its record is dominated by district-level restrictions, not national prohibitions. Two countries can both rank high and mean completely different things by it; the action and scope fields are what keep the distinction honest.

Reproduce it

Every figure above comes from Verboten's static JSON API—no key, no rate limit. The reasons ranking and the LGBTQ+ concentration are two short calls:

import requests

BASE = "https://ai-analytics.org/verboten"

# The stated reasons, ranked by distinct banned/restricted titles.
m = requests.get(f"{BASE}/api/index.json", timeout=30).json()
for r in m["reasons"]:
    print(f'{r["books"]:>6}  {r["label"]}')

# Which countries drive a given reason? LGBTQ+ is the sharp case.
lgbtq = requests.get(f"{BASE}/api/reason/lgbtq.json", timeout=30).json()
print(lgbtq["bookCount"], "titles across", lgbtq["countryCount"], "countries")
for c in lgbtq["topCountries"][:5]:
    print(f'{c["bookCount"]:>6}  {c["name"]}')

What the numbers cannot say

A censorship index measures documentation as much as censorship. The United States looms large partly because its school-board fights are exhaustively recorded by library associations and the press; the most opaque censors are precisely the ones least likely to leave a citable record, so a low count for a closed country is a statement about transparency, not freedom. All counts here rank distinct titles and distinct countries rather than raw events, so a title challenged in a hundred districts is one banned title, not a hundred. Read with those limits, the shape holds: politics leads the world, the American wave is historically singular, and the 2020s are unlike any decade before them. The full record—by country, title, and reason—is open under CC BY 4.0.