← Back to Home
ON THIS DAY 2026-10-26

1983: Alfred Tarski dies in Berkeley

On 26 October 1983, the Polish-American logician Alfred Tarski died in Berkeley, California, at the age of 82. Born in Warsaw in 1901, Tarski fled to the United States just weeks before the Nazi invasion of Poland in 1939, a stroke of timing that saved his life while much of his family perished in the Holocaust. He spent the rest of his career at the University of California, Berkeley, building one of the twentieth century's most influential schools of logic.

Tarski's most consequential idea is deceptively simple: a rigorous, mathematical definition of truth for formal languages. In his 1933 paper "The Concept of Truth in Formalized Languages," he showed that a sentence can be called true within a language only by stepping outside it, into a richer metalanguage, and defining truth recursively through the notion of "satisfaction." This move founded model theory, the branch of logic that studies the relationship between formal symbols and the structures they describe.

That relationship turned out to matter enormously for computing. Tarski's algebra of relations fed directly into Edgar Codd's relational model for databases, the ancestor of nearly every SQL system storing and retrieving information today, including the knowledge bases behind modern search and recommendation engines. His semantic framework also underlies logic programming languages like Prolog, formal methods for verifying software and hardware, and the description logics that power ontologies and knowledge graphs in symbolic AI.

  • Model theory: formal semantics for logical and programming languages
  • Relational algebra: the mathematical backbone of databases
  • Knowledge representation: description logics behind the Semantic Web

Tarski never built a machine or wrote a line of code, but his insistence on distinguishing language from the language that talks about it anticipated a problem that still troubles artificial intelligence: how a system can represent, let alone verify, that one of its statements is actually true. Today's large language models generate fluent sentences without any Tarskian notion of satisfaction behind them, which is precisely why researchers keep returning to symbolic techniques, knowledge graphs, and formal verification to ground AI's claims in something more rigorous than plausibility.