SQL, or the language that outlived one of its authors
At a 1973 IBM research seminar, Donald Chamberlin sat through Edgar Codd’s presentation on his relational calculus — the formal notation Codd had designed for querying his new kind of database — and understood immediately that it would never leave the building. The symbols were precise. The syntax was borrowed from mathematical logic. And no one who wasn’t already a mathematician was going to write it. Chamberlin left the seminar with a clear task: same theory, different language.
The theory itself had arrived three years earlier in the June 1970 issue of Communications of the ACM. Codd, an IBM mathematician working at the company’s San Jose Research Laboratory, had proposed organizing databases not in hierarchical trees but in flat tables linked by shared columns — the relational model (Wikipedia). A query would describe the desired result, not the retrieval procedure. The paper was thorough enough to win Codd the Turing Award in 1981. IBM’s initial response was more cautious: the company worried the new model would undercut its existing database products. In 1973, with competitors circling, IBM launched System R, a prototype to test whether relational databases could be made fast enough to matter.
Chamberlin and his colleague Raymond Boyce were assigned the query language. Boyce, who had recently joined IBM’s research staff, shared Chamberlin’s diagnosis: the gap to close was not technical but ergonomic. They designed SEQUEL — Structured English Query Language — with the explicit goal of letting non-mathematicians retrieve data without first earning a degree in predicate calculus. Their 1974 paper, presented at an ACM symposium in Ann Arbor, Michigan, introduced a grammar that looked more like English than algebra (Wikipedia; LearnSQL.com). The key operation, SELECT, asked for specific columns; WHERE filtered them; FROM named the table. A query read like a question.
Boyce died of a ruptured brain aneurysm in 1974, shortly after the paper was presented. He was twenty-six. He had no idea whether the thing would work.
SEQUEL did not survive as a name. “SEQUEL” was already a registered trademark of Hawker Siddeley Dynamics Engineering, a British aerospace company. IBM dropped the vowels and the language became SQL — officially Structured Query Language — though the pronunciation debate (“sequel” versus three separate letters) started immediately and never quite resolved (LearnSQL.com).
By 1979, Relational Software Inc. — shortly to become Oracle Corporation — had shipped the first commercial SQL database, beating IBM’s own System R derivative to market by two years. IBM’s SQL/DS followed in 1981. ANSI made SQL an official standard in 1986; ISO adopted it in 1987. Every major database system built since — Oracle, PostgreSQL, MySQL, Microsoft SQL Server, SQLite — speaks some version of the grammar Chamberlin and Boyce sketched in San Jose in 1973.
Boyce’s name survived in a different form: the Boyce-Codd normal form, a foundational concept in database theory, bears his name beside Codd’s. A theorem is a more durable monument than a version number.
Codd had answered the structural question: how should data be organized? Chamberlin and Boyce answered the human one: how should a person ask for it? The answer — in plain sentences rather than symbols — turned out to matter as much as the theory it was built to express.
Sources
- SQL — Wikipedia — Codd’s 1970 relational model paper, SEQUEL development at IBM San Jose, Hawker Siddeley trademark history, ANSI 1986 standardization, commercial timeline.
- History of SQL — LearnSQL.com — System R project, Chamberlin and Boyce’s roles, Ray Boyce’s death, Oracle’s 1979 commercial release, IBM’s 1981 SQL/DS.