Martin Odersky: The Man Who Built a Rival to His Own Compiler
Every Java programmer who has ever typed `javac` has run code whose reference implementation Martin Odersky originally wrote. He also co-designed Java's generics, the feature that let the language finally express typed collections. Having done more than almost anyone to make Java respectable, he concluded that the deeper problem could not be patched from inside — and, from a professorship in Lausanne, spent the next twenty years building the language he thought should have existed instead.
Zurich to Lausanne, by way of Adelaide
Odersky took his doctorate at ETH Zurich in 1989, and his early career reads as a tour of the places where programming-language research was actually being done. He went straight to IBM's T. J. Watson Research Center in 1989 and to Yale in 1991, took a professorship at the University of Karlsruhe in 1993 and another at the University of South Australia in 1997, and in 1999 arrived at EPFL as full professor, where he has been ever since. He heads the Programming Methods Laboratory in the School of Computer and Communication Sciences.
His stated research interests are unglamorous and exact: semantics, type systems, programming language design, compiler construction. These are the four things you need if you intend to build a language that works rather than one that reads well in a paper, and it is worth noting that he has spent his career holding all four at once.
Rewriting Java from the Inside
The 1990s were Java's decade, and Odersky spent them making the language better while becoming steadily more convinced of its limits. He built Pizza, an extension that added functional programming capabilities to an aggressively imperative language. He co-designed Java's generics — a genuinely brutal engineering problem, since parametric polymorphism had to be retrofitted into a language already deployed on millions of machines without breaking any of the code already written against it. And he wrote the original javac, the reference compiler.
That is an unusual position to occupy. He was not an outsider throwing rocks at Java; he was one of the people responsible for it, working on the inside, and he came away thinking the inside was not where the answer lay. "For a long time I've been between the two worlds," he has said, "and thinking that there should be a way to combine the best elements of both structured programming and functional programming."
The Blank Page
EPFL gave him the freedom to stop patching. "I decided that we could start from scratch," he recalled of his arrival in 1999, "and ask, with a blank page, 'how would we put together elements of structured, or object-oriented, programming, and functional programming?'"
The result was Scala, first released publicly in 2004. The name is a compression of "scalable language" — meaning, in his account, "a programming language where you can start small with just a few lines of code and grow bigger." It merges object-oriented and functional programming in a single type system while remaining interoperable with Java, which is the design constraint that made it usable and also made it hard: the new language had to run on the Java Virtual Machine and speak to existing Java code, so every elegant idea had to survive contact with an installed base.
He did not expect what happened next. "At the beginning, we didn't expect anything," he said, "but five years later, suddenly a lot of people in San Francisco, in Silicon Valley and elsewhere jumped on it."
What Industry Did With It
Scala became one of the very few languages to escape the academy. Twitter and LinkedIn were early adopters at serious scale; today thousands of companies use it, among them Duolingo, Volkswagen and X. It also stayed a research vehicle: the SIGPLAN award committee noted that Scala "has been a fertile research ground for metaprogramming, macros, staging, and embedded domain-specific languages," including work on machine learning and GPU execution.
The governance reflects the double life. Scala is co-owned by EPFL and Lightbend, the commercial company built around it, and refined by contributors worldwide. Odersky is Academic Director of the Scala Center, the open-source foundation based at EPFL that stewards the language.
In 2019 Scala received the SIGPLAN Programming Languages Software Award, given for "the development of a software system that has had a significant impact on programming language research." The committee's summary is the most precise praise on the record: "Scala is one of the few programming languages from academia that has had a significant impact on the world." Odersky is a Fellow of the ACM, and in 2025 received the Programming Languages Achievement Award for what was described as a profound and lasting impact on the field. Alongside all of it he has supervised more than forty doctoral students at EPFL, served as associate editor of the Journal of Functional Programming, sat on IFIP Working Group 2.8, chaired ICFP in 2000 and run the programmes for ECOOP 2004 and ETAPS/CC 2007.
Why Martin Is Called a Genius
The specific intellectual difficulty here is not writing a compiler, which many capable people can do. It is reconciling two type disciplines that make contradictory assumptions. Object-oriented programming is built on mutable state, identity and subtyping; functional programming is built on immutable values and parametric polymorphism. Making both sound inside a single static type system — so that the compiler can actually prove things about programs that mix them — is a hard theoretical problem, and Odersky solved it twice under different constraints. The first time was inside Java, where generics had to be added to a live, deployed language without invalidating existing code. The second was on a blank page, where the constraint was the opposite: total freedom of design, but the finished language still had to run on someone else's virtual machine and interoperate with someone else's libraries. Very few people can hold formal type theory and industrial compatibility in mind simultaneously. His whole career is that combination.
The counter-case deserves stating plainly. Nobody, including his award citations, uses the word genius; the committees say "significant impact" and "profound and lasting impact," which is a different and more defensible claim. He did not invent either of the paradigms he fused, and the fusion itself had precedents. Java generics were co-designed, not designed alone. Scala is co-owned with a company and shaped by contributors across the world, so treating it as one mind's artifact is a simplification he has never made himself — and by his own account its industrial success arrived unexpectedly, five years after release, driven by adopters he did not court. A great deal of that success rests on a platform decision, the JVM, which gave Scala an installed base no new language could otherwise buy. What he unambiguously has is design judgment of a very high order sustained over decades, which is rarer than a single brilliant idea and much harder to fake.
Legacy
Scala's influence runs wider than its own user base, because the pattern it proved — functional constructs living comfortably inside a mainstream object-oriented language on a mainstream runtime — became the default expectation for languages that followed. Odersky's career is also an argument about where good languages come from: not from committees and not purely from industry, but from a university lab willing to spend twenty years on one idea while remaining answerable to people shipping production code. That EPFL still hosts the Scala Center, decades on, is the point rather than a footnote.
Achievements
- Affiliated with Swiss Federal Institute of Technology in Lausanne
- Educated at ETH Zurich
- Worked as computer scientist, engineer and professor
