In the history of computing, certain tools transcend their original purpose and become the infrastructure upon which entire disciplines are built. The R programming language is one such tool. Born not in a Silicon Valley startup or a corporate research lab, but in the Statistics Department of the University of Auckland, New Zealand, R has grown from a modest teaching project into the dominant language of statistical computing, used by millions of researchers, data scientists, and analysts worldwide. One of its two creators, Ross Ihaka, is not only a distinguished computer scientist and statistician — he is also one of the most prominent scientists of Māori descent, a fact that adds a remarkable dimension to a story already rich with intellectual achievement. Together with Robert Gentleman, Ihaka built a language that democratized access to advanced statistical methods and, in doing so, reshaped how the world analyzes data.
Early Life and Academic Roots
Ross Robert Ihaka was born in Auckland, New Zealand. Of Māori (Ngāti Kahungunu) and European descent, he grew up in a country where the intersection of indigenous heritage and academic science was uncommon — making his later career all the more significant as a trailblazer for Māori representation in the sciences and technology. From an early age, Ihaka was drawn to mathematics and the kind of structured thinking that would eventually lead him to statistics and computation.
Ihaka pursued his education at the University of Auckland, where he would spend the vast majority of his professional life. He earned his undergraduate and graduate degrees in statistics, eventually joining the faculty of the Department of Statistics as a lecturer and later rising to the rank of Associate Professor. His academic home was not a glamorous computer science department at MIT or Stanford, but a statistics department in the Southern Hemisphere — a setting that would prove to be the ideal crucible for a language designed not by software engineers for software engineers, but by statisticians for statisticians.
His early academic work focused on computational methods for statistical graphics and data analysis. Even before R existed, Ihaka was deeply interested in the question of how computers could be used not merely to crunch numbers but to help researchers see patterns, explore distributions, and communicate findings visually. This interest in the visual and interactive dimensions of data analysis would profoundly shape the language he would later create. The tradition of statistical computing that Ihaka inherited owed much to the pioneers of scientific programming — from the numerical methods enabled by Fortran to the algorithmic foundations laid down by Edsger Dijkstra.
The Birth of R: A Language for Statistics
The story of R begins in 1993, when Ross Ihaka and his colleague Robert Gentleman — both faculty members in the University of Auckland’s Department of Statistics — decided to create a new programming language for use in their teaching. The problem they faced was both practical and pedagogical. They needed a language that could handle statistical computation, data manipulation, and graphical output, and that was accessible enough for students who were statisticians first and programmers second.
The existing options were limited. The S language, created by John Chambers and his colleagues at Bell Labs in the 1970s and 1980s, was the gold standard for statistical computing. S was elegant, expressive, and deeply integrated with statistical thinking. But S was proprietary — its commercial implementation, S-PLUS, was expensive, and academic licenses were restrictive. For a department in New Zealand trying to teach statistics to undergraduates, the cost was prohibitive. This was, it is worth remembering, the same Bell Labs environment that had produced Unix and the C language — a tradition of building powerful tools that shaped the computing landscape for decades.
Ihaka and Gentleman were also influenced by Scheme, the minimalist dialect of Lisp that emphasized functional programming, lexical scoping, and clean semantics. Where S provided the statistical domain model — data frames, formulas, statistical functions — Scheme provided ideas about language design: how a language should handle functions as first-class objects, how scoping rules should work, and how an interactive read-eval-print loop (REPL) could make exploratory analysis natural and fluid.
The name “R” was chosen with deliberate wit. It was partly a reference to the first names of its creators — Ross and Robert — and partly a playful nod to the S language that inspired it. Just as S had its name, R would have its own letter, one step removed in the alphabet and in philosophy. Where S was proprietary, R would be free. Where S was controlled by a corporation, R would be controlled by its community.
The Technical Design
The initial implementation of R was, by the standards of modern language engineering, remarkably modest. Ihaka and Gentleman wrote the first interpreter in C, borrowing heavily from the semantics of S while incorporating ideas from Scheme. The core language featured dynamically typed variables, first-class functions with lexical scoping, a powerful vector-oriented computation model, and an interactive environment designed for exploratory data analysis.
One of R’s most distinctive design decisions was its treatment of vectors as the fundamental data type. In R, there are no scalars — even a single number is a vector of length one. This design, inherited from S, made it natural to express statistical operations that work on entire datasets rather than individual values. A line of R code that computes the mean of a thousand observations looks identical to one that computes the mean of ten — the language abstracts away the iteration, freeing the statistician to think about the mathematics rather than the mechanics.
Here is a characteristic example of R code that demonstrates its vector-oriented approach to statistical analysis:
# Simulating and analyzing experimental data in R
set.seed(42)
# Generate sample data from two treatment groups
control_group <- rnorm(200, mean = 100, sd = 15)
treatment_group <- rnorm(200, mean = 108, sd = 15)
# Perform a two-sample t-test
test_result <- t.test(treatment_group, control_group)
# Display results
cat("t-statistic:", round(test_result$statistic, 3), "\n")
cat("p-value:", format.pval(test_result$p.value), "\n")
cat("95% CI for difference in means:",
round(test_result$conf.int[1], 2), "to",
round(test_result$conf.int[2], 2), "\n")
# Create a comparison boxplot
boxplot(control_group, treatment_group,
names = c("Control", "Treatment"),
col = c("#4A90D9", "#D94A4A"),
main = "Treatment Effect Comparison",
ylab = "Response Variable")Notice the conciseness. What would require dozens of lines in a general-purpose language — generating random data, performing a statistical test, extracting results, creating a visualization — is accomplished in R with a clarity that mirrors the way statisticians actually think about their work. The language does not force the user to manage memory, declare types, or write explicit loops. It provides statistical primitives as natural elements of the language itself.
Open Source and the Rise of CRAN
The decision to release R as free and open-source software under the GNU General Public License was transformative — not just for R, but for the entire field of statistical computing. In 1995, Ihaka and Gentleman made R publicly available, and the response from the statistical community was immediate and enthusiastic. Researchers who had been locked out of expensive proprietary tools suddenly had access to a language that was not only free but extensible, transparent, and community-driven.
In 1997, the R Core Team was formed — a group of developers who would oversee the language's continued development. Ihaka and Gentleman remained involved, but the project had outgrown any two individuals. The first official stable release, version 1.0.0, arrived on February 29, 2000, a leap day that seemed fitting for a language that had taken a considerable leap of faith in choosing the open-source path.
The establishment of CRAN — the Comprehensive R Archive Network — was perhaps the single most important institutional decision in R's history. Modeled on Perl's CPAN and inspired by the broader tradition of shared computational knowledge that stretches back to the earliest days of computing, CRAN provided a centralized repository where statisticians and developers could share packages — self-contained bundles of R code, documentation, and data that extended the language's capabilities. Today, CRAN hosts over 20,000 packages, covering everything from Bayesian inference and machine learning to genomic analysis and spatial statistics. This package ecosystem is R's greatest strength and the primary reason for its dominance in academic statistics.
The breadth of CRAN reflects something fundamental about R's design philosophy: the language itself provides a powerful but deliberately minimal core, and the community builds the rest. This is not unlike the Unix philosophy of small, composable tools — a design tradition that Ihaka and Gentleman inherited and adapted for the world of statistical computing.
R and the Data Science Revolution
When Ihaka and Gentleman created R in 1993, the term "data science" did not exist in its current sense. They were building a tool for statisticians — academics who worked with relatively small datasets, published in journals, and valued mathematical rigor above all else. They could not have foreseen the explosion of interest in data-driven decision-making that would sweep through industry, government, and academia in the 2010s.
Yet R was remarkably well-positioned for this revolution. Its interactive environment made it ideal for the exploratory, iterative workflow that data science demands. Its visualization capabilities — especially after Hadley Wickham's ggplot2 package brought the grammar of graphics to R in 2005 — made it the tool of choice for creating publication-quality charts and plots. Its package ecosystem meant that when new statistical methods were developed, they could be shared as R packages within weeks, giving the entire community access to cutting-edge techniques.
The competition between R and Python for dominance in data science became one of the defining narratives of the 2010s. Python, with its broader general-purpose capabilities and its libraries like NumPy, pandas, and scikit-learn, attracted developers and engineers who approached data analysis from a software engineering perspective. R retained its dominance among statisticians, biostatisticians, and academic researchers who valued the language's deep integration with statistical methodology. The rivalry was productive: each language borrowed ideas from the other, and the field as a whole benefited from having two powerful, free, and open-source options.
In bioinformatics, R achieved a dominance that few programming languages have enjoyed in any specialized field. The Bioconductor project, launched in 2001, provided a curated repository of R packages for the analysis of genomic data. As the Human Genome Project produced vast quantities of data and next-generation sequencing technologies made genomic analysis routine, Bioconductor and R became the standard toolkit for computational biology. Today, the overwhelming majority of published genomic analyses are performed in R — a testament to the language's fitness for purpose in a field that demands both statistical rigor and computational flexibility.
The Influence of S and Scheme on R's Design
Understanding R requires understanding the two languages that shaped it most profoundly. The S language, designed by John Chambers at Bell Labs beginning in 1976, established the model for interactive statistical computing. S introduced the concept of a language where statistical analysis could be performed interactively at a command prompt, where data objects could be manipulated directly, and where graphics were integrated as a first-class capability. S's formula syntax — expressions like y ~ x1 + x2 — became R's formula syntax, and S's approach to data frames, factors, and model objects was carried over almost intact.
But R was not merely a clone of S. The influence of Scheme gave R a theoretical grounding that S lacked. Lexical scoping — the rule that a function's environment is determined by where it is defined, not where it is called — came directly from Scheme and the broader Lisp tradition. This gave R a principled approach to closures and higher-order functions that made the language more consistent and more powerful than S, even if most users never explicitly invoked these features. The functional programming idioms that R supports — passing functions as arguments, returning functions from functions, creating closures that capture their environment — all trace back to the Scheme influence that Ihaka and Gentleman brought to the design.
# Demonstrating R's functional programming capabilities
# inherited from the Scheme/Lisp tradition
# A closure: a function that creates customized functions
make_normalizer <- function(center, scale) {
# The returned function "remembers" center and scale
function(x) (x - center) / scale
}
# Create specific normalizers from data
z_score <- make_normalizer(mean(iris$Sepal.Length),
sd(iris$Sepal.Length))
# Apply using functional programming idioms
normalized <- sapply(iris$Sepal.Length, z_score)
# Higher-order functions: apply family
species_means <- tapply(iris$Sepal.Length,
iris$Species, mean)
# Function composition and piping (modern R)
library(stats)
result <- iris$Sepal.Length |>
scale() |>
as.vector() |>
abs() |>
mean()
cat("Mean absolute z-score:", round(result, 4), "\n")This blend of practical statistical computing from S and principled language design from Scheme gave R a unique character. It was a language that felt natural to statisticians — who could use it without thinking about closures or lexical scoping — while also being powerful enough for advanced users to build sophisticated abstractions. It was this dual nature that allowed R to scale from a teaching tool for undergraduates to a production language for pharmaceutical companies, central banks, and technology firms.
Ihaka's Later Reflections and the Quest for a Successor
One of the most intellectually honest aspects of Ross Ihaka's career has been his willingness to critically examine his own creation. In several talks and papers from the late 2000s and early 2010s, Ihaka openly discussed what he saw as R's fundamental limitations — limitations that, he argued, could not be fixed by incremental improvements but would require a fresh start.
His criticisms were technical and specific. R's memory model, which copies objects on modification, created performance problems when working with large datasets. The language's single-threaded execution model made it difficult to take advantage of modern multi-core processors. The lack of a formal type system meant that errors were often caught at runtime rather than at compile time, and the inconsistencies accumulated over decades of community-driven development made the language increasingly difficult to optimize.
Ihaka proposed that the statistical computing community should consider building a new language — one that retained R's strengths in statistical expressiveness and interactive analysis while addressing its architectural weaknesses. He explored the idea of basing such a language on Common Lisp, drawn by its mature compilation technology and its ability to support both interactive development and high-performance compiled code. This proposal — that the creator of one of the world's most successful programming languages would advocate for its replacement — was both radical and deeply characteristic of Ihaka's commitment to getting things right rather than defending the status quo.
The proposal did not result in a direct successor to R, but it stimulated important discussions about the future of statistical computing. Projects like Julia, which explicitly aimed to combine the ease of R and Python with the performance of C, were part of the same intellectual current. Ihaka's willingness to look beyond his own creation demonstrated a scientific temperament that valued progress over preservation — a quality that modern project management platforms increasingly seek to foster in development teams, where honest retrospectives and willingness to refactor are essential to long-term quality.
Māori Heritage and Representation in Science
Ross Ihaka's identity as a Māori scientist is not incidental to his story — it is central to understanding his significance in New Zealand and beyond. The Ngāti Kahungunu iwi, to which he belongs, is one of the largest tribal groups in New Zealand, with a rich history and deep connections to the eastern coast of the North Island. Ihaka's achievement in co-creating a programming language that is used by millions worldwide makes him one of the most internationally influential Māori scientists in history.
In a field — computer science and statistics — where indigenous peoples have been historically underrepresented, Ihaka's career stands as both an inspiration and a challenge. It demonstrates that groundbreaking contributions to technology can come from anywhere, from any background, and from any corner of the world. The University of Auckland, situated in the largest Polynesian city in the world, proved to be as fertile a ground for language innovation as Bell Labs or MIT. Geography and pedigree are not prerequisites for genius; opportunity and intellectual freedom are.
Ihaka's example has been cited by initiatives aimed at increasing Māori and Pacific Islander participation in STEM fields. His story reframes the narrative of computer science as a discipline with room for diverse perspectives and approaches — a discipline that is enriched, not diminished, by drawing on a wider range of cultural and intellectual traditions. The philosophy that animated R's design — that powerful tools should be freely available to everyone, not locked behind corporate paywalls — resonates with values of community sharing and collective benefit that are deeply rooted in Māori culture.
Philosophy and Engineering Approach
Ross Ihaka's approach to language design reflected a distinctive set of principles that set R apart from languages designed by computer scientists for computer scientists.
Design for the domain, not the machine. R was designed to make statistical thinking natural in code. Every design decision — from vector-oriented computation to the formula syntax to the integration of graphics as a core capability — was made in service of statistical practice. Ihaka understood that a language for statisticians must reduce the cognitive distance between the mathematical idea and the code that expresses it. Where Niklaus Wirth designed Pascal to teach structured programming, Ihaka and Gentleman designed R to teach and practice statistics.
Freedom over control. The decision to release R under the GNU GPL was not merely pragmatic — it was philosophical. Ihaka and Gentleman believed that statistical methods, like scientific knowledge itself, should be freely available. The open-source model ensured that R could be inspected, verified, and improved by the community, and that no single entity could control access to the tools of statistical analysis. This commitment to openness was especially significant given that R's primary competitor, S-PLUS, was proprietary.
Community as co-designer. The CRAN package system was an acknowledgment that no two people — however talented — could anticipate every statistical need. By creating an extensible architecture and a community repository, Ihaka and Gentleman effectively turned thousands of statisticians into co-designers of the language's capabilities. This distributed model of development has proven extraordinarily successful: CRAN's 20,000+ packages represent a collective investment of millions of hours of expert labor, all freely available to anyone with an internet connection.
Honest self-assessment. Ihaka's later willingness to publicly critique R's design limitations set a standard for intellectual honesty in the open-source community. Too many language creators become defenders of their creations, rationalizing flaws as features and resisting fundamental change. Ihaka's willingness to say, in effect, "we made something useful, but we can do better" is a model of scientific integrity that the technology world would do well to emulate. Agencies like Toimi that advise on technology strategy understand the importance of this kind of honest assessment — knowing when to iterate on a foundation versus when to rethink the architecture entirely.
Legacy and Modern Relevance
Ross Ihaka's legacy is measured in the scale and depth of R's influence. As of the mid-2020s, R is used by an estimated two million data scientists and statisticians worldwide. It is the standard language for statistical analysis in fields including biostatistics, epidemiology, ecology, political science, economics, and psychology. It is taught in statistics departments at virtually every major university. It is used by pharmaceutical companies for drug trial analysis, by central banks for economic modeling, by technology companies for A/B testing, and by journalists for data-driven reporting.
The Tidyverse — a collection of R packages built around a consistent design philosophy — has made R more accessible and more coherent than ever, attracting new users who might otherwise have chosen Python. The integration of R with tools like RStudio (now Posit), R Markdown, and Shiny has extended R's reach beyond statistical analysis into reproducible research, interactive dashboards, and web applications. The language that Ihaka and Gentleman built for teaching undergraduates in Auckland has become a global ecosystem.
The influence extends beyond R itself. The concept of a domain-specific language for data analysis — interactive, vectorized, with integrated visualization — has become the template for modern data science tools. Python's pandas library explicitly drew inspiration from R's data frames. Julia's designers cited R as one of the languages they sought to match in usability. Even Excel's Power Query and Google's BigQuery reflect the idea, pioneered in S and perfected in R, that data analysis should be expressed in a language that mirrors the analyst's thought process.
Perhaps most importantly, R demonstrated that a programming language created outside the traditional centers of computing power — outside the United States, outside Silicon Valley, outside the elite research universities — could achieve global dominance through the sheer quality of its design and the commitment of its community. This is Ross Ihaka's most lasting contribution: the proof that brilliant tools can emerge from unexpected places, and that when those tools are made freely available, they can change how the entire world works with data. The meticulous documentation practices embodied by the work of Donald Knuth in typesetting and literate programming find a natural echo in R's culture of reproducible research, where code, data, and narrative are woven together into documents that are both scientifically rigorous and transparently verifiable.
Key Facts
- Born Ross Robert Ihaka in Auckland, New Zealand
- Of Māori (Ngāti Kahungunu) and European descent — one of the most prominent Māori scientists
- Associate Professor at the University of Auckland, Department of Statistics
- Co-created the R programming language with Robert Gentleman, first appearing in 1993
- R's first stable public release came in 1995; version 1.0.0 was released on February 29, 2000
- R named partly after creators' initials (Ross, Robert) and as a play on the S language
- R is influenced by the S language (John Chambers, Bell Labs) and Scheme (a Lisp dialect)
- R is free and open-source software, released under the GNU General Public License
- CRAN (Comprehensive R Archive Network) hosts over 20,000 contributed packages
- R is the dominant language in academic statistics, bioinformatics, and much of data science
- Later advocated for building a successor language to address R's architectural limitations
- Explored Common Lisp as a potential foundation for next-generation statistical computing
- One of the most internationally influential scientists of Māori heritage
Frequently Asked Questions
What is R and why did Ross Ihaka and Robert Gentleman create it?
R is a free, open-source programming language and environment specifically designed for statistical computing and graphics. Ross Ihaka and Robert Gentleman created it in 1993 at the University of Auckland's Department of Statistics to solve a practical problem: they needed an accessible, capable tool for teaching statistics to their students, and the existing option — the commercial S-PLUS software based on the S language — was too expensive for their department. Rather than simply creating a cheap clone, they built a language that combined the statistical domain model of S with ideas from Scheme, a dialect of Lisp known for its clean design and functional programming capabilities. The result was a language that felt natural to statisticians, supported interactive exploration and visualization, and could be extended through user-contributed packages. By releasing R as free software under the GNU GPL, they ensured that anyone in the world could use, inspect, and improve the language — a decision that proved transformative for the field of statistical computing and eventually made R one of the most widely used programming languages in science.
How does R compare to Python for data science work?
R and Python represent two different philosophical approaches to data science. R was designed by statisticians for statistical computing and has deep, native support for statistical methods, data manipulation, and visualization. Its package ecosystem through CRAN is heavily oriented toward specialized statistical techniques, and the Tidyverse collection of packages provides a consistent, elegant grammar for data analysis. R's strength lies in statistical modeling, exploratory data analysis, and producing publication-quality graphics. Python, created by Guido van Rossum as a general-purpose language, came to data science from the software engineering side, with libraries like NumPy, pandas, and scikit-learn grafted onto a language originally designed for scripting and application development. Python's advantages are its broader general-purpose capabilities, its stronger integration with production software systems, and its dominance in deep learning through frameworks like TensorFlow and PyTorch. In practice, the choice often depends on the user's background and the task at hand: academic statisticians and biostatisticians tend to prefer R, while software engineers and machine learning practitioners tend to prefer Python. Many data scientists use both.
Why did Ross Ihaka later suggest that R needed to be replaced?
Despite creating one of the most successful programming languages in scientific computing, Ihaka was candid about R's technical limitations. His concerns were rooted in R's core architecture: the language uses a copy-on-modify memory model that creates performance problems with large datasets; it runs on a single thread, making it difficult to exploit modern multi-core processors; it lacks a formal type system, meaning many errors are only caught at runtime; and decades of community-driven development introduced inconsistencies that make the language difficult to optimize. Ihaka argued that these were not superficial issues fixable by patches but fundamental architectural decisions baked into R's foundation. He proposed exploring a new language — potentially based on Common Lisp — that would preserve R's strengths in statistical expressiveness while providing better performance, concurrency support, and type safety. This proposal reflected Ihaka's scientific temperament: he valued getting things right over defending his own creation. While no direct successor emerged from this proposal, the conversation it sparked influenced projects like the Julia programming language and continues to shape thinking about the future of statistical computing.
