Les Perelman, MIT Writing Assessment Pioneer and Critic of Automated Grading, Dies at 77

Les Perelman, a longtime MIT faculty member whose work fundamentally shaped writing instruction at the Institute and who became a nationally recognized critic of automated essay grading, died on November 12, 2025, at his home in Lexington, Massachusetts. He was 77.

Perelman's career at MIT spanned 35 years, during which he established the Writing Across the Curriculum program and became known for his pointed critiques of standardized writing assessments. His work challenged the education technology industry's approach to evaluating student writing, arguing that many automated systems rewarded formulaic responses over genuine thinking and communication.

Born in Los Angeles, Perelman earned his PhD in English from the University of Massachusetts at Amherst in 1980 after participating in Berkeley's activist years as an undergraduate. He joined MIT in 1987 after teaching positions at the University of Southern California and Tulane University.

His widow, MIT Professor Emerita Elizabeth Garrels, recalls Perelman's ability to demonstrate the flaws in automated grading systems. At one conference, he arranged to stand at one end of a room and "grade" essays held up at the other side, consistently calling out the scores that standardized systems would assign. "He was practically blind without his glasses," Garrels says, "but he could predict automated grading outcomes with remarkable accuracy."

Perelman's most famous critique involved his BABEL (Basic Automatic B.S. Essay Language) Generator, developed in 2020. The software produced nonsense essays that commercial autograders nevertheless scored highly. One example he shared included absurd claims like "Teaching assistants are paid an excessive amount of money" and non-factual references to Dickens' "Great Expectations," yet still received top marks from automated systems.

His criticism extended to academic research supporting automated grading. He notably challenged a University of Akron study claiming autograders performed as well as human graders, questioning the methodology and conclusions.

Not all organizations rejected his critiques. The Educational Testing Service (ETS) allowed Perelman to test its autograder, where he discovered he could achieve perfect scores even with essays containing gibberish and typographical errors.

Within MIT, Perelman's most lasting contribution was his push to embed writing instruction throughout the curriculum rather than treating it as a standalone subject or a hurdle to be cleared. With a $325,000 National Science Foundation grant, he convinced MIT to hire writing instructors who were also subject matter experts, often with STEM PhDs.

This initiative evolved into the Writing Across the Curriculum program, now called Writing, Rhetoric, and Professional Communication, which employs more than 30 instructors. The program's goal was to ensure students developed communication skills alongside technical expertise.

Building this infrastructure took nearly 15 years, according to his successor Suzanne Lane. Perelman first had to demonstrate the unevenness of writing instruction at MIT, where some students received extensive communication training while others graduated with minimal writing experience.

Alumni surveys proved crucial to his case. While MIT graduates rated their technical skills highly, they consistently reported that their communication abilities were poorly developed despite being critical to career success. This gap between importance and preparation helped drive support for a formal communication requirement.

A 1997 pilot program, supported by the NSF grant, tested communication-intensive courses across all departments. The experiment involved 24 subjects and approximately 300 students. Following "lively" faculty discussion, MIT approved the communication requirement in 1999, with formal implementation beginning in fall 2001.

The requirement has since expanded from the initial 24 subjects to nearly 300 across the Institute, ranging from civil engineering design courses to linguistics workshops.

Perelman's influence extended beyond MIT. He co-authored "The Middle English Letter of Alexander to Aristotle" with Vincent DiMarco and worked as a technical writer for Wang Computers. He edited books on writing studies with New Jersey Institute of Technology professor Norbert Elliot.

In 2018, he collaborated with the New South Wales Teachers Federation to convince Australia to reject automated essay grading systems.

Nancy Sommers, whose work on undergraduate writing assessment at Harvard paralleled Perelman's, described him as "brilliant, with a Talmudic way of asking questions and entering academic debates." She recalled his generosity as a friend, including waiting at an airline gate with food after a canceled flight.

Perelman's critiques of automated grading systems highlighted a fundamental tension in education technology: the balance between scalable assessment and meaningful evaluation of complex skills like writing. His work suggested that many automated systems, despite their sophistication, could be gamed through formulaic responses that satisfied algorithmic criteria without demonstrating genuine understanding or communication ability.

His legacy at MIT represents a broader recognition that technical education must include strong communication skills. The communication requirement ensures that MIT graduates can not only solve complex problems but also explain their solutions effectively to diverse audiences.

Donations in Perelman's name can be made to UNICEF's work supporting children in Ukraine, the Lexington Refugee Assistance Program, Doctors Without Borders, and the Ash Grove Movie Finishing Fund.

Perelman's career demonstrates how one person's persistent questioning of educational practices can lead to institutional change. By combining technical critique with practical program development, he helped MIT recognize that writing instruction isn't a luxury or a hurdle, but an essential component of technical education.

His work continues to influence debates about the role of technology in education, particularly regarding the assessment of skills that resist simple quantification. As automated grading systems become more sophisticated, Perelman's critiques remain relevant to discussions about what we value in student writing and how we measure it.