Blog
0 comments

An Unassuming Giant in Her Field: Radia Perlman's Work Speaks Volumes

March 8, 2021
Radia Perlman circa 1976

By Dan Rosenheim

Let it be said, loudly and unambiguously, that Radia Perlman’s unparalleled contribution to the existence of a functioning Internet is a product of her remarkable intellectual gifts.

The creator of important innovations that make modern computer networks reliable and easy to manage, including the well-known Spanning Tree algorithm, and author of the seminal network book, Interconnections, Perlman has been widely recognized  in articles, awards and in membership in the National Academy of Engineering, Inventor Hall of Fame and Internet Hall of Fame.

But in an interview from her home in suburban Seattle, Perlman is strikingly self-effacing, going so far as to suggest that happenstance  even in the form of an unwelcome #MeToo experience  may have helped open the door to her stellar career.

“I guess I just don’t project an aura of self-importance,” she said during an hour-long session on Zoom. “I'm sort of glad I don't because it means that I'm approachable, although it is sometimes frustrating when people don’t hear you.”

Perlman’s #MeToo experience in 1976 was a final affront in her pursuit of a graduate degree in math from MIT. It prompted her to leave school and go to work in the private sector  where, within a short period of time, she had come up with important network innovations that are still used by the Internet today  and when she finally did return to MIT, it was for a PhD in computer science. 

The first time in graduate school, after completing all other requirements, Perlman couldn’t find a professor to serve as her thesis advisor.

“The Math Department at that time didn’t assign you an advisor  you had to find one,” she recalled. “I was very shy, but I would knock on the doors of professors, and each one would say, ‘I don’t have time for you.’ So, I was really struggling to get my thesis started.”

But the last straw, she said, was when she told a professor she had discovered a new way to solve a math problem that had recently been published and asked to discuss it with him.

He, too, said he was busy during the day and suggested talking over dinner.

“Colleagues routinely had dinner together, so it didn’t occur to me this could turn out badly. When he said it would be quieter at his house and didn’t mind driving me back after dinner, I asked whether his wife had dinner planned or we should pick something up, and he said, ‘Oh, it’s all planned.’

“And then of course I show up, and his wife and kids were out of town. I didn’t know what to do because I was stuck in Concord with this disgusting man, and I tried all sorts of polite ways to tell him no, but I was afraid he could destroy my career by just telling people I wasn’t that bright.” 

Spurred by this unwelcome experience, Perlman dropped out of grad school, taking a job in software development for government contractor Bolt Beranek and Newman (BBN). Before long, her work drew the attention of Digital Equipment Corp., which she joined in 1980, and it was there that she did a lot of foundational work in computer networks.

“So, the ugliness with the professor actually ended up great because, had I stayed at MIT, I would have done some tiny incremental thesis,” Perlman said. “Instead, as it turns out, dropping out landed me in the world's best job at the best time in history.” 

The #MeToo incident was hardly the only time Perlman felt the extra weight of being female in a field long notorious for its domination by males. She was one of only 50 women in a class of 1,000 at MIT, for starters.

And there was more than one occasion at school and in the business world when her words  and her work  failed to draw the response that a male colleague might have elicited.

While employed by BBN, for instance, Perlman solved a network routing problem that no one else had figured out. But when she presented her findings at a technology conference, the audience was silent; it was as though she hadn’t said a word. 

“After the meeting, an executive from Digital came up to me and said ‘That’s incredible  we’ve been thinking about the problem for so long! Your solution is so simple, and…once you see it…so obvious. But you were just completely ignored. Doesn’t that bother you?’

“And I said: ‘I’m used to it.’” 

But while Perlman acknowledges the extra burden women face, she is disinclined to emphasize it.  

“You know, whether it’s a gender thing, I’m not sure,” she said. “It certainly doesn’t help, but I don’t like to play the gender card. I more like to say it’s the personalities.”

Indeed, Perlman said, her concern is less with gender bias than with those who elevate themselves by being pushy and self-promoting.

“I have worked with women, as well as men, who are unbelievably self-promoting,” Perlman said. “And people get taken in by them  I tend to get taken in by them, too, at first blush. But I’ve never seen anyone like that, man or woman, who is actually any good technically.”

And if Perlman’s quiet voice had trouble making itself heard for a bit, her work ultimately has had no trouble speaking for her  coming through loud and clear. 

After 10 years as a professional, Radia took a paid educational sabbatical from Digital and returned to MIT for her doctorate. There, in 1988, she completed a thesis that disclosed a way to make networks function properly even if some switches are “malicious,” or misbehaving. To this day  32 years later  Perlman meets professors at the school who say that thesis is the first thing their students are told to read.

And what really put Perlman on the programming map  even more than Spanning Tree  was the 1992 publication of her book Interconnections, which is regarded as a kind of Bible for network theory and practice.

“The things I invented are not called ‘Perlman’s thing,’ but they are the way connections occur. It’s just the way things work,” she said matter-of-factly.

“The only reason anybody realizes I did anything is that luckily I wrote the book,” Perlman added, “and it became the book everybody had to have on their bookshelf.”

Perlman subsequently published a second book, Network Security, co-authored with engineers Mike Speciner and her current life partner Charlie Kaufman.  And she has been in demand as an educator, having taught at Harvard, the University of Washington  and developed computer programming tools even for young children. Indeed, Perlman is as passionate about educating as she is about the science of computer networks.

“I think I have a feel for people, and I love teaching,” she said. “When they said (the software program) Logo would revolutionize education, I noticed that kids weren’t really understanding it.”

So, Perlman developed a new, child-friendly way of teaching the Logo concepts, using a specialized keyboard and a system where computer commands were objects that could physically be combined to form programs.

Today, Perlman’s instruction has taken a new form. Employed by Dell as a Fellow with responsibility for network protocols and cryptography, she is also referred to by the company’s CTO as “Dell’s BS Detector.”

As a current example of overhype in the public mind, Perlman points to blockchain, the mechanism that enables cryptocurrencies like bitcoin.

“Bitcoin is unbelievably expensive,” she adds, “but ask yourself why you need it.

“We can do digital currency with credit cards and wire transfers  why do we need this other stuff? Pretty much the only killer app for it is ransomware and other kinds of illegal things.” 

Quantum computing is another area where Perlman finds misconceptions alarming.

“People hear that a quantum computer is simply a faster version of current computers…like a quadrillion zillion times faster, and eventually we’ll all be using quantum laptops.  But that is utterly wrong.  There are only a very narrow set of problems that a quantum computer might be better at solving. 

"So, again, I sort of take it upon myself to understand these things and speak sanity to people when they’re going off in the wrong direction.”

A central tenet in Perlman’s approach is the need to define an issue clearly before seeking a solution.

“That’s a real problem in the industry where people start writing code before they clearly understand the problem,” she said, “and when the code doesn’t work, they add more and more stuff, and whatever.”

By way of illustration, Perlman offered an example from her own personal life, one that has nothing whatsoever to do with data processing, “but that will forever make you realize why you should know what problem you are solving before you try to solve it.

“When my son was three, he ran up to me crying, holding up his hand and saying, ‘my hand, my hand.’

“So, I took his hand and kissed it a few times, saying, ‘What’s the matter, honey, did you hurt it?’

“And he said, ‘No, I got pee on it.’"

That son, Ray, is now an engineer in his own right and will be co-author number four on the next edition of Network Security.  A daughter, Dawn, an accomplished musician with an advanced degree in marketing, raises three children on the East Coast. Radia, meanwhile, lives happily with Charlie Kaufman, whom she met at Digital where he was the network security architect and she was the routing architect.

“I happen to be quarantined with my soulmate for 24 hours a day, which is not enough time to spend with him,” she said. “The only time I’ve ever criticized him is if he’s hugging me, I say, ‘I wish you had more arms.’ We’re just disgustingly cute.”

Perlman spends the pandemic working for Dell, attending remote conferences, working on the next edition of the book, playing piano, walking, cooking, watching “way too much TV with Charlie” and Zooming with family. 

All of those things, along with her intrinsic energy and strong sense of humor, help to temper the growing pessimism she feels about the Internet and the general state of the world.

“Ten years ago, I would talk with great enthusiasm about the Internet  you can reach a worldwide audience, you can buy any widget you want, you can get information for free, and all this stuff is just marvelous,” she said. 

“But now I think the Internet is the end of civilization. Anyone can say anything, and A.I. figures out what you like and gives you more and more extreme things, not to mention hostile foreign powers. It’s terrifying, and I don’t really know how to fix that.

“In my darker times I think, well, all this hate, the end of truth, climate change  maybe we don’t deserve to live any more. Maybe we’ll just go extinct, and some better species will come along.”

Dour thinking indeed, but the fatalism it expresses is belied by Perlman’s warmth, resilient sense of humor and an apparently undiminished well of enthusiasm.

And along with the joys in her personal life, Perlman still seems motivated by the knowledge-based passions that have driven much of her professional life: the delight in discovery, the desire to solve problems, the satisfaction of teaching others. 

“Once you get to be fairly senior, she said, “people assume you know everything. But if you believe you know everything, you should retire, because you’re dangerous. I still have things to discover and, hopefully, a lot to share. No plans for retiring.”

 

Dan Rosenheim was managing editor of The San Francisco Chronicle and vice president/news for KPIX-TV. Retired from full-time work, he is currently business development director for Bay City News and LocalNewsMatters.org.