The Conway Disappearance Effect

Over Thanksgiving weekend, Lynn Conway sent me a link to an article that she'd written for the October edition of IEEE Computer, The Disappeared: Beyond Winning and Losing.

Mead & Conway

When I interviewed Rob Rutenbar when he received last year's (2017) Kaufman Award, he told me "I was one of the Mead & Conway generation." I, too, am a member of the Mead & Conway generation. If you are the right age, and you ended up in the semiconductor industry, then almost certainly you learned your trade from Introduction to VLSI Systems. This was written by Carver Mead at Caltech and Lynn Conway at Xerox PARC. If you want to read more about this book then see my Breakfast Bytes post The Book That Changed Everything. In fact, an earlier version of that post appeared during the period that I was blogging for EDN Magazine (remember them?) and is one of the references from Lynn's article.

In her piece, she describes how the program was set up:

In 1976, Bert Sutherland of Xerox Palo Alto Research Center (PARC) and Ivan Sutherland of Caltech launched an effort to attack this problem. A collaborative project began—I led the team at PARC using my expertise in computer architecture, and Mead led the team at Caltech using his expertise in semiconductor device physics.

But the point of the article is not that the VLSI system's program that Carver Mead and Lynn Conway "changed everything" (the title of my earlier blog post) with the book they wrote and the work on which it was based, but on how she "disappeared". For example, as a result of the work, Carver was a recipient of the Kaufman Award in 1996, but Lynn was not. Indeed, some of the awards Carver received cited work that he didn't even do, Lynn did. The nadir was:

In 2009, my disappearance was complete after the Computer History Museum’s gala celebration of the 50th anniversary of the integrated circuit. Sixteen men were described by the media as “the Valley’s founding fathers.” They were inducted into the National Inventors Hall of Fame for their contributions to microelectronics. Top billing went to Gordon Moore and Carver Mead. I was not invited to the event, and didn’t even know it was happening. Pat Castro was not mentioned, either.

Pat Castro (who is a woman, in case it's not obvious despite the androgynous name) ran the HP's Integrated Circuit Processing Lab that manufactured all the chips for some of the earliest students of the new methodology.

Irene Buchanan

I have a confession. I disappeared a Mead & Conway woman too. I provided a link above to my post The Book That Changed Everything. In both that post and the earlier one that Lynn referenced, I said that John Gray came back from Caltech and taught a course on VLSI design using the galley proofs of the book. Actually, John stayed longer, and it was Irene Buchanan, who had also gone to Caltech and came back with the galley proofs and taught the course. She politely pointed this out when she ran across the post, and I fixed it. But it is an example of how easily these things happen without the slightest bad intention. Memory is fickle 30 years later.

Other "Disappeared" People

I happened to be an undergraduate at Cambridge University when Anthony Hewish was announced as the recipient of the 1974 Nobel Prize for physics for the discovery of pulsars (along with Martin Ryle for other work in radio-astronomy not directly related to pulsars). But it was Jocelyn Bell who pretty much built the telescope, found the "scruff" on the chart paper, dug into it, and did all the work. Hewitt was her supervisor. He was also the first author on the paper announcing the discovery, and Jocelyn was the second (there were 3 others).  As a computer science undergraduate, with nothing to do with astronomy, it must have created quite a stir within the whole university for me to notice at the time. It was generally considered that "she wuz robbed" since she did all the work.

As it happened, a couple of days before Lynn sent me her article, I came across an article by a professor at Northeastern, Albert-László Barabási who had developed an algorithm for determining attribution (who "really" did the work) in scientific disciplines that flagged that Douglas Prasher should have won the Nobel Prize in 2008 for work that was given to three other scientists. They tried to track him down but he'd vanished, and didn't seem to be affiliated with any academic institution nor industrial laboratory. Then they realized that this scientist, who their algorithm considered at least Nobel-worthy, hadn't published in a decade.

But I'm falling into the same trap here, since above I said that the author of the article I was reading "had developed an algorithm", but in fact it was Hua-Wei Shen, who had joined "his" lab who developed the algorithm. Eventually, they tracked down Prasher, who turned out to be driving a courtesy van at a Toyota dealership in Huntsville Alabama. The reason he "should" have won a Nobel prize is that

Today, virtually all molecular biology labs depend on his discovery.

But he didn't even get tenure. You can read the whole story at the link above.

This algorithmic work is interesting since it is quantitative. If you are in the mood, you can dismiss almost any of these stories, Lynn Conway and Carver Mead, or Jocelyn Bell and Anthony Hewish as "sour grapes". And as the story of Douglas Prasher shows, it's not exclusively a female thing to be passed over for recognition. But Albert-László's group looked at male and female economists' tenure publication records and found:

Regardless of gender, every solo paper an economist writes increases his or her chances of tenure by 8 or 9 percent. Yet a gap suddenly appears once a woman coauthors a paper, and the chances only widen with each collaborative project she participates in. Instead of increasing her odds, every coauthored paper she contributes to lowers them. The research shows that when women coauthor, they’re accorded far less than half the usual benefits of authorship. From a tenure perspective, if you’re a female economist publishing with men, you might as well not publish at all.

So there are real biases and it is certainly not just sour grapes.

As it happens, the top female economist is probably Deirdre McCloskey and solved the tenure issue by doing it while still the male economist Donald McCloskey (if you want to read all the details she has a book Crossing: A Memoir). She has some interesting comparisons of things like faculty meetings as both sexes, having spent time doing many of the things (interrupting, speaking over people, mansplaining) that she now is on the receiving end of. Of course, I was being flippant above when I said that Deirdre solved the tenure issue by still being a man. But it raises the obvious question of whether she'd have fared differently if she'd always been a woman.

Lynn Conway wasn't so lucky, and was fired from IBM when she transitioned. She was a key contributor on the ACS team that came up with many of the ideas of modern out-of-order instruction processing. IBM didn't really make use of them immediately, but every modern microprocessor does. She then had to start again living like a foreign spy in her own country with a covert identity, terrified she might lose her career again. Of course, in that sense, she disappeared herself, and it was relatively recently that she could add her early career as a leading computer architecture researcher to her later résumé in VLSI design.

The Conway Effect

Of course, a good way to undissappear is to have a generalization named after you, like Moore's Law. Or create one yourself. Lynn has extended the Mathilda Effect (female scientists' contributions being attributed to their male colleagues) and the Mathew Effect (eminent scientists on a project get more credit than junior researchers who did the work, just because they are eminent) to what she calls the Conway Effect: 

People tend to be blind to innovations made by “others,” or those they don’t expect to make innovations. People usually don’t notice when something that has never been done before is happening right in front of their eyes. Even if people sensed that it was an innovation, they’d think a “known innovator” was responsible, not a person who isn’t expected to make innovations.

I'm not sure that is quite right since that implies that women are equally blind to innovations made by women, or Asians to innovations made by Asians, and so on. It could be true, or alternatively, it could be that people are blind to innovations made by people not like themselves.

In any case, I think the effect needs a bit of punchy editing if it's going to catch on. That worked for Fred Brooks in his book The Mythical Man Month, with Brooks Law: "adding more manpower to a late software project makes it later." Maybe the Conway Effect just needs to be: "we are blind to innovations made by people not like us." Prize committees, Swedish and otherwise, take note.

Rosalind Franklin

There is another Cambridge-related story of a woman who is sometimes considered passed over. Francis Crick and James Watson won the 1962 Nobel Prize for Biology for the discovery of the double helix structure of DNA, and some people believe Rosalind Franklin should have got the award too.

Crick and Watson's work depended crucially on her X-ray crystallography. Rosalind, in addition to being a woman, was at the less prestigious King's College London. Her "photo 51" to the right was critical evidence in the structure of DNA. But just to show you how complicated these things get, Rosalind Franklin didn't create the image, her student Raymond Gosling did. And by the time Crick and Watson saw it (actually just Watson), Franklin had left the lab and Gosling was supervised by Maurice Wilkins who had produced some of the earlier, but less clear, X-ray images.

Everybody knows that Crick and Watson were awarded the Nobel Prize for Biology, as I said above. But everybody is wrong, or at least only half-right. There is no Nobel Prize for Biology, it is actually the Nobel Prize for Physiology or Medicine. Further, it actually went to Crick, Watson, and Wilkins. In some ways, Wilkins is another disappeared person. I'm pretty sure you've never heard of him.

Rosalind, unfortunately, died in 1958 from cancer, and the Nobel Prize is not awarded posthumously, so she was not eligible in 1962.

So I leave it as a final exercise for the reader to decide whether, had she lived, she would have shared the Nobel Prize with Crick and Watson (three is the maximum number of people who can share a Nobel Prize, so she would have had to replace Wilkins). Or whether she'd get hit by the Mathilda Effect (as a woman), the Matthew Effect (less senior), and the Conway Effect (not like the Nobel Prize committee for Biology...I mean Physiology or Medicine).

Trivia

And a final trivia question: who is the only person to win two Nobel Prizes for science in two different scientific disciplines? You can read the answer in my post Discovery of the Electron.

Want a hint? I said "person" very deliberately, and given the topic of this post, it will help you to guess why.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.