And other views and questions to ask

Dr. Claudine Gay, the 30th president of Harvard University, was accused of plagiarizing by Dr. Carol Swain, who retired in 2017 from Vanderbilt University. Swain said that Gay used sections of her 1993 book and a 1997 article without proper citation.

This followed allegations by others earlier this month, culminating in a column by Ruth Marcus in yesterday’s Washington Post calling on her to resign.

Marcus does not mention Swain, but Ken covers their main academic examples below. Marcus also references the Free Beacon article linked under “month,” which has more-recent examples than the others. Update 1/2/24: Gay resigned today after new plagiarism instances surfaced—see section of our new post here.

Plagiarism

I thought that plagiarism is a pretty standard concept. Here is a definition of plagiarism:

It means using someone else’s work without giving them proper credit. In academic writing, plagiarizing involves using words, ideas, or information from a source without citing it correctly. In practice, this can mean a few different things.

Gay is accused of lifting words, phrases and sentences from other sources without proper attribution. Her papers sometimes lift passages verbatim from other scholars and at other times make minor adjustments, like changing the word “adage” to “popular saying” or “Black male children” to “young black athletes.” Does this make it okay? Not clear.

What Should Happen?

The most extreme suggestion is:

The Crimson—the Harvard student paper—says here:

A Harvard web page titled “What Constitutes Plagiarism?” says that when copying language “word for word from another source,” scholars “must give credit to the author of the source material, either by placing the source material in quotation marks and providing a clear citation, or by paraphrasing the source material and providing a clear citation.”

Here is one of their examples of Gay’s copying a large piece of text from another. The passage in question from a 1993 paper by D. Stephen Voss and Bradley Palmquist is:

• Second, the average turnout rate seems to decrease linearly as African Americans become a larger proportion of the population. This is one sign that the data contain little aggregation bias. If racial turnout rates changed depending upon a precinct’s racial mix, which is one description of bias, a linear form would be unlikely in a simple scatter plot (resulting only when changes in one race’s turnout rate somehow compensated for changes in the other’s across the graph).

The corresponding passage in Gay’s dissertation is:

• The average turnout rate seems to increase linearly as African-Americans become a larger proportion of the population. This is one sign that the data contain little aggregation bias. (If racial turnout rates changed depending upon a precinct’s racial mix, which is one way to think about bias, a linear form would be unlikely in a simple scatter plot. A linear form would only result if the changes in one race’s turnout were compensated by changes in the turnout of the other race across the graph.

The typo of a missing closing parenthesis is in Gay’s thesis. Her next sentence is: “This kind of perfect symmetry is unlikely.” This strikes us as being within the intended parenthetical comment by Gay, while the next sentence mentions aggregation bias and seems meant to be outside it.

The correspondences are highlighted in this graphic taken from the Daily Mail article:

Here is the correspondence in the other accused passage in Gay’s thesis, two pages earlier:

One thing that needs to be said right away is that this is not a wholesale lifting of an argument or example from the Palmquist-Voss paper. The Crimson quotes Voss as saying the usage was “technically plagiarism” but “fairly limited,” “not at all sneaky,” and “minor-to-inconsequential” in substance. A further quote from Voss, that “similar descriptions of technical methods are common throughout academia”—here the quotes are ours not theirs—leads into questions that I (Ken writing from this paragraph onward) will pose about the boundary to what is “common.”

Ken’s Questions

Seeing some of the other accused examples brought three further questions to my mind. The first relates to matters Dick and I encounter on this blog, the second to instances I’ve observed in our own field of computational complexity theory, and the third to both.

1. When might it suffice to say that certain people said something, without putting their words in quotes?
2. When are words giving definitions of common concepts an author’s own material?
3. When should statements of fact—apart from argumens r proved theorems—be credited to a source?

The next example cited in the Crimson, which is also featured in a story by the NY Post (via MSN) whose graphic we give below, is an instance of question 1:

The Crimson article says that Bobo and Gilliam did not feel that Gay had plagiarized their work. This last sentence is a piggy-backed instance on my part: the Crimson article has those exact words (including Gay’s thesis advisor Gary King) except that I’ve improved the grammar by inserting “that” and “had.” In any event, the article is linked at the top of this post, so anyone can check it.

The second question is something I wondered about while writing papers in computational complexity early on. Perhaps because ours was a young field, we were in the habit of having a long section “2. Preliminaries” that defined everything from scratch, culminating with definitions of complexity classes being analyzed. These definitions were formulaic and often repeated. I even recall instances of authors copying basically the entire section from another one of their own papers—though that is another matter.

The point is that nobody was thought to have purchase on common concepts. We don’t credit Jack Edmonds (or whomever) for the first definition of for polynomial time. We forget that was named as “Nick’s Class” for Nick Pippenger. When I attended a talk on the 1993 paper “On Span Programs” by Mauricio Karchmer and Avi Wigderson, I forgot that I had coined the term “span program” in 1986—until Uwe Schöning reminded me of that a few years later. The concept itself was not mine, and I did not do anything with it. Examples of community phrases are less common. For one example, although the vivid phrase “appeal to the church of the higher Hilbert space” was coined by IBM’s John Smolin in quantum computing, the trick it references is common and antedates him.

The question is over standards in other fields. Here is the first of two examples from Swain that are covered in the early article by Christopher Rufo and Christopher Brunet:

They jump in right after with: “Gay’s use of Swain’s material is a straightforward violation of the university’s rule on ‘verbatim plagiarism,’ which states that one ‘must give credit to the author of the source material, either by placing the source material in quotation marks and providing a clear citation, or by paraphrasing the source material and providing a clear citation’—neither of which Gay followed.” But this bypasses the question about the nature of the source material. These are definitions, which Swain herself cites from Hanna Pitkin. Are they common currency? I find at least one person in a forum saying so (the owner or realness of the name “Claudia” are beside the point):

The aforementioned later Free Beacon article cites Gay’s colleague Jeffrey Liebman as holding it “defensible for scholars to crib technical descriptions from each other.”

The second of the Crimson article’s two examples from a 1990 paper by David Covin involves pinching a phrase that certain centers “were to be formed in work areas, villages, prisons, candomblé and umbanda temples, samba schools, afoxés, churches, and favelas” that clearly goes beyond commonality. I find Covin properly cited for this phrase in a 2014 dissertation by Juan Diego Diaz Meneses (p53), though without exact verbatim indication:

The first Covin example touches my third question; here is the graphic from the Crimson:

Is Covin or Gonzalez the source for these facts (a) and (b)? Or were they knowable more widely? Certainly Gay should have written, “As recounted in a 1990 paper by Covin…” even though her article had no formal citations.

Finally, the other example accused by Swain—as presented in the Rufo-Brunet article—is a simple example of a common fact: “Since the 1950s the reelection rate for House members has rarely dipped below 90 percent.” Arguably the 1950s startpoint and the words “rarely dipped” are distinctive to Swain’s wording. But this leads into my last matter.

One More Angle

My question 2 on definitions of terms widens into larger ones on sourcing background. I offer the following also as a continuation of the principle of my last post that wider context needs to be taken into account. Gay’s thesis is available via ProQuest. The context leading up to the two accused paragraphs on pages 32 and 34 begins with the start of her long Chapter 3 on page 18. On page 20, she writes:

She then discusses “two stages of estimation” in her model, through page 28 where she introduces the example from Pennsylvania:

Note the relation to her advisor King’s methodology at the bottom. My point is to note how pages 32 and 34 (page 33 is a full-page figure) are bracketed by this relation:

That is, the paragraph accused as not citing Voss and Palmquist—they are nowhere cited in her thesis—ends with reference to her advisor’s method.

The angle by which her possible misconduct should be appraised is that this is really a continuing section of preliminaries building from her advisor. It is possible that this came from a series of instruction—such as in seminars—over which the reference to Voss and Palmquist dropped off. Embellishing this picture is that Voss himself was one of Gay’s instructors at Harvard over this time of development.

This is no total excuse—even granting my angle, the word “sloppiness” at best applies—but this is also the kind of thing that can happen over a chain of lectures and seminars in our field.

Open Problems

Ken and I hope we have helped with this issue. I also hope that we have not violated any rules on citations ourselves.

[add sentence after mention of “span programs”]