Craig Wright Plagiarized Significant
Portions of His PhD Thesis and Tried to Hide It
Note: This article has been updated since its original publication. The update can be found at the end of the article.
Craig Wright was awarded his PhD by Charles Sturt University in 2017. His doctoral thesis (archived link) is titled The Quantification of Information Systems Risk: A Look at Quantitative Responses to Information Security Issues.
Substantial, deliberate plagiarism is present in at least thirty pages of Wright’s thesis, including almost the entirety of Chapter 6.¹
Wright plagiarized huge swaths of content and reworded it to avoid automated detection tools. In most cases, he simply substituted synonyms every few words. For example:
Source (first published in 2004) (page 112):
“In building each decision tree model based on a different random subset of the training dataset a random subset of the available variables is used to choose how best to partition the dataset at each node.”
Wright (page 32):
“In the construction of each decision-tree model, an individual random subset of the training dataset uses a random subset of the presented variables to decide where to partition the dataset at each node.”
In almost every case, he did not even bother to change the sentence order.
_________________________
¹ Wright has a well-documented history of plagiarism. See, e.g. — (1, 2, 3, 4, 5, 6, 7, 8). Most recently, he was caught having plagiarized the majority of his 2008 LLM thesis.
Below is a collection of comparison images detailing this plagiarism. The stolen parts are color-coded to make it easier to see how Wright kept the same sentences in the same order. Wright’s thesis will be on the left, with the original source on the right. The words and phrases in black boxes indicate the parts that Wright stole verbatim or near-verbatim.
(Note: None of the sources mentioned here were cited by Wright in his thesis, nor did he cite works that substantially include the material in the form which it appears here. This and other potential excuses will be discussed at the close of the article.)
As we progress through these examples, the reader might keep in mind the pledge Wright made in the Certificate of Authorship, signed and included in the opening pages of the thesis:
I hereby declare that this submission is my own work and to the best of my knowledge and belief, understand that it contains no material previously published or written by another person, nor material which to a substantial extent has been accepted for the award of any other degree or diploma at Charles Sturt University or any other educational institution, except where due acknowledgement is made in the thesis.
Errors in Transcription
As in previous instances of Wright’s plagiarism, Wright usually introduced errors when he copied material, especially when it contained math. For example, in the next image, Wright took a figure originally captioned “optimal choice” and erroneously wrote “optional choice”. He also attempted to obfuscate the equations by choosing different variable notation, but confused himself in the process and made a few errors, which are highlighted in red boxes.
Sometimes, Wright even forgot to define terms that he stole from the source material, or he introduced an equation without including the intermediate steps necessary for it to make sense.
In other cases, Wright copied clear errors that were already present in the source material and did not correct them.
Many Different Sources
Wright did not limit his plagiarism to published papers or books. He also plagiarized extensively from webpages, as the following image shows:
Wright stole approximately seven pages of material for his thesis from an ornithology professor’s webpage:
https://web.archive.org/web/20050710084210/http://people.eku.edu/ritchisong/homepage.htm
Wright also stole from slideshow presentations, and on occasion he mixed his stolen content from two sources simultaneously.
Stolen Figures
At least ten different figures were stolen, most of which were deliberately modified to avoid detection. For example, Wright stole a figure from a 1982 textbook about bird nest defense and re-labeled the X-Axis to read “Time spent not defending systems” instead of “Time Away from Nest”. The hastily-edited version first appeared in Wright’s 2012 paper, “Territorial Behavior and the Economics of Botnets”. Wright created a new version of the same figure for his PhD thesis in an attempt to hide the plagiarism.
There are multiple other examples of the same behavior:
The plagiarism was extensive
The following images are additional instances of plagiarism:
Response to Potential Excuses
Potential excuse 1: “I cited the Blanchard book in two of my papers. They are cited in this thesis.”
Response: The Blanchard book (the source for only 10% of the discovered plagiarism) is not cited in the thesis itself, and while it is true that two cited papers themselves cite Blanchard, the citations to Blanchard in those two papers do not cover the plagiarized material at all, nor would a simple cite be enough to cover the extensiveness of this plagiarism. Also, nearly all of the 50+ citations in those two papers were moved over to the dissertation’s bibliography; the Blanchard book was specifically left out. If anything, it simply proves that Wright had the source book in his possession, and purposely tried to hide it.
Potential excuse 2: It was the fault of a copy editor / style editor / someone else who forgot to include these citations.
Response: The thesis includes the signed Certificate of Authorship from Wright:
It also includes a statement about the copy-editor indicating that the edits were “limited to formatting, grammar and style”:
Finally, Wright himself admits that “[a]ll remaining errors are [his] own” in his acknowledgements:
Potential Excuse 3: These are just a few mistakes. It’s difficult to include every citation in a work like this. Mistakes are inevitable.
Response: The extensiveness of the plagiarism makes it clear this excuse is bogus. Further, the changes to avoid detection prove that the plagiarism was intentional and not a “mistake”.
Potential Excuse 4: The citations are all there, but they’re covered in the papers that I did cite. If you “follow the citation trail”, you’ll eventually hit the source material.
Response: This is simply false. While Wright will most likely try to rely on this excuse, as it requires significant effort to disprove, it should be noted that, even if this claim were true, it would not excuse the plagiarism found here. However, it is decidedly not true.
One easy way to prove this claim false is to search for some of the verbatim-copied text. Anyone with access to University resources can also verify this by searching academic publications. For instance, a search for the exact phrase “right-censored at some censoring time” only bears two results: Wright’s thesis and the Cook (2003) source.
Similarly, searching for “optimal time in near patch at high attack rate” results in only two distinct results: the ornithology professor’s website and Wright’s papers.
Even if some of his citations touch on the subject matter, it is clear that he lifted the actual content from the sources identified here, and he did not cite them. Finally, most of the plagiarized text from the thesis has no citation at all.
Potential Excuse 5: All the sources actually plagiarized Wright.
Response: While the notion that all ten currently identified sources plagiarized Wright (and fixed his many errors!) is absurd on its face, it is easily disprovable in most cases. For example, the highly technical Blanchard book was written in 1981 when Wright was approximately eleven years old.
As he has done before, Wright may pretend that he possesses old drafts that are inaccessible or were not publicly available until now. However, the burden is clearly on him to prove that these are authentic.
In addition, there are several papers Wright has published that contain the plagiarized material even more directly. For example, this paper took three figures from Horvath (2006) without attribution. Notice, as before, Wright modified the X-Axis to remove the original caption from one of the figures. (This particular figure also appears in the PhD thesis on page 32.)
Original (notice the caption about kidney cancer):
And after Wright’s modification:
Potential Excuse 6: This doesn’t affect the content of the thesis. He still deserves a PhD.
Response: Significant portions of the identified plagiarism cover material that Wright pretended to be original research. Furthermore, the examples of plagiarism shown here were identified using simple Google searches (then carefully verified). Given the effort Wright put into obfuscation, there is undoubtedly much more plagiarized material in the thesis.
Potential Excuse 7: “My source papers actually had the citations in them.”
Response: They do not have the citations in them. However, Wright used this excuse after he was caught plagiarizing his 2008 LLM. He uploaded a new version and pretended the citations were always there. To avoid this, the Appendix (4.) contains archived versions of Wright’s texts as they appear at the time this was written.
Appendix
1. Full-size versions of the images found here can be accessed here.
2. (Likely incomplete) list of pages in Wright’s thesis that contain plagiarized material:
Pages 27–33
Pages 94–95
Pages 97–98
Pages 151–152
Pages 232–235
Pages 237–238
Pages 240–241
Page 245
Pages 247–248
Pages 252–254
Pages 343–345
3. (Likely incomplete) list of sources that Wright plagiarized from:
Anderton, Charles & Carter, J.R.. (2009).
Principles of Conflict Economics: A Primer for Social Scientists.
10.1017/CBO9780511813474.
Baumann, Reto. Ethical Hacking. (2002)
https://www.giac.org/paper/gsec/2468/ethical-hacking/104292
Blanchard and Fabrycky. Systems Engineering and Analysis (1981)
Burroni and Sarraute, “Using Neural Networks for remote OS Identification (2005)
Accessible at:
https://www.coresecurity.com/sites/default/private-files/publication/2016/05/Burroni-Sarraute-PacSec2005.pdf
Cook, Richard & Lawless, Jerald & Lee, Ker-Ai. (2003).
Cumulative processes related to event histories.
Statistics and Operations Research Transactions. 27.
Available at:
http://dmle.icmat.es/pdf/SORT_2003_27_01_02.pdf
Horvath, Steve
Unsupervised Learning with Random Forest Predictors: Applied to Tissue Microarray Data
Biostatistics and Human Genetics
University of California, LA
2006
Accessible at: https://gradebuddy.com/doc/2584667/unsupervised-learning/
Ritchison, Gary (Eastern Kentucky University Professor) webpages:
https://web.archive.org/web/20060925174020/http://people.eku.edu/ritchisong/behavecolnotes2.htm
https://web.archive.org/web/20071123053409/http://people.eku.edu:80/ritchisong/RITCHISO/behavecolnotes4.htm
Shor, Mike. GameTheory.net (2003 Version)
https://web.archive.org/web/20030816145520/http://gametheory.net/Dictionary/Games/StagHunt.html
Shvarts, Alexander. “The Russian Mafia: Do Rational Choice Models Apply?”
Michigan Sociological Review, vol. 15, 2001, pp. 29–63. JSTOR, www.jstor.org/stable/40969058
Williams, Graham.
https://web.archive.org/web/20080719184320/http://datamining.togaware.com/survivor/Random_Forests.html
taken from his book “Data Mining: Desktop Survival Guide” (2004) (ISBN 0975710923)
https://zoo.cs.yale.edu/classes/cs445/misc/mar13lae08.pdf
4. List of archived versions of Wright’s texts
A comparative study of attacks against Corporate IIS and Apache Web Servers (2011) (cache)
Compliance or Security, What Cost? (2011)
Criminal Specialization as a Corollary of Rational Choice (2010)
The Economics of Developing Security Embedded Software (w/ Tanveer Zia, 2010) (cache)
The IT Regulatory and Standards Compliance Handbook (2008)
Modeling System Audit as a Sequential test with Discovery as a Failure Time Endpoint (w/ Tanveer Zia, 2012)
Of Black Swans, Platypii and Bunyips: The outlier and normal incident in risk management (2011)
A Preamble Into Aligning Systems Engineering and Information Security Risk (2011) (cache)
A Quantitative Analysis into the Economics of Correcting Software Bugs (2011)
Random Forests in Decisions: Abstract
Rationally Opting for the Insecure Alternative: Negative Externalities and the Selection of Security Controls (2011) (cache)
Software, Vendors and Reputation: An Analysis of the Dilemma in Creating Secure Software (2010)
UPDATE: 11 May 2020
Craig Wright has responded in part to certain items in this article in a blog post. It can be found here (archive here)
Wright planted his limited rebuttal in the weeds of a rambling diatribe addressing a host of unrelated grievances. It appears he views this article as part of some long-running and sinister grand conspiracy against him — a “hit piece” — but the reality is less glamorous. It is merely a factual report on plagiarism in Wright’s PhD thesis. Accordingly, this update is concerned with Wright’s challenges to points raised in this article.
First must be addressed Wright’s newly-narrowed, softer take on plagiarism in light of this report. He had earlier taken a more expansive view:
Plagiarism varies in its extent. It goes from simply rephrasing the ideas of another without referencing your sources right through to the literal block copy of paragraphs of text and the theft of entire passages.
Wright also had taken a hardline stance on the issue, calling plagiarism a “criminal fraud”. He was able to call out specific instances of plagiarism by others in detail, as here.
Now, to set up a basis to deny the instances of plagiarism reported in this article, Wright picks out the first sentence of Oxford University’s guidance on plagiarism: “Plagiarism is presenting someone else’s work or ideas as your own, with or without their consent, by incorporating it into your work without full acknowledgement.”
Looking further down that same guide, we can find a more comprehensive explanation. Much of these cautions were ill-observed in Wright’s thesis.
Wright claims (wrongly) that this author uses a special definition of plagiarism, one based on a faulty understanding due to lack of expertise. But even if that were true, it wouldn’t matter. Both Oxford and this author have no bearing on the facts. Charles Sturt University is the institution involved here, so we can point to CSU’s misconduct policy for guidance in this matter.
Indeed, this policy is reflected in Wright’s signed Certificate of Authorship reproduced in this article.
Now that we’re all on the same page, we can proceed to the rebuttal.
Wright did not address all the examples of plagiarism, but chose to focus on only two specific examples. (There is no defense of the stolen ornithology content, for example.) His main points raised are (paraphrased):
- Writing down well-known equations is not plagiarism, and they don’t need to be spelled out or derived in full.
- I’m not claiming credit for the ideas described, and they’re well-known in the field, therefore citations are unnecessary.
- I originally had a lot more content and citations in a much larger draft, but my supervisor told me to cut out most of it.
- The figures I’m accused of plagiarizing are common types of charts. Anyone making charts illustrating these concepts would end up with substantially identical figures, which is why they appear in lots of papers without any citations.
Wright’s post completely avoids engaging the main point of the original report: that he engaged in extensive and deliberate plagiarism, copying ideas, lengthy segments of text, and illustrations from various authors, and trying to cover his tracks. Instead, he miscasts the report as somehow being about a few inconsequential academic nitpicks. Wright does not give any answers to the big questions, and instead reframes the issue and focuses on small ones.
Before addressing specifics, the following general response applies to excuses 1, 2 and 4:
Wright didn’t just copy ideas and equations; he copied the extensive explanation of those ideas and equations, in nearly the same exact words and mathematical notation, that someone else worked hard to create.
Excuse 1: Writing down well-known equations is not plagiarism, and they don’t need to be spelled out or derived in full.
Wright’s argument here is:
In the first example, shown below, Mr Maxwell [sic] has pulled out a text from Blanchard & Fabrycky (1981) and commented that I copied the formula without going through the full derivation. … I’m not attempting to derive the formula for the net present value, but I’m merely putting it down. You may ask, why have I not quoted the text he references? The simple answer here is that Blanchard & Fabrycky themselves did not derive the present equivalence function or net present value calculation and formula. So, you could ask, why have they not referenced it?
Are the authors of such an esteemed textbook also plagiarists?
The figure he uses is:
Here Wright completely miscasts the issue. It is not simply that he omitted the derivation. It is that he failed to define the very terms he uses, which shows it was not his own work. Even if his assertion is correct — that he is merely presenting the equations as a conclusion or reminder to the reader — it’s impossible for the reader to follow the equations because they are not fully defined. Wright simply copied it.
Similarly, Wright omits the fact that while the math — that is, the relationships between variables — may be common knowledge, the expression of that math is not. Different authors often use different variable names or notations to denote existing concepts, and this is not a problem because each author always walks their reader through the necessary definitions. Wright’s thesis, by copying only parts of another author’s equations and definitions, is left with a broken train of thought that cannot be properly followed; it is yet another indicator that he did not write these parts himself.
The whataboutism around Blanchard’s “plagiarism” is just a red herring. Blanchard does nothing remotely similar to what Wright did in terms of copying other material.
The reader is encouraged to click through Wright’s links that purport to show additional people “plagiarizing” the formulas. Do try to find an instance where the formulas are shown exactly the way they are in Wright’s and Blanchard’s presentation, with the same notation and concept flow — that is, the expression of the math. You will not find it, as Wright copied directly from Blanchard, who did not copy from others. Copying the expression of the math is plagiarism.
Again, Wright may have us think that he didn’t do anything amiss here, that he just used a well-known standard formula. This is simply not the case. Wright copied a specific author’s expression of the math, including the surrounding text, and pretended this was Wright’s own explanation. Passing off someone else’s work as your own is the textbook definition of plagiarism, as Wright well knows.
Excuse 2: I’m not claiming credit for the ideas described, and they’re well-known in the field, therefore citations are unnecessary.
As is apparent from the links Wright provides, the concept of ‘net present value’ can be explained in many different (basically equivalent) ways. Wright was free to present the ideas and formulas in his own way, but chose to (incompletely) copy from an already-made version without citing it, which made it more confusing. This is undeniably plagiarism.
This is perhaps the heart of Wright’s excuse:
When cherry-picking, what [they] are neglecting is that no source is seeking to cite the origin of everything. To get into such level of detail, I would have to start citing the origin of the term plus. Where did basic algebra come from, where did the sigma symbol come from, what is the history of interest? No, at a certain point, you do not quote ideas as anything other than given.
This is quite absurd. To be clear: Wright did not merely repeat well known ideas. He copied extensive swaths of content, including equations, figures, and the accompanying textual descriptions.
The excuse is laid most bare here:
I say it again: nobody at a university level would believe that you are saying you are creating net present value calculations by putting it in a thesis. It is simply the use of a common formula.
If Wright merely mentioned a well-known idea in his own words, this might apply. However, that’s far from the case here. Wright didn’t just mention a well-known idea; he copied the extensive explanation of that idea, in the same exact words and mathematical notation, that someone else worked to create.
Wright goes on (emphasis added):
Predator prey games, Jean-Jacques Rousseau’s 18th-century concept of a stag hunt, and many other topics that I have covered in my thesis may be covered without going into the detail of the source. There is a simple reason for doing so: the people evaluating it know that it is not my idea. I am not claiming to have invented a stag hunt; I am using a standard model. I am not claiming to have invented predator prey games; I am using a standard model.
Again, Wright was free to present the ideas that he’s talking about here, but instead he chose to copy someone else’s work of presenting them in an accessible and understandable way, and thus acted as if the presentation were his own. The text was copied almost verbatim, which could be why Wright didn’t actually include the image, reproduced here.
Excuse 3: I originally had a lot more content and citations in a much larger draft, but my supervisor told me to cut out most of it.
Wright next tries to claim that his PhD supervisor, Tanveer Zia, recommended he remove citations simply because there were originally too many.
Besides, when I initially submitted my thesis for examination, on 17th December, 2012, it was, summarily, returned to me without much review because of its length. I had created a 636-page monster. The first thing that my supervisor, Tanveer Zia, did was to return it to me. Prof Zia and others acted to supervise.
While there is no evidence given for the 636-page claim, the implication is that Wright originally had more citations, and his advisors recommended he remove them. Wright makes this more explicit later in his post (emphasis added):
As explained, I had to reduce the size of my thesis. In my original section of notes, that was then removed, I had a whole lot of extra material. Some of it included details of the papers I got calculations from. You could say, they should be in my thesis submission, yet, I was told to take them out. Both of my supervisors and the head of school and later the reviewers from other universities, such as the University of Melbourne, told me that the section was unnecessary and to be removed.
It also means that once you remove the notes, you remove the references in the notes.
As evidence for this accusation, Wright gives the following screenshot:
It’s not clear what significance is supposed to be derived from this screenshot; Tassey is not related to any of the uncovered examples of plagiarism, and that exact reference is still present in the final version of the thesis on page 358:
Not only is Wright mistaken about this being an instance of a reference cut from the paper, there is also no reason to believe he cut relevant references at all.
Finally, we’re being asked to take Wright’s word for these supposed additional references having ever existed, but they don’t exist in the thesis for which Wright was awarded a PhD.
Excuse 4: The figures I’m accused of plagiarizing are common types of charts. Anyone making charts illustrating these concepts would end up with substantially identical figures, which is why they appear in lots of papers without any citations.
Wright next tries to argue that some of the figures he stole are also so ubiquitous that he didn’t need to cite them. He gives many examples of “consumer choice”-type charts.
He writes:
So yes, my diagram is a horrible, ugly version of the general methodology of consumer choice theory, the same every single microeconomist uses. My only apology is that I am not a better artist.
Wright has rounded up a number of similar graphs of the same topic, but the wide variation visible in these graphs only serves to underscore how conspicuously near-identical Wright’s graph is to the source, matching it point for point. More importantly, Craig also copied the surrounding accompanying text which is unique to Anderton and doesn’t make sense for the majority of the other images Wright has now dug up.
Fact is, there is nothing “general” about the figures he took. They are undeniably taken from Anderton and Carter (2009).
