Anatomy of a fraud — A deep dive into one of Craig Wright’s plagiarized papers
I want to focus on one of the newer instances of plagiarism first brought to light earlier this month by Zectro. This particular instance is more remarkable because Wright just presented this paper at a conference, and it’s been published by Springer. Although both the conference organizers and publisher had been notified weeks ago, neither has acted yet. At the end of the article, I will provide some contact information for anyone who would like to hold them accountable.
Here is a pre-print version of Wright’s paper, “A Proof of Turing Completeness in Bitcoin Script”. The differences from the final version are minimal.
To get right to it, almost the entire Wright paper was plagiarized from a 1964 paper by Corrado Böhm. The following picture shows Wright’s complete, published paper with the plagiarized parts highlighted.
The source paper is not cited at all, not that it could possibly forgive all this copying, even if he did cite it. The Böhm source paper cites a previous paper Böhm wrote, which bears no resemblance to the one in question, and Wright copied that citation into his paper (along with every other citation from the Böhm source paper).
The plagiarized source is completely contained within Wright’s paper. That is, he included the entire thing (with minor rewording) and only added brief (wholly nonsensical) asides about how it’s relevant to Bitcoin Script.
In the process, Wright also introduced several errors that were not present in the original.
Why would he do this?
Let’s dig in to the details.
One of the first times Wright was in the spotlight was because of a 2015 conference where he claimed that Bitcoin was Turing Complete. While this was and continues to be false, he has since modified the claim to say that Bitcoin Script is a Turing Complete language, which the paper in question purports to prove. Of note, he previously published a different paper to prove the same thing, and it was also plagiarized.
Since 2015, he’s tried in vain to prove that Script is Turing Complete. This is doomed to be a Sisyphean task, as it’s ridiculous on its face. At minimum (a necessary but insufficient condition), a Turing Complete language must be able to enter an infinite loop. Bitcoin Script is not capable of that, so it cannot be Turing Complete.
Nevertheless, Wright continues the charade. His latest attempt, this paper, is a hilarious demonstration of a fraud’s attempt to deceive.
Why plagiarize this particular source paper?
Wright likely chose this paper to copy because its author is deceased, it is not copyrighted, and it’s not available (in full) online.
Maybe it’s still valid even if it’s plagiarized?
No, it’s not. Very much not.
To be clear, the source paper has no relation whatsoever to Bitcoin Script. It describes how a particular class of Turing Machines is universal. Wright merely waves his hands to try to explain how Bitcoin Script is relevant:
This process is simulated using the primary Bitcoin script stack coupled with the use of the Alt stack as a dual counter machine. It would be sufficient to demonstrate that the use of a dual stack system, which is in effect, a multi-tape 2-symbol Post-Turing machine with its behavior restricted so its tapes act like simple “counters”.
This, of course, is absolute bullshit. Script could have a thousand stacks and still not be Turing Complete. It’s not the number of stacks; it’s that Script lacks the ability to boundlessly loop! The language the source paper describes has the ability to boundlessly loop, of course.
That doesn’t stop Wright from simply inserting references to Bitcoin Script in various places where it doesn’t belong:
Here is the original:
Another hilarious example from Wright’s paper:
Mathematically, what Wright did with this paper is basically the equivalent of copying a proof that the square root of 2 is irrational, and using it to say, “and so is the number 3.5”.
One strange thing I noticed was that Wright inserted some completely out-of-place tirade on security right in the middle of Böhm’s proof:
No Wright plagiarism would be complete without his introduction of errors not present in the original. Thanks again to Jorge Stolfi for pointing some of these out.
Wright screwed up the very first figure. Here is Wright’s version:
He apparently simply didn’t understand Böhm’s branching notation:
Next, we have Wright’s attempt:
to copy the original:
There’s much more.
If you noticed that Wright omitted a tiny part of that section (the part that begins ‘(or over…’), don’t fear. He added that part in a footnote:
Wright attempted to copy the discussion text, but made a grammatical mess of things:
Here’s what he was trying to say:
There are many more examples, but I’ll just do one more. This is one of my favorites:
There are a few things to note. First is the strange term ‘altering machines’. Let’s see what that’s about:
Ooooh, “all Turing machines”, not “altering machines”. He also chooses a different character (theta), likely because he couldn’t find the original on his computer. Finally, he seems to have inserted his own sentence! Highlighted in light blue: “Another way of stating this is that…”
Oh, wait, he just copied the footnote:
Feel free to leave additional examples in the comments. There are plenty left.
Though these examples are funny, let’s remember that Wright committed a serious academic crime, one he himself calls “criminal fraud”:
Worse, perhaps, is that conference organizers and publishers seem uninterested in policing themselves.
If anyone is interested in contacting them, here is the contact information for:
The IntelliSys Conference (email@example.com)
Springer Nature (firstname.lastname@example.org)
The latter two are included because Wright ‘wrote’ the paper apparently under their auspices:
Late edit: it appears that Craig Wright’s CNAM thesis title is: Bitcoin: SEIR-C propagation models of block and transaction dissemination. This happens to be identical to the title of the ‘paper’ from my fourth example of his plagiarism. CNAM ought to be very interested in that.