Court Reveals Craig Wright's Bitcoin White Paper LaTeX Files as Recent Forgeries

(ii) Dr Wright’s “electronic cash” LaTeX file “E-Cash-main.tex”, “Candidate C” {H/331/1}

58. This file is a LaTeX source file which is very similar to main.tex. This file is Candidate C in Mr Rosendahl’s report. The file presents as very similar to main.tex, but it is entitled “Electronic Cash without a trusted third party”, the draft title for the Bitcoin White Paper which was communicated by Satoshi Nakamoto to Adam Back and Wei Dai in September 2008. The abstract of this paper also matches the preview abstract in those Satoshi emails.

(a) COPA’s Reasons for Alleging Forgery

59. The Bitcoin White Paper was not written in LaTeX. It was written and produced in OpenOffice 2.4. Its metadata record that it was produced by that means. Examination by both parties’ experts has led them both to conclude, and agree, that it was produced by that means based on every level of the pdf from the fine details of its typographical presentation, down to the binary digits of the PDF. [Rosendahl1 [66-68] G/7/23] [Rosendahl1 [191] G/7/58] [Lynch1 [117-120] and [122.c-d] I/5/36] [Rosendahl / Lynch1 [2] Q/5/1].

60. The encoding and embedding of fonts within the original Bitcoin White Paper, and their kerning, is fundamentally incompatible with LaTeX, especially LaTeX in 2009. [Rosendahl1 [66-68] G/7/23 and [156-157] G/7/49].

61. Dr Wright’s selected LaTeX file does not, when compiled, produce the Bitcoin White Paper (and neither does any other file in the TC folder). It compiles to a PDF which only superficially resembles the Bitcoin White Paper but which is in fact substantially different [Rosendahl1 [120 G/7/43] [Lynch1 [105] I/5/29] [Rosendahl / Lynch Joint report [3] Q/5/1]. Further, and for the avoidance of doubt, the differences are not even explained by the belated indications given by Dr Wright’s solicitors on 29 December 2023 M2/763/1 that there were certain reasons why Dr Wright’s file would not compile to a replica of the Bitcoin White Paper (contrary to his previous statement on oath).

62. Dr Wright’s LaTeX files from the TC folder are not original to 2008 or 2009. Using LaTeX software that is up to date for 2009, none of them compiles properly, instead producing errors and warnings. The files L and C could not be compiled at all. In order to compile them at all, it is necessary to fix the files or use a much more modern version of LaTeX software [Rosendahl1 [106-112] G/7/38]. This is because Dr Wright’s LaTeX files include software packages and options which could not have been used in 2009 [Rosendahl1 [158] G/7/50].

63. The images that result from compiling Dr Wright’s identified file are different from the images in the original Bitcoin White Paper (as would be the images resulting from compiling any of Dr Wright’s LaTeX files), including in the precise location and angle of the lines and arrows. Dr Wright relies on the precise location and angle of lines and arrows being the same [Wright 6 E/21/3; Field1 [27] P3/13/10], but they are not the same and exhibit subtle but important differences [Rosendahl 1 [98-100] G/7/35 and [181] G/7/55] [Lynch1 [105-115] I/5/29] [Rosendahl / Lynch1 [3] Q/5/1]. In one figure, even the label text does not match [Rosendahl1 [98] and fig 3.4 G/7/36].

64. Dr Wright’s LaTeX files are recent creations:

64.1. The files are modern files written in modern LaTeX code. Paragraph 62 above is repeated.

64.2. Dr Wright’s Overleaf account, from which they were sourced, was created on 8 August 2023. There is no previous record of the existence of these files [Shoosmiths’ first letter of 8 January 2024, [6] M2/820/2].

64.3. When the Overleaf Metadata was produced, it showed that:

64.3.1. Dr Wright had first created main.tex on 19 November 2023 as a blank document. Dr Wright had then edited the document main.tex over a course of at least 211 edits in the period between 19 November 2023 (2023-11-19 at 18:23:52.252 UTC, the date the document was created) and 1 December 2023 (2023-12-01 at 14:01:48.428 UTC, the date that documents were provided to and compiled by Stroz Friedberg). Dr Wright had then engaged in over 150 further edits in the following days up to 12 December 2023 (2023-12-12 at 15:02:45.176UTC).

64.3.2. The file E-Cash-main.tex had been created at 17.09 on 24 November 2023 (2023-11-24 at 17:09:43.829UTC) and had been populated 11 seconds later with over 700 lines of code. The document had then been edited in a series of 7 further edits up to 17.12 (2023-11-24 at 17:12:15.559 UTC).

64.4. The files were not disclosed at the time for initial disclosure, nor extended disclosure. Further, the files were not disclosed at any time in the intervening period up to 17 November 2023, during which Dr Wright provided no fewer than 12 more rounds of disclosure.

64.5. Dr Wright has sought to explain away his failure to disclose the files at the time of extended disclosure by blaming his former solicitors, Ontier. Dr Wright states that Ontier advised him that the files were not disclosable and could not be disclosed [Wright6 E/21/3; Field1 [27 P3/13/10]. However, this account is inherently implausible (presupposing bizarre and negligent advice), and Ontier have confirmed that Dr Wright’s account is false in each and every particular. [Shoosmiths’ letter of 18 December 2023 [[4]].

64.6. Despite the order of the Court to do so, Dr Wright has failed to produce any written record of such advice. Given the supposed importance of the material, it is to be inferred that the issue of its disclosure and Dr Wright’s ability to rely upon it would have been addressed in writing and that Dr Wright is unable to provide such a written record because such advice was never given.

64.7. It is to be inferred that Dr Wright’s LaTeX files were not disclosed at that time because they did not yet exist at that time.

65. Contrary to Dr Wright’s assertions, it is not difficult to reverse engineer the publicly available Bitcoin White Paper to create a LaTeX source file that compiles to a PDF file similar to Dr Wright’s identified file and that is superficially similar to the Bitcoin White Paper. This is agreed by the parties’ experts. [Rosendahl / Lynch1 [4] Q/5/2]

66. There are strong indications that the images in Dr Wright’s LaTeX files have been automatically generated, using an online tool called Aspose PDF to LaTeX. The result of converting the original Bitcoin White Paper to LaTeX using Aspose is an exact match, with nearly identical low-level coding. While being long and verbose and not simple to create by hand in LaTeX, the code can be created by Aspose instantly. [Rosendahl1 [197] G/7/60].

67. The code which causes Dr Wright’s LaTeX files to resemble the original Bitcoin White Paper in their textual output calls on packages and options which did not exist in 2009. These include instructions for: setting the font (with fontspec); how references are displayed (with the hidelinks option); the font of the letter λ (with unicode-math); and the drawing of diagrams (with the package TikZ, coded to load the TikZ library arrows.meta). They include the use of Lua code (requiring the package luacode with LuaLaTeX). [Rosendahl1 [section 3.7], G/7/43-50].

68. As well as the textual content of the output PDF from Dr Wright’s LaTeX files, also the location of elements on the pages cannot be reproduced without modern code. The code calls again on software packages and options which did not yet exist in 2009, including to set the geometry of the page and margins (in the case of main.tex, the command \newgeometry), the location of images on the page (with the command \AddToShipoutPictureBG*), the location display of margins. [Rosendahl1 [section 3.7], G/7/43-50].

69. In each case, those packages and options provide the very functionality which causes Dr Wright’s LaTeX files to output something resembling the original Bitcoin White Paper. If they were not used, the output PDF from Dr Wright’s LaTeX files would not resemble the original Bitcoin White Paper at all. [Rosendahl1 [158], G/7/50] [Rosendahl / Lynch1 [5] Q/5/2].

70. In each case, by contrast, each of the observed characteristics of the Bitcoin White Paper were supported natively by OpenOffice in 2008-2009 [Rosendahl1 [160], G/7/50] [Lynch1 [116-120] I/5/33].

71. In respect of E-Cash-main.tex (file C), the file cannot be compiled at all on a 2008-2009 version of LaTeX software, even with concerted effort to make the file compatible. [Rosendahl1 [106-109] G/7/38].

72. Dr Wright has stated that these files should be compiled on Overleaf, a modern system which did not exist in 2008-2009.

(b) COPA’s Reasons for Inferring Dr Wright’s Knowledge / Responsibility

73. The effect of creating these documents is to support Dr Wright’s story that the Bitcoin White Paper was created in LaTeX. The story is untrue, and is designed to be supportive of Dr Wright’s claim to be Satoshi Nakamoto (suggesting both special knowledge of the production of the paper and possession of material which only Satoshi could supposedly possess), contrary to fact.

74. E-Cash-main.tex (file C) is created by copying from Satoshi Nakamoto’s draft abstract (sent by email to Adam Back and Wei Dai in a form available to Dr Wright) in reference to a document that was never published. The effect of creating this document is to support Dr Wright’s claim to be Satoshi Nakamoto by suggesting that he is in sole possession of the source of that unpublished document, contrary to fact.

75. Dr Wright has admitted to having created these files and being in sole possession of these files since they were created. Further, Dr Wright has averred that he has created and been in sole possession of these files since their creation, claiming that the “mere possession of these files is evidence of authorship of the White Paper” [Wright 6 E/21/3; Field 1 [33], P3/13/11], and that they are “unique”. It is true only that the files are unique to Dr Wright. His possession and deployment of these files is indicative of a willingness to forge documents to be used as evidence before this Court.

76. On 30 September 2023, Christen Ager-Hanssen tweeted a screenshot indicating that Dr Wright’s browsing history showed that he had searched Google leading to the web page “Was anything in Satoshi Nakamoto’s original Bitcoin paper compiled in LaTeX?” L20/195/1 [Sherrell19 [65-66] P1/18/24]. Dr Wright has stated that the screenshots of his browsing history posted in tweets by Mr Ager-Hanssen were obtained from his (Dr Wright’s) laptop computer [Wright3 [18-19] E/3/6].

77. These documents were then created during the course of the present proceedings using current technology: paragraphs 63, 64, and 68 above are repeated. Dr Wright has created numerous other documents during the course of proceedings, as is a matter of agreement between all four forensic experts (including Dr Wright’s own instructed experts).

78. Dr Wright has stated in clear terms that “all graphical images in the White Paper were produced using LaTeX code” [Wright4 [6.c.i] E/4/5]. However, the graphical images in the White Paper were not produced by LaTeX code. Paragraphs 62 to 64 above are repeated.

79. When ordered by this Court to describe the software environment required to compile the document, Dr Wright did not simply state what software is required. Instead, he provided a statement (Wright8) that was verbose and vague. It is to be inferred that the purpose of the statement was to obscure relevant technical detail and supply irrelevant, technical-sounding information in its place and/or to provide scope for Dr Wright to give excuses for failures of his LaTeX files to compile into a true replica of the Bitcoin White Paper.

80. The files were taken from Dr Wright’s own LaTeX editing account, hosted on Overleaf. Dr Wright claimed that no metadata existed in relation to the files hosted on that account, but the claim was untrue and metadata does exist and had been accessible to Dr Wright prior to the time that he made that claim. Further, when ordered to provide access to the metadata, account history, and history of editing those files on Overleaf, Dr Wright did not disclose that data until compelled by further application to the Court. Yet further, while refusing to provide the metadata associated with his Overleaf account, Dr Wright has sought to alter and embellish his story in respect of the number and type of accounts hosted on Overleaf and the supposed method of transmission of the LaTeX files.

81. The metadata, as summarised in paragraph 64.3 above, directly implicate Dr Wright as being responsible for the recent creation of the Overleaf LaTeX files as forgeries.

82. Dr Wright has sought falsely to blame his former advisors for his failure to disclose these supposedly important documents at an earlier stage:

82.1. The story provided by Dr Wright in respect of failings by Dr Wright’s former solicitors, Ontier, is false. Paragraph 64.5 above is repeated.

82.2. The story provided by Dr Wright in respect of failings by Dr Wright’s former edisclosure providers, Alix Partners, is unsubstantiated and was not foreshadowed. When attempts were made to investigate the matter with Alix Partners, Dr Wright sought to frustrate those efforts [Macfarlanes’ letter to Alix Partners dated 15 January 2024; Shoosmiths’ letter to Alix Partners dated 16 January 2024]. Further, when an application was made to enable Alix Partners to provide that information, Dr Wright opposed the application. Yet further, when the Court made the order sought so as to enable the required information to be provided by Alix Partners with Dr Wright’s consent, Dr Wright resisted that order.

83. It was on the basis of these documents that Dr Wright attempted to delay the trial, by offering the Court a fait accompli in respect of a trial adjournment on grounds of fairness [Wright6 E/21/3; Field1 [34] P3/13/11].

(c) Dr Wright’s Explanations and COPA’s Rebuttal

84. Dr Wright sought to explain the findings indicating forgery by repeating his claim to have created the Bitcoin White Paper in LATEX, but adding that he had used MikTeX (which Mr Rosendahl had not used). He also claimed that the way in which LATEX works (i.e. how it compiles) on Overleaf is different from the versions of LATEX software he was using back in 2008/09, hence why they now appeared differently. He also claimed that the versions in Overleaf that were disclosed had been altered due to (1) him making a number of typographical corrections to the published version of the Bitcoin White Paper and (2) various demonstrations carried out for the benefit of his lawyers. He also said that he had made very extensive modifications to the LATEX code so as to allow his version of the code to create the Bitcoin White Paper in the format published. He explained away the unusual nature of his code as being a ‘digital watermark’. See Field1 at {E/24/6}, Shoosmiths’ letter of 13 December 2023 {AB-A/2/67} and {Day5/121:10} and following.

85. COPA submitted that this explanation should be rejected as dishonest for the following reasons:

85.1. Despite having given numerous public accounts, including under oath in other proceedings and in his earlier witness statements in this case, Dr Wright had never mentioned that he had used LATEX to create the Bitcoin White Paper until his fourth statement dated 23 October 2023 {E/4/5} ([6c(i)]) (and even there the reference was elliptical). Given the emphasis which he now places on the use of LATEX (in particular in his evidence relied upon to adjourn the trial at the PTR – Field1 [19.2] {E/24/7}), it is implausible that he did not mention it earlier.

85.2. The Bitcoin White Paper was not written in LATEX, but in OpenOffice 2.4 as agreed by both Rosendahl and Lynch in their Joint Report {Q/5/1}.

85.3. Most of the disclosed LATEX files in the TC folder would not compile at all using a 2008/09 version of LATEX. Mr Rosendahl found that no less that 14 software packages referenced in Dr Wright’s LATEX files could not have been used in 2009. Rosendahl {G/7/38} and {G/7/50}.

85.4. To do as Dr Wright claimed, and make the extensive modifications to allow the creation of the Bitcoin White Paper in LATEX, would involve spending months creating the necessary software and then further weeks just to make the Times New Roman font. Mr Rosendahl noted that there were only a few people in the world with the knowledge and skills to do this work: Rosendahl1 {G/7/24}; {Day17/33:14} to {Day17/35:16}.

85.5. The Bitcoin White Paper, in contrast, can be produced in Open Office in a couple of hours for the text and around a day’s work including the diagrams. Rosendahl {Day17/33:24} to {Day17/34:6}.

85.6. Dr Wright never properly explained the nature of the digital watermark he claimed to have embedded in the Bitcoin White Paper. He claimed that the ‘digital watermark’ was the unusual way in which he had written the LATEX code, but he could not point to any older version containing that code, as he had to admit to having edited the code in the Overleaf folder once he was caught red handed doing that.

85.7. None of the LATEX versions of the Bitcoin White Paper in Overleaf, including Candidates L and C, compile into a replica of the actual Bitcoin White Paper. There are differences in the bibliography, spacing, formula, symbols, punctuation and other content differences: see Rosendahl1{G/7/36}

85.8. Dr Wright admitted to having carried out a web search in September 2023 asking if Satoshi had written the Bitcoin White Paper in LATEX: {Day5/122:20} and following.

85.9. The evidence of Mr Rosendahl (agreed by Mr Lynch in the joint statement) was that the Bitcoin White Paper could be replicated using software such as Aspose, and that reverse engineering the Bitcoin White Paper to create a LATEX source file that outputs something superficially similar is “not too difficult” – see Joint Report of Rosendahl and Lynch {Q/5/2}. It appears that Dr Wright did indeed use Aspose but failed to make all the corrections to the code that it produced to camouflage that use: {Day15/203:5} and following.

85.10. The Overleaf metadata shows Dr Wright actually forging the LATEX documents in real time, unaware that every edit was being recorded. The animation shown to the Court by the Developers {L21/13} shows that process and how he moved through the document fixing spacing and making changes line by line.

(d) Conclusion

86. In the main body of the Judgment in [426]-[558] I have addressed a detailed set of submissions made by the Developers which overlap with many of the points relied upon by COPA, as set out above. The findings I made in [426]-[558], in conjunction with COPA’s submissions set out above, demonstrate that the evidence was overwhelming in showing that the LATEX documents were created by Dr Wright in September 2023 by way of his reaction to Madden1. Madden1 taught him the pitfalls of documents containing metadata. So he pivoted to a set of documents which (he thought) either contained no metadata or much less than the documents from his original disclosure. These included the LATEX documents.

87. In fact, contrary to his expectation, Dr Wright’s Overleaf account did store metadata. As I stated in the main Judgment at [480], I agree with the Developer’s submission that their animation {L21/13} shows at least part of his forgery in real time.

88. Accordingly, I am in no doubt that all of Dr Wright’s LATEX documents are recent forgeries created by Dr Wright. This finding is also reinforced by the analysis of various particular LATEX documents in their own sections below.