1
00:00:05,140 --> 00:00:08,920
Well, good afternoon, everyone, and thank you so much for coming. This is very special.

2
00:00:09,760 --> 00:00:15,219
It's not the first time I've introduced a lecture, but I'm introducing it. The first what we think is a new cycle of a lectures.

3
00:00:15,220 --> 00:00:18,340
So that's a very special moment for the Voltaire Foundation.

4
00:00:18,790 --> 00:00:25,959
We've had for 25 years or more. We've had an annual lecture called The Best one lecture that I know wrong with all us here.

5
00:00:25,960 --> 00:00:31,120
But, um, where we invite enlightenment historians and literary critics to come and give a talk.

6
00:00:31,630 --> 00:00:38,530
But in recent years, for lots of reasons that we know about, the Voltaire Foundation has been moving more and more towards digital research.

7
00:00:39,250 --> 00:00:45,639
Um, when I first became director, we, I was given to, to sell on my first day in the office, a CD-Rom.

8
00:00:45,640 --> 00:00:50,770
That's a bit historic, isn't it? And then we produced a database in the early 2000s, the electronic enlightenment.

9
00:00:51,090 --> 00:00:57,430
Now we're working on digitising from scratch in a since the 201 volumes of the complete works of old.

10
00:00:57,640 --> 00:01:04,450
So there's always been although we've been primarily a paper publisher until recently, we've always been doing digital research.

11
00:01:05,140 --> 00:01:10,570
And over the last few years, particularly working with Glenn, the digital research has become terribly important.

12
00:01:10,570 --> 00:01:16,300
As we've accumulated more data, there's been more opportunities and more excitement in exploiting that data and thinking about how to use it.

13
00:01:16,690 --> 00:01:24,280
So it seemed in the in the evolution of the Voltaire Foundation, this is a good moment for us to sort of make the point by having an annual lecture.

14
00:01:24,640 --> 00:01:32,950
And, um, last year we launched a new online open access journal, Digital Enlightenment Studies.

15
00:01:34,020 --> 00:01:41,819
I'm, I can't say take a copy with you because it's online, but you can take with you a postcard, a beautiful design postcard, uh, outside.

16
00:01:41,820 --> 00:01:44,010
So please have a look. It's a fantastic journal.

17
00:01:44,010 --> 00:01:52,530
It's the it's establishing a whole area of research, and it contains wonderful articles, including one by Nico and his team in Helsinki were here.

18
00:01:52,800 --> 00:01:56,490
And if you're inspired to send us an article, we'd like that too.

19
00:01:56,610 --> 00:02:02,939
So, um, we've begun by, uh, founding this new online journal, Open Access.

20
00:02:02,940 --> 00:02:11,700
It looks beautiful. And we're now in, in the same spirit, beginning a new series of, uh, annual lectures.

21
00:02:12,150 --> 00:02:17,100
Um, it's a huge pleasure to have, Glenn to give this first lecture in the series.

22
00:02:17,610 --> 00:02:24,299
Um, I think we will. We've met probably before, but the first time we met and worked together for a long time was when you were Mellon,

23
00:02:24,300 --> 00:02:31,470
um, fellow in digital humanities, uh, funded, funded by the Mellon Foundation in 2011 to 2013.

24
00:02:31,890 --> 00:02:39,990
And although you then left us and you went to the A and you were you were at the centre for research and you, of course, before that.

25
00:02:39,990 --> 00:02:47,550
And since then you've been working in Chicago on the Art Artful project, producing amazing resources, and not least the Encyclopédie.

26
00:02:48,090 --> 00:02:54,030
Um, and since 2018, Glenn has been professor of D.H. in the French department at the Sorbonne.

27
00:02:54,810 --> 00:02:57,990
Um, but in all that time, you've also had a connection with the.

28
00:02:57,990 --> 00:03:00,180
But we've maintained the connection with the Voltaire Foundation.

29
00:03:00,810 --> 00:03:07,140
And, uh, you're currently, since 2021, the Astra Foundation research fellow in digital humanities.

30
00:03:08,100 --> 00:03:13,440
And I also want to say a little bit of celebrating the way that the Voltaire Foundation is moving into digital research.

31
00:03:14,130 --> 00:03:20,490
I do want to pay particular thanks to the Astra Foundation, um, who's represented by Elisabet.

32
00:03:20,850 --> 00:03:28,350
The also Foundation three, four years ago made us an extraordinary grant which has enabled us to really establish ourselves in this area.

33
00:03:28,740 --> 00:03:34,379
And this is a really transformational grant which has enabled us first to establish ourselves

34
00:03:34,380 --> 00:03:39,360
in the area of digital research and to employ lots of younger researchers and younger people.

35
00:03:39,600 --> 00:03:46,780
So it's played a really key part in helping us develop in that direction and encourage young researchers and scholars in Oxford.

36
00:03:46,800 --> 00:03:50,880
We're incredibly grateful. Um, some of that was such as a hit, some lights,

37
00:03:51,270 --> 00:04:01,320
and Glen has been the Astra Foundation research fellow and has been a key player in helping us directs that, um, that new direction of AI research.

38
00:04:01,770 --> 00:04:08,700
Um, I also should mention Glenn's publications. Um, he published his thesis interest on the Literary Theory of Peggy, published by Oxford.

39
00:04:09,090 --> 00:04:14,520
Um, I know that you were just a specialist for humanities research on political vocabularies.

40
00:04:14,880 --> 00:04:20,940
Um, we work together on a book on Voltaire's correspondence and readings of quotation in Voltaire's correspondence.

41
00:04:21,390 --> 00:04:27,930
Uh, this is a lovely digitising. Had to stick with Glenn Row, you know, monograph series published by Oxford.

42
00:04:28,380 --> 00:04:33,030
Um, I'm here to advertise. Obviously. Um, we have copies on sale afterwards.

43
00:04:33,300 --> 00:04:38,250
Um, and recently, I think Glenn's published a book called Observability of Earth, which we published.

44
00:04:38,250 --> 00:04:46,890
We've done it with digital Zoom. So there's a hugely impressive list of publications, not to mention immensely distinguished articles.

45
00:04:47,340 --> 00:04:50,760
Glenn and his colleagues from Chicago over here,

46
00:04:50,760 --> 00:05:00,120
Robert Silk Lucas here of have done really pioneering work in thinking about data reuse and citation strategy and the Oscar party and the beast.

47
00:05:00,750 --> 00:05:04,970
Um, and some of those things I think may come back into this lecture today.

48
00:05:05,010 --> 00:05:12,659
I'm not sure I can get us clear what he's going to talk about. But crucially, Glenn's also just pulled off the the magic for everybody's session.

49
00:05:12,660 --> 00:05:18,240
And, you know, so you consolidate a grant I think in year two which has a great title and a great logo.

50
00:05:18,510 --> 00:05:26,220
Modelling enlightenment is modern. I should have said modern modelling enlightenment, reassembling networks of modernity through data driven research.

51
00:05:26,910 --> 00:05:34,020
So the idea here is to use big data and to use that to shape that with thought experiment to to try and remap.

52
00:05:34,110 --> 00:05:40,790
I think, well, I'm not Glenn, but I think the attempt will be to remap the way we can eventually think about French Enlightenment,

53
00:05:40,800 --> 00:05:46,260
French Enlightenment writing. Um, I'm not sure how much today's lecture will draw on that.

54
00:05:46,260 --> 00:05:52,980
We're going to see, but you have a great title, Poetics of Text Reuse Digital Intertextuality in the 18th Century archive.

55
00:05:53,250 --> 00:06:03,350
Glenn is a fantastic pleasure to have you. Thank you very much. But I thank you so much, Nicholas.

56
00:06:03,350 --> 00:06:04,729
And thank you all to to coming.

57
00:06:04,730 --> 00:06:11,420
It's it's a really, really genuine and great pleasure to be here today to to to give this first annual lecture and digital,

58
00:06:11,780 --> 00:06:15,319
uh, enlightenment studies uh, which is, as Nicholas just mentioned,

59
00:06:15,320 --> 00:06:19,060
a field in which we're heavily invested, uh, in, in that collectively at the vault,

60
00:06:19,070 --> 00:06:25,490
our foundation in Chicago, thanks to the Astra Foundation, uh, at the Sorbonne, uh, sort of a bit all over.

61
00:06:25,790 --> 00:06:31,879
Uh, we're committed to, to to promoting, uh, this new field of research and to present to the world,

62
00:06:31,880 --> 00:06:37,520
both in the form of this public lecture, but also the new open access, the journal that Nicholas mentioned.

63
00:06:38,090 --> 00:06:45,860
The reason that so for both of these initiatives, uh, is the realisation that that this field of study, uh,

64
00:06:45,860 --> 00:06:52,100
digital approaches to the 18th century in general has reached a certain maturity and digital collections,

65
00:06:52,130 --> 00:06:59,000
uh, a certain critical mass, uh, which warrants and welcomes new scholarship and forms of academic outreach.

66
00:06:59,720 --> 00:07:06,350
Uh, this is not to say that digital, uh, research has fully and finally been accepted by the humanities mainstream,

67
00:07:06,470 --> 00:07:10,940
uh, but simply that we're perhaps on a better footing today than at any time in the past two decades.

68
00:07:11,720 --> 00:07:18,020
Uh, certain barriers remain, uh, not just for digital 18th century studies, but for the wider digital humanities world,

69
00:07:18,110 --> 00:07:23,510
uh, in general, who are still in some venues, uh, kept at a disciplinary distance.

70
00:07:24,320 --> 00:07:28,820
It is often said, for example, that, uh, D.H. lacks sufficient theoretical grounding.

71
00:07:29,750 --> 00:07:33,230
It's just the new positivism and all praxis, no theory.

72
00:07:33,740 --> 00:07:39,590
Uh, and this is somewhat true if you attack literary and historical questions from a purely computational perspective,

73
00:07:40,070 --> 00:07:47,570
uh, one text is the same as another, whether it's a novel or a grocery list once reduced to ones and zeros.

74
00:07:48,200 --> 00:07:52,939
Uh, if, however, uh, you start from a theoretical framework such as intertextuality,

75
00:07:52,940 --> 00:07:58,159
as we'll see today, or the idea of modelling as an epistemological exercise to,

76
00:07:58,160 --> 00:08:03,770
uh, follow Willard McCarty, then the conversation between human and machine becomes all the more engaging.

77
00:08:04,280 --> 00:08:08,510
And this is especially true in our current age of so-called artificial intelligence,

78
00:08:09,080 --> 00:08:17,600
where large language models capable of reading whole universes of text and generating humanlike responses based on an almost infinite archive,

79
00:08:18,170 --> 00:08:22,610
and are in dire need of theoretical, not to mention ethical, grounding from the humanities.

80
00:08:23,150 --> 00:08:26,810
These models are the fever dream, I would argue, of post-structuralism,

81
00:08:27,770 --> 00:08:34,850
and we need scholars conversant in theory with a capital T to understand these new artificial text realities.

82
00:08:35,770 --> 00:08:41,680
What I'd like to present to you today, then, is perhaps surprisingly, much less technical and much more theoretical.

83
00:08:41,920 --> 00:08:46,090
Uh, in terms of scope, uh, and this is quite intentional on my part, uh,

84
00:08:46,090 --> 00:08:53,170
in a modest effort to demonstrate that thinking with and thinking through computational methods and approaches

85
00:08:53,740 --> 00:09:00,219
to literary historical data can lead to real insights and uncover the rich complexity of humanistic,

86
00:09:00,220 --> 00:09:05,530
generative intelligences, uh, that are all too often hidden in our ever growing digital archives.

87
00:09:06,130 --> 00:09:09,250
So one such computational approach that we've been, uh,

88
00:09:09,250 --> 00:09:14,320
using for years at the University of Chicago and elsewhere is the automatic detection of text reuse.

89
00:09:15,100 --> 00:09:20,919
And this is a as I said, an old, old, old in computational terms goes back to 2002,

90
00:09:20,920 --> 00:09:27,489
at least in its first official, uh, naming of text to use, uh, and it's quite, uh, prosaic in its definition.

91
00:09:27,490 --> 00:09:31,930
It's the reuse of existing written sources in the creation of a new text.

92
00:09:32,210 --> 00:09:37,420
Uh, and in this paper, they talk about a sort of similarity spectrum where you have information retrieval,

93
00:09:37,420 --> 00:09:43,090
which is the classic, uh, document retrieval methods that go back to the 60s and 70s,

94
00:09:43,120 --> 00:09:47,199
uh, bag of words models, if that's if that means anything to you, uh,

95
00:09:47,200 --> 00:09:52,780
where we think about how documents are related to each other in a sort of semantic similarity space.

96
00:09:53,260 --> 00:09:56,140
Text reuse is somewhere in the middle between the two,

97
00:09:56,350 --> 00:10:01,570
and then on the right is the near duplicate detection, which we know and love as plagiarism detection.

98
00:10:01,990 --> 00:10:10,060
Uh, in all of its glory. To think about this, theoretically, we might say that text reuse is a form of intertextuality.

99
00:10:10,750 --> 00:10:15,330
Uh, and you could, which is a theory that comes out of the 60s and 70s in France.

100
00:10:15,340 --> 00:10:21,980
Uh, Julia Kristeva is the first to use the term, uh. She uses it as a translation of Maxine's notion of dialogue ism.

101
00:10:22,060 --> 00:10:25,270
Uh, and then it sort of takes off in the telco circle from there.

102
00:10:25,660 --> 00:10:31,360
Uh, you could have a sort of conservative approach to intertextuality, which would be the Harold Bloom School of influence,

103
00:10:31,870 --> 00:10:36,729
uh, where the, uh, criticism of the art of knowing the hidden roads that go from poem to poem,

104
00:10:36,730 --> 00:10:42,280
where poetic geniuses sort of have to grapple with the great giants of the past,

105
00:10:42,280 --> 00:10:49,210
and they necessarily borrow from these, uh, from these, uh, predecessors to assert their own poetic authority.

106
00:10:49,810 --> 00:10:53,799
More radically, uh, is the Berkeley and School of of intertextuality,

107
00:10:53,800 --> 00:10:58,750
where every text becomes the sort of inter texts where other texts are always present in it at various levels,

108
00:10:59,110 --> 00:11:04,060
in more or less recognisable forms, texts from the previous culture and those from the surrounding culture.

109
00:11:04,270 --> 00:11:08,890
Every text then becomes the new fabric or tissue of quotations from the past.

110
00:11:09,490 --> 00:11:15,140
Uh, between these two. I wouldn't say our position is between these two views.

111
00:11:15,470 --> 00:11:24,080
I'd take a more middle road. Not surprisingly either more Ginetta in in my in my take of of intertextuality engineered to in his

112
00:11:24,080 --> 00:11:29,240
great work on palimpsest two defines it as a relationship of co-presence between two or more texts.

113
00:11:29,720 --> 00:11:36,290
That is to say, identically and most often by the actual presence of one text in another in its most explicit and literal form.

114
00:11:37,130 --> 00:11:40,910
It is the traditional practice of citation, which I now put in the middle of our spectrum,

115
00:11:41,840 --> 00:11:45,140
with or without inverted commas, with or without precise reference,

116
00:11:45,620 --> 00:11:51,830
when the less explicit and less canonical form that of plagiarism, uh, which is an undeclared but still literal borrowing.

117
00:11:52,310 --> 00:11:55,580
Clearly this today we think of text reuse a bit.

118
00:11:56,210 --> 00:12:01,490
It's a bit sad, actually, as a form of plagiarism and almost exclusively as a form of plagiarism.

119
00:12:01,940 --> 00:12:08,270
And this is because we have this notion that words belong to certain authors and can't be reused without their authorisation.

120
00:12:08,670 --> 00:12:10,909
Uh, this is played out recently in academia.

121
00:12:10,910 --> 00:12:21,290
And if you follow the poor plight of Claudine Gay, the first African-American woman president of Harvard who was fired, uh, ostensibly for plagiarism.

122
00:12:21,320 --> 00:12:28,580
Uh, this article in the Guardian tells you exactly what a what a broad section this notion of plagiarism is.

123
00:12:29,180 --> 00:12:33,200
And a certain Barbara Glatt, who was responsible for the turn it in software.

124
00:12:33,200 --> 00:12:39,229
If you've ever used it, you know about it can be direct copying or something word for word without attribution.

125
00:12:39,230 --> 00:12:42,290
Indirect the wholesale theft of ideas. Mosaic.

126
00:12:42,290 --> 00:12:49,099
And this is a term that will come back today. Hold it in your mind, uh, changing some words while copying others, or even an honest mistake,

127
00:12:49,100 --> 00:12:55,190
an error of omission and execution all things gay was accused to have done, even as she continues to stand by her scholarship.

128
00:12:55,190 --> 00:12:58,790
These are all things I think we all do as, uh, as scholars.

129
00:12:59,180 --> 00:13:05,750
Uh, and it's a it's once you move into the legalistic realm of plagiarism, that text reuse becomes a problematic practice.

130
00:13:05,960 --> 00:13:12,260
But in the same article, if you move down, uh, the professor of uh, linguistic anthropology at Notre Dame, Susan Blum,

131
00:13:12,500 --> 00:13:18,980
says that there's a sort of continuum, that it's too cut and dry to think of plagiarism just as as it's theft of other,

132
00:13:19,310 --> 00:13:23,000
other people's words, and that somewhere between originality and complete copying,

133
00:13:23,000 --> 00:13:28,729
there's a language and culture that moves into the middle, and that it is in this language and culture.

134
00:13:28,730 --> 00:13:35,060
That term, perhaps text reuse inhabits, uh, this middle, this middle ground.

135
00:13:35,330 --> 00:13:38,090
And I would say that that's exactly the case in the 18th century,

136
00:13:38,090 --> 00:13:46,819
where they didn't yet have this notion of plagiarism as the theft of another author's word and that the culture, uh, gravitated.

137
00:13:46,820 --> 00:13:51,560
The culture of citation gravitated between these two poles of originality and imitation.

138
00:13:52,280 --> 00:13:56,210
And in a nutshell, the 18th century clearly is, uh.

139
00:13:57,480 --> 00:14:01,440
Is dominated still at the beginning of the 18th century by this classical notion of imitation,

140
00:14:01,710 --> 00:14:06,030
uh, which aesthetically was the model for, uh, hundreds of years.

141
00:14:06,660 --> 00:14:14,370
And, uh, it it's predicated on this notion of the value of a work, uh, came from its conformity to the great works of the past.

142
00:14:15,210 --> 00:14:20,280
And it was through these works that we could gain access to truth, beauty, nature, uh,

143
00:14:20,280 --> 00:14:26,370
all of these higher level ideas of which the ancients were somehow the strict equivalent.

144
00:14:26,910 --> 00:14:35,700
So, speaking of Virgil, for example, Pope remarks, uh, that perhaps he seemed above the critics law and but from nature's fountain scorn to draw.

145
00:14:36,300 --> 00:14:40,710
But went examined. On every party came nature. And Homer were he found the same.

146
00:14:44,200 --> 00:14:51,129
From the 18th century onwards and under the influence of German Romanticism, in particular literary works,

147
00:14:51,130 --> 00:14:58,230
we understand as more or less the emanation of what was unique in each individual their genius, and at the same time the post.

148
00:14:58,720 --> 00:15:05,470
The poet developed a new and immediate relationship with nature. Uh, one that didn't have to be based on representations of the ancients.

149
00:15:05,890 --> 00:15:14,860
To see this in action. We look to Racine's preface to Britannicus from from 1676, where he praised the conformity of his work to that of Tacitus.

150
00:15:15,180 --> 00:15:19,090
I copied my characters after the greatest painter of antiquity, Tacitus,

151
00:15:19,990 --> 00:15:25,600
and I was then so full of reading of this excellent historian that there was not a striking feature in my tragedy,

152
00:15:25,900 --> 00:15:29,560
for which he did not give me the idea for the model clearly of imitation.

153
00:15:29,590 --> 00:15:36,850
Compare this with just 100 years later with Rousseau, as we know in his opening volley of the confessions,

154
00:15:36,850 --> 00:15:40,840
where he takes pride in having no predecessor and indeed no successor.

155
00:15:41,340 --> 00:15:48,130
Uh, I am creating a work that is without prior example and whose execution will have no imitators.

156
00:15:48,610 --> 00:15:57,130
Never mind that generically he borrows his title from Saint Augustine, and then thematically he is very much continuing just the work of Montana.

157
00:15:57,850 --> 00:15:59,950
Uh, Montana is in fact, uh,

158
00:16:00,100 --> 00:16:07,120
key here because he demonstrates nicely for us the inherent ambiguity of a writer who is at once imbued with the tradition of imitation,

159
00:16:07,840 --> 00:16:14,470
inserting his flirt into his text and already aware of his identity as a unique and original author,

160
00:16:14,740 --> 00:16:17,880
giving himself over to the reader to turn to a new.

161
00:16:19,400 --> 00:16:24,230
This notion of originality is inseparable from, I would say,

162
00:16:24,620 --> 00:16:29,840
the idea of our idea of the formation of an idea that we can have of text for use in the 18th century.

163
00:16:30,830 --> 00:16:33,440
And this is the notion that Ronan McKay, for example,

164
00:16:33,440 --> 00:16:38,570
is demonstrated as being rather new to the 18th century and in fact tributary of the French Enlightenment.

165
00:16:38,990 --> 00:16:47,900
Uh, this notion of originality as an aesthetic category, and in fact, the word isn't used in French until the until the very end of the 17th century.

166
00:16:48,320 --> 00:16:54,470
Uh, by, uh, talking about the originality of paintings, one of a kind that can't be copied.

167
00:16:56,520 --> 00:17:03,300
Zero goes on to use it in 1759, uh, as describing someone of a very original character.

168
00:17:03,660 --> 00:17:09,470
The originality of the Baron declaration. The Danish minister to France to Sophie Valon.

169
00:17:10,950 --> 00:17:14,249
The first dictionary enters the lexicon in 1743.

170
00:17:14,250 --> 00:17:19,840
In the dictionary rival, who proudly proclaims that you won't find this word in any other dictionary.

171
00:17:20,010 --> 00:17:28,400
So the originality of the definition of originality is here proclaimed the then to be repeated, but, uh,

172
00:17:28,530 --> 00:17:34,019
a bit laconically in the dictionary Academy fonzie's as just the character of that which is original, uh,

173
00:17:34,020 --> 00:17:40,710
of people or things, the the great Encyclopédie of which there will certainly be question today, uh,

174
00:17:41,010 --> 00:17:46,799
plays with this semantic shift between original and originality, and it can be seen in these two entries.

175
00:17:46,800 --> 00:17:53,040
And this is exactly how the UN security works, uh, which seems to be, uh,

176
00:17:53,040 --> 00:17:57,779
which presents both the aesthetics of imitation that is still operative in the 18th century,

177
00:17:57,780 --> 00:18:02,000
but also presents a that's sort of almost already exhausted, uh, by this time.

178
00:18:02,010 --> 00:18:06,960
Huh. So it's in the first design or authentic, uh, instrument of something original,

179
00:18:07,230 --> 00:18:11,430
which must serve as a model or an example of something to be copied or imitated.

180
00:18:11,430 --> 00:18:16,950
Today we can hardly find any ancient title of possession, subjugation, etc. that is original there.

181
00:18:16,990 --> 00:18:22,170
They're only or copies collated from the originals. So we've spent our originality.

182
00:18:22,170 --> 00:18:26,430
And then the most important thing, of course, in the antiquity are the cross references.

183
00:18:26,430 --> 00:18:29,940
So there's a cross reference to the originality, which we think is by the.

184
00:18:31,400 --> 00:18:37,250
And he says, uh, it's a way of doing a common thing in a singular and distinguished way.

185
00:18:37,370 --> 00:18:40,550
Originality is very rare, and most people are nothing but copy of each other.

186
00:18:41,110 --> 00:18:44,300
Uh, the title of the original is given in good and bad ways.

187
00:18:44,450 --> 00:18:50,390
One can only think of Rousseau there in the sense of an original, both good and bad.

188
00:18:50,660 --> 00:19:00,800
Uh, so this, uh, this tension between imitation and originality, between copying, compiling and eventually plagiarism is central to the project.

189
00:19:01,070 --> 00:19:06,950
Uh, uh, as the editor, zero and d'Alembert were dogged from the outset by accusations of impiety,

190
00:19:06,950 --> 00:19:13,520
improper use of authority of authorities, and flat out plagiarism that on their response to their detractors.

191
00:19:13,820 --> 00:19:18,950
In the introduction editor's introduction to the third volume, where he says, by its very nature,

192
00:19:19,340 --> 00:19:22,940
therefore the Encyclopédie must contain a large number of things that are not new.

193
00:19:23,940 --> 00:19:29,520
Among the various works at the Institute. He has been accused of borrowing from other dictionaries have been singled out.

194
00:19:30,360 --> 00:19:37,979
The resemblance that is sometimes found between an article and an article in some dictionary is forced by the nature of the subject,

195
00:19:37,980 --> 00:19:43,080
especially when the article is short and consists of only a definition of a minor historical fact.

196
00:19:43,500 --> 00:19:49,649
This is so true that most dictionaries resemble each other in a large number of articles, because they simply cannot do otherwise.

197
00:19:49,650 --> 00:19:52,770
So we are talking here about how you construct a dictionary, how you build it,

198
00:19:53,130 --> 00:19:57,030
and typically practically how you build it is you started with the list of headwinds,

199
00:19:57,240 --> 00:20:00,390
and you inherited this list of headwinds from other dictionaries.

200
00:20:00,690 --> 00:20:04,440
And it was understood that if an older dictionary had a word in it,

201
00:20:05,010 --> 00:20:08,640
you probably needed it in your dictionary whether you thought it was a good idea or not.

202
00:20:09,150 --> 00:20:16,590
And DiDio takes this head on, and this is quite possibly my favourite article in the 74,000 articles of the Encyclopédie.

203
00:20:16,800 --> 00:20:19,740
So here is the article. Egoism Zima in the dictionary how to travel.

204
00:20:19,740 --> 00:20:27,630
So the direct predecessor of the encyclopaedia, which tells us it's a plant in Brazil or the islands of South America, and that's about it.

205
00:20:28,410 --> 00:20:33,030
Diderot includes this article in the first volume, and he says he repeats it.

206
00:20:33,480 --> 00:20:39,360
He just reuses the first bit of the article. Exact word for word. It's a plant that grows in Brazil, on the islands in South America.

207
00:20:39,690 --> 00:20:44,820
This is all that we are told about it. And I would like to know for whom such descriptions are made.

208
00:20:45,180 --> 00:20:51,509
It cannot be for the natives of the countries concerned, who are likely to know more about the agua Zima than is contained in this description,

209
00:20:51,510 --> 00:20:54,900
who do not need to learn that the magazine grows in their country.

210
00:20:55,620 --> 00:21:02,069
It is. If you said to a Frenchman that the pear tree is a tree that grows in France and Germany, etc., and it's not meant for us either.

211
00:21:02,070 --> 00:21:09,300
For what do we care that there's a tree in Brazil named Zima? If all we know about is its name, what is the point of giving the name?

212
00:21:10,080 --> 00:21:14,729
It leaves the ignorant just as they were, and teaches the rest of us nothing, if all the same.

213
00:21:14,730 --> 00:21:18,790
I mentioned this plant here, along with several others that are described just as poorly then,

214
00:21:18,810 --> 00:21:23,070
as that of consideration for sun readers who prefer to find nothing in a dictionary article,

215
00:21:23,400 --> 00:21:32,660
even to find something stupid, then to find no article at all, really, uh, takes some license with this poetics of Reus to it.

216
00:21:32,670 --> 00:21:40,379
But this really gets to the heart of this notion of compilation of how you had to construct a dictionary in the 18th century.

217
00:21:40,380 --> 00:21:44,220
And so if we look at the article compiler, compiler, the compiler,

218
00:21:44,820 --> 00:21:50,490
which is by sort of ironically, one of the great texture users of the answer to the Abbe melee,

219
00:21:51,450 --> 00:21:59,040
uh, he tells us that it's a writer who does not compose anything of genius, but who is content to collect and repeat what others have written.

220
00:21:59,040 --> 00:22:05,010
Very much not originality. However, most, like lexicographers, he tells us, are merely compilers.

221
00:22:05,940 --> 00:22:07,530
And he goes on to talk about this notion.

222
00:22:07,530 --> 00:22:17,519
And so but clearly somebody we would assume zero take some umbrage with this, with this idea of lexicographers being merely compilers.

223
00:22:17,520 --> 00:22:22,290
And there is another cross-reference at the bottom, probably added by literal to plagiarist.

224
00:22:22,740 --> 00:22:28,740
And so we get finally to plagiarism, and then it's a and it's a it's defined quite clearly at the beginning as a writer

225
00:22:28,740 --> 00:22:33,840
who plunders other authors and gives their productions of being their own work,

226
00:22:33,840 --> 00:22:39,840
this notion of plundering will come back in one of our titles, which was somewhat badly taken, but it's another story.

227
00:22:41,360 --> 00:22:50,030
And then DeNiro goes on and he says. But lexicographers, at least those who deal with the arts and sciences, seem to be exempt from the law.

228
00:22:50,060 --> 00:22:51,440
The common laws of mine and yours.

229
00:22:51,680 --> 00:22:58,190
They do not claim to build their own land, nor to draw from it, from it the materials necessary for the construction of their work.

230
00:22:58,790 --> 00:23:02,989
Indeed, the character of a good dictionary, such as we would like this one to be,

231
00:23:02,990 --> 00:23:06,870
consists largely in making the use of the best discoveries of others.

232
00:23:06,890 --> 00:23:16,240
What we borrow from others, we borrow openly in broad daylight, starting from the writers from which we have drawn our status as compilers.

233
00:23:16,250 --> 00:23:23,270
And he repeats the term gives us a right and or title to take advantage of everything that we can contribute to the perfection of our purpose,

234
00:23:23,270 --> 00:23:26,450
wherever it may be, may be found. If we steal.

235
00:23:27,690 --> 00:23:31,380
It is only in imitation of the bees who gather.

236
00:23:31,800 --> 00:23:38,000
Uh, for the profit, for the public good. And this notion of the of bees, I gather it goes back to the Renaissance.

237
00:23:38,010 --> 00:23:44,999
Clearly, Montana even brings this up, as he's just a bee that's taking nectar from wherever he finds it for the better.

238
00:23:45,000 --> 00:23:45,240
Good.

239
00:23:47,850 --> 00:23:56,590
And it cannot exactly be said that we plunder authors again back to our title, but that we draw contributions from them for the benefit of letters.

240
00:23:56,700 --> 00:24:02,250
So this, then, is how plagiarism is presented in the encyclopaedia at once.

241
00:24:02,700 --> 00:24:10,470
Something that's bad, but given the right context, could be potentially quite good and in fact almost a moral duty.

242
00:24:10,920 --> 00:24:18,690
Uh, and we'll see how this evolves. Uh, Voltaire takes this definition back up in his great work of, uh, text reuse.

243
00:24:18,690 --> 00:24:21,840
The question answered in 1772.

244
00:24:22,240 --> 00:24:30,330
Uh, and he has a sort of, uh, mercantile definition of plagiarism, uh, when an author sells the thoughts of another for his own.

245
00:24:30,900 --> 00:24:33,809
This petty theft is called plagiarism. We could call plagiarism.

246
00:24:33,810 --> 00:24:38,230
All the compilers, all the dictionary makers who do not, who do no more than repeat, uh,

247
00:24:38,250 --> 00:24:43,800
opinions, errors, uh, impostors and truths already, uh, printed in previous dictionaries.

248
00:24:44,280 --> 00:24:48,970
Uh, but they are at least plagiarism in good faith. They do not claim credit for invention.

249
00:24:48,990 --> 00:24:55,229
They do not even claim to be, uh, uh, to have on earth the materials they have assembled from the ancients.

250
00:24:55,230 --> 00:25:01,050
They have merely copied in the laborious compilers of the 16th century. They sell you a quote on what you already have in folio.

251
00:25:01,440 --> 00:25:07,590
Uh, call them booksellers, if you like or not. Authors classify them as second hand writers rather than plagiarism.

252
00:25:07,650 --> 00:25:13,440
This brings to mind the work of Antoine Compagno. I'm still going to do what I do the literature.

253
00:25:14,310 --> 00:25:22,170
Finally, and most interestingly, uh, my model will take up this notion of plagiarism in the supplement copied in 1777.

254
00:25:23,160 --> 00:25:27,130
And he, uh, is quite clear that it's no great thing to be.

255
00:25:27,270 --> 00:25:30,809
It's no great tragedy to be to be a plagiarist.

256
00:25:30,810 --> 00:25:32,220
And it's a sort of literary crime.

257
00:25:32,370 --> 00:25:40,890
And it's not it's not so serious for which, uh, pedants, envious people and fools do not fail, uh, to put famous writers on trial.

258
00:25:41,220 --> 00:25:43,620
So same true today as it was then.

259
00:25:44,310 --> 00:25:50,370
Plagiarism is the name they give to a theft of thoughts, and they cry out against this theft as if they themselves were being robbed,

260
00:25:50,910 --> 00:25:57,660
or as if it were essential to public order and peace, uh, that the properties of the mind should be invaluable.

261
00:25:57,690 --> 00:26:01,620
Clearly this is before the law and modern laws on copyright.

262
00:26:01,680 --> 00:26:09,540
Anyone who brings to light, uh, either to expression or appropriateness, I thought that was not as, um, but which would be otherwise lost.

263
00:26:09,570 --> 00:26:14,390
Makes it his own by giving it a new being. For oblivion is like nothingness.

264
00:26:14,490 --> 00:26:18,510
So here, plagiarism is a positive thing, and in fact, you have to do it.

265
00:26:19,140 --> 00:26:24,960
You have a right to a posterity to reuse others in public law.

266
00:26:24,990 --> 00:26:30,600
Uh, ownership of land is conditional and it not being cultivated on it.

267
00:26:30,840 --> 00:26:33,560
It's being cultivated by the owner. Uh, uh.

268
00:26:34,650 --> 00:26:39,930
And if the owner let it follow the society, which would have the right to demand, uh, that he give it up or sell it.

269
00:26:40,380 --> 00:26:47,280
The same is true of literature. Uh, he who has seized the happy and fruitful idea, uh, and does not develop it,

270
00:26:47,280 --> 00:26:54,450
leaves it as a common property for the first occupant who will know better than he does how to develop its richness.

271
00:26:54,870 --> 00:27:01,229
So this is the sort of notion of plagiarism that we've we've taken on this notion of text reuse as a productive and in fact,

272
00:27:01,230 --> 00:27:11,160
almost moral imperative in the 18th century, which clearly will be, uh, destroyed in the laws of 1791, 1793 of copyright and the right of the author.

273
00:27:11,340 --> 00:27:15,900
But for most of the of the 18th century, this sort of free and open, uh,

274
00:27:16,170 --> 00:27:22,400
plagiarist text reduces, is is operative, and nowhere more than in this in the antiquity.

275
00:27:23,280 --> 00:27:24,810
And it follows Diderot.

276
00:27:25,200 --> 00:27:33,059
d'Alembert continues in his advertisement from these reflections that the committee must often contain either extracts or even sometimes in full,

277
00:27:33,060 --> 00:27:40,680
several pieces of of the best works of each genre. It is only important to the public that the choice be made with clarity in economy.

278
00:27:41,190 --> 00:27:48,780
Uh, but it is also important for the authors to quote the originals accurately, both to enable the reader to consult them and to give each his due.

279
00:27:49,080 --> 00:27:52,830
This is how several several of our colleagues have done it.

280
00:27:53,040 --> 00:27:59,760
We wish that they had all done so. But in any case, uh, when an article is well done, it is equally enjoyable to read it.

281
00:28:00,150 --> 00:28:02,490
Uh, whichever form, whichever hand it comes.

282
00:28:02,920 --> 00:28:10,350
Uh, and the disadvantage of, uh, failing to quote which is always great in relation to the author, it's much less so in relation to the dictionary.

283
00:28:10,780 --> 00:28:16,050
Let's bring this back to this, sort of this, this sort of provoked us years ago to, to to think about how,

284
00:28:16,410 --> 00:28:22,170
how we can find these the quotations, uh, which were marked or sometimes not marked.

285
00:28:22,650 --> 00:28:27,180
Uh, and these different reuses in the encyclopaedia and how we could put the answer properly,

286
00:28:27,180 --> 00:28:33,719
which is the huge text, which is 74,000 article articles, 28 volumes in folio on this similarity spectrum.

287
00:28:33,720 --> 00:28:41,130
To come back to this. So somewhat and we did tests and an information retrieval and similar article, uh, comparisons.

288
00:28:41,550 --> 00:28:45,390
And we felt we tried plagiarism detection. It was a bit too strict.

289
00:28:45,390 --> 00:28:53,420
And so this is how all of. This is the sort of preface to how we got to investigating automatic text reads, and we published two articles.

290
00:28:53,430 --> 00:29:00,030
Again, I mentioned the first in Plundering Philosophers. And now you see where we get this title from, uh, in 2010, uh,

291
00:29:00,030 --> 00:29:05,100
which experimented with all these different sorts of, uh, of, of, uh, information retrieval.

292
00:29:05,550 --> 00:29:08,240
And then there's something borrowed which is more or less, uh,

293
00:29:08,250 --> 00:29:14,820
outlining how sequence alignment, uh, an approach to text reuse can be particularly fruitful,

294
00:29:15,330 --> 00:29:24,840
uh, for us, uh, and these are slides that I've, uh, in true text reuse manner has reused for at least 15 years.

295
00:29:26,340 --> 00:29:33,030
Uh, the investigation of these sorts of relationships begins with identification of, of, of similar passages using sequence alignment,

296
00:29:33,030 --> 00:29:39,750
which is a, which is a technique that's, that's used in other domains, uh, to find different strings of similarity.

297
00:29:39,990 --> 00:29:46,410
Uh, so it's used a lot in bioinformatics to think about similar DNA sequences, a plagiarism detection, which I've mentioned.

298
00:29:46,710 --> 00:29:55,200
And then inversely in text collation. So looking for uh, zones of, of uh variants into light traditions.

299
00:29:55,560 --> 00:30:03,450
Uh, we were inspired Michelsen in particular an article at the time, it was inspired by the bioinformatic approach, uh, of sequence alignment.

300
00:30:04,050 --> 00:30:05,430
This is still in use today.

301
00:30:05,790 --> 00:30:12,780
Uh, colleagues in Helsinki are here that use the Blast algorithm, which comes directly out of, uh, the bioinformatic, uh, world.

302
00:30:13,230 --> 00:30:20,820
Uh, at artful, we've developed Mark Olsen and now Carlos Gladstone, who is here, our own system called, uh, text pin.

303
00:30:21,300 --> 00:30:24,660
This is as technical as I'll get today. It's not very technical at all.

304
00:30:24,720 --> 00:30:28,830
Uh, but this is it. Give you an idea of what tax payer does, what the system does.

305
00:30:29,190 --> 00:30:36,569
Uh, we take, uh, large databases, we pre-treat them, turn them in what's called n-gram.

306
00:30:36,570 --> 00:30:43,559
So groups of words, uh, we look for the common sequences of these ngrams, uh, and then we adjust parameters.

307
00:30:43,560 --> 00:30:47,040
There's a high flexibility. Again, plagiarism detection was a bit too strict.

308
00:30:47,040 --> 00:30:53,159
What we're interested in, especially in my new project, is, uh, zones of, uh, where the same text is there,

309
00:30:53,160 --> 00:30:59,730
but there might be changes or errors or things, you know, you can jump out of. So this, this idea of, of, uh, flexible parameters is very important.

310
00:31:00,540 --> 00:31:06,269
And so if you take the first two lines of Rousseau's Social contract, learning about to read only fair, uh,

311
00:31:06,270 --> 00:31:14,610
this is how we pre-treat the, the, uh, the document, uh, it becomes omnibus to leave about to fair, uh, you can see how this works.

312
00:31:14,760 --> 00:31:21,780
The idea being that, um, Libor part two, uh, if shared between two documents that are not the same document, is a fairly rare event.

313
00:31:22,080 --> 00:31:26,489
Uh, and given the rarity of this event, the algorithm will then think about, uh,

314
00:31:26,490 --> 00:31:30,690
are there other n n-gram shared either to the left or right in a sliding window that we define?

315
00:31:31,380 --> 00:31:37,080
Uh, so this is a deceptively straightforward, uh, notion of intertextuality.

316
00:31:37,080 --> 00:31:40,200
It's the same set text, more or less in two different places.

317
00:31:40,530 --> 00:31:47,640
Uh, but you'll as you'll see today, the range of text we use, uh, the detection of reuses was only the first step.

318
00:31:47,790 --> 00:31:55,320
And in fact, it's understanding how these reuses are employed and deployed in the 18th century that it becomes really interesting.

319
00:31:55,770 --> 00:32:05,550
So, for example, there's no real surprise that. The answer cites or uses that it was a known predecessor of the non-security.

320
00:32:06,270 --> 00:32:10,230
And we find a very long segment at the end of the article on Spinoza.

321
00:32:10,500 --> 00:32:17,700
But what is surprising is that at the end of the article, the event finishes in Bell's voice.

322
00:32:18,330 --> 00:32:21,930
So he concludes the article in a funny budget.

323
00:32:22,230 --> 00:32:25,530
So there's no quotation marks. Bell is mentioned at the beginning.

324
00:32:25,860 --> 00:32:32,519
Uh, so there's something going on here, but it's a way of sort of putting Bailey in the anti-gravity.

325
00:32:32,520 --> 00:32:36,720
His voice becomes part of the of the authorial fingerprint of the committee.

326
00:32:37,260 --> 00:32:44,190
So he concludes the article whether or not the readers of the time knew this remains to be seen, but for us,

327
00:32:44,190 --> 00:32:51,090
it was a clear indication that we had no idea what was going on here until the computer, uh, found this match.

328
00:32:52,080 --> 00:32:54,809
It can do. We've we found this with lots of different work.

329
00:32:54,810 --> 00:32:58,950
So, for example, the idea of using lock in the encyclopaedia is really, really interesting.

330
00:32:59,250 --> 00:33:06,150
Uh, and we can do this using, uh, using text from, from echo, uh, which is a translation obviously.

331
00:33:06,540 --> 00:33:10,770
Uh, and using lock in the encyclopaedia was a bold move.

332
00:33:10,890 --> 00:33:14,280
Uh, lock second treatise was in France banned and burned.

333
00:33:14,640 --> 00:33:15,600
It was on the index.

334
00:33:15,840 --> 00:33:23,220
So if you said I'm going to say lock, uh, chances are the censor would come and say, you know, either take it out or you'll lose your privilege.

335
00:33:23,700 --> 00:33:28,799
Uh, but, uh, is a man possessed, and he says, I'm going to use lock.

336
00:33:28,800 --> 00:33:30,480
I just won't tell anybody what's going on here.

337
00:33:30,480 --> 00:33:38,280
And so all through the article, uh, government is lots and lots of lock, uh, with no indication whatsoever, uh, that this is where it comes from.

338
00:33:38,310 --> 00:33:46,650
Uh, and again, it's more or less it's not a copy. Equally, it's not, uh, that he's taking a bit of lock and just pasting it in for an article.

339
00:33:46,650 --> 00:33:50,640
It's that he's weaving it into his own articles to support his own argument.

340
00:33:51,060 --> 00:34:00,960
And it's it's very much how text reuse worked in, uh, in the 18th century as a way of generating meaning through the use of other models.

341
00:34:01,380 --> 00:34:05,880
This, uh, caused us to think about the status of different works. What works could you cite in the notes you prepared?

342
00:34:06,150 --> 00:34:10,410
Couldn't you could cite the only add, for example, was less controversial.

343
00:34:10,770 --> 00:34:17,850
Uh, but you could never cite we shouldn't ever cite the, uh, Latrobe philosophy because, again, like, good luck, it was abandoned, burned.

344
00:34:19,050 --> 00:34:24,840
And then there was a whole sort of spectrum of dangerous works that would make their way into obscurity.

345
00:34:25,020 --> 00:34:30,149
It was really stupid to think about citing The Last Plague, because the list from 11 issues, in fact,

346
00:34:30,150 --> 00:34:35,549
causes the Nazi Germany to lose its privilege, uh, because of its association with materialism.

347
00:34:35,550 --> 00:34:45,540
But once again, so cool. Uh, he, uh, is quite insistent, so he uses a bit of, uh, the recipe in his article on, uh, the, uh, mine.

348
00:34:45,840 --> 00:34:49,169
Uh, but he gives a tip of the hat here. He says to people, oh, here it is.

349
00:34:49,170 --> 00:34:56,579
There's no there's no quotation marks. Uh, there's just this passage and he says, see, decomposition in the social class.

350
00:34:56,580 --> 00:35:02,100
So it's up to the reader to understand who is this Bush-Cheney? Uh, what's going on?

351
00:35:02,610 --> 00:35:06,120
Is it? And would they recognise this is Eldridge shows. Not so sure.

352
00:35:06,450 --> 00:35:09,780
Uh, it's true. There's a whole system of coding that goes on the property.

353
00:35:10,080 --> 00:35:15,150
So at least on philosophies Voltaire, Lupu, Euclid's, Montesquieu.

354
00:35:15,450 --> 00:35:18,389
Uh, there's lots of different other ways of referring to people.

355
00:35:18,390 --> 00:35:24,870
But this is, again, a case where perhaps the readers of the 18th century saw what was going on.

356
00:35:25,500 --> 00:35:35,160
So short readers today certainly do not some specialist maybe, uh, but it's a real area where, uh, the computer can bring us to new knowledge.

357
00:35:35,520 --> 00:35:38,430
And this was published it, of course, with Dana Justin and Robert Marcy.

358
00:35:38,670 --> 00:35:42,330
And then we continued along this vein, thinking about hidden voices in The Secret,

359
00:35:42,510 --> 00:35:45,720
which is a polyphonic work, as you saw with the sort of incorporation of Bell.

360
00:35:46,170 --> 00:35:52,380
Uh, maybe we know those 140 authors, but maybe there's many, many, many more that we just don't know about.

361
00:35:52,440 --> 00:35:55,350
Uh, and one of these, of course, was was Emily Chatelain.

362
00:35:56,130 --> 00:36:03,780
And somebody knew that the article was, in fact, drawn more or less from her, uh, instead situated on the physique.

363
00:36:04,230 --> 00:36:10,800
And so we digitised or we took a digitised version of the institution, the physique, and we compared it to the encyclopaedia.

364
00:36:11,370 --> 00:36:16,769
And what was it? It was really interesting. We found, I think, 13 articles with at least some Emily Chatelet.

365
00:36:16,770 --> 00:36:23,940
She is, of course, not an author. There's only one female author in the annals seem to be any and Darrow refers to as an anonymous woman.

366
00:36:24,150 --> 00:36:28,470
It's a bit sad. We think it's called wife. That's an article on fashion.

367
00:36:29,010 --> 00:36:35,100
Uh, that's the only woman in the list. But nonetheless, you can see that some of these articles and we've given them the scores.

368
00:36:35,100 --> 00:36:40,140
And how much Du Chatelet is in there. Contradiction is 95% du Chatelet.

369
00:36:40,890 --> 00:36:46,140
Uh, and the reason that form is mentioned everywhere is that there was this guy named Senator for me,

370
00:36:46,650 --> 00:36:51,930
was the secretary of the Academy of Berlin, and he had he was going to make his own dictionary.

371
00:36:52,200 --> 00:36:55,259
And then he heard about delivering that on Bell, and he got discouraged.

372
00:36:55,260 --> 00:36:58,470
And so he gives his papers to that on there. He says, do what you want with them.

373
00:36:58,770 --> 00:37:04,009
But if. He just cribs and steals from me today in many places and so on.

374
00:37:04,010 --> 00:37:07,400
Bill doesn't quite know that what's going on, and so he puts this into the answer, Billy.

375
00:37:07,670 --> 00:37:13,310
But in fact all of these articles continue, for example, then on verse as these come from the papers of Samuel Fournette.

376
00:37:13,790 --> 00:37:19,999
But it's a media shack today. Uh, so this sort of gives us this notion of, uh, there are these articles that should rightly,

377
00:37:20,000 --> 00:37:24,290
somehow or we should somehow mention them today as being at least,

378
00:37:24,710 --> 00:37:34,610
uh, a co-author, uh, with, uh, with these others and then, uh, recently or somewhat recently, uh, we thought about translation as read.

379
00:37:34,610 --> 00:37:42,500
So if, you know, sort of the history of the institute, it started out as a translation project of the English Cyclopaedia, Ephraim Chambers,

380
00:37:42,680 --> 00:37:50,930
and then eventually quickly, uh, it got out of hand and they decided, well, this is it's going to be much more than just a translation.

381
00:37:51,110 --> 00:37:56,330
But they had this, this block of text, of translations of articles that were started by different people.

382
00:37:56,720 --> 00:38:00,530
Uh, and we're never quite sure where they occur in the security.

383
00:38:00,740 --> 00:38:09,290
There are articles that have a mention, uh, and here Paolo says that many times they left these articles and just just used them as filler,

384
00:38:09,740 --> 00:38:12,740
uh, almost unchanged without significant modifications.

385
00:38:12,960 --> 00:38:17,600
Uh, but there's never been a sort of systematic undertaking of how to find these articles.

386
00:38:17,810 --> 00:38:22,890
Some are marked with this and, uh, and reference to chambers.

387
00:38:22,910 --> 00:38:31,220
There's about 1100 of these. And, uh, this is a good article, uh, a guide to the law by women.

388
00:38:31,520 --> 00:38:40,030
And you can see, uh, what we wanted to do was somehow draw this connection between, uh, the chambers at the bottom and the, uh, at the top.

389
00:38:40,040 --> 00:38:46,840
And at the time, I think it's 2018. 2019. It was before these great multilingual language models.

390
00:38:46,850 --> 00:38:53,899
I'm sure now today we could do it much more easily. Levels probably have different ways, but we did it in a in a pretty hand-cranked way.

391
00:38:53,900 --> 00:38:59,000
As we said, as we digitised the 1742 chambers, which was the chambers that Darrow owned.

392
00:38:59,480 --> 00:39:08,690
Uh, we ran it through Google Translate automatically, and we kept the structure and we built a virtual dictionary, a sort of fake French chambers.

393
00:39:09,020 --> 00:39:13,790
And with that French chambers, we ran the sequence alignment against it to find articles.

394
00:39:13,790 --> 00:39:22,399
So, for example, an accuracy comes back and we had to really, uh, get fine grained on our comparisons because obviously it's a modern translation.

395
00:39:22,400 --> 00:39:26,660
So for example, it totally has no idea what a petticoat government might or might not be.

396
00:39:26,960 --> 00:39:30,650
It's a uppity motto. It makes no sense whatsoever.

397
00:39:30,650 --> 00:39:34,430
It's a terrible translation, but there's enough. Uh n-gram.

398
00:39:34,430 --> 00:39:37,700
There's enough content in there that we're able to to get a match.

399
00:39:38,000 --> 00:39:43,310
Uh, and again, uh, it was already signed by chambers, uh, and we only found about 800 of the 1100.

400
00:39:43,970 --> 00:39:46,910
So it could be due to translation, could be due to lots of different things.

401
00:39:47,180 --> 00:39:52,520
But we found like 2500 more articles that had no mention to chambers whatsoever.

402
00:39:52,700 --> 00:39:58,760
So, for example, a friend, the MLA, writes the article demoniac, which is a word for word,

403
00:39:59,060 --> 00:40:03,410
uh, the same amount of paragraphs, the same more, uh, cross-references that chambers.

404
00:40:03,620 --> 00:40:07,880
It's taken clearly directly from chambers. And he signs his name to it.

405
00:40:08,510 --> 00:40:11,730
So again, is this the chambers article, or is this the animal?

406
00:40:11,750 --> 00:40:14,959
It's, uh, moving on.

407
00:40:14,960 --> 00:40:20,180
This is sort of to give you a very synthetic view of how we envision text reuse detection.

408
00:40:21,970 --> 00:40:29,910
We're progressively moving out of the city rapidly and into the sort of larger communication world of of print culture in the in the 18th century.

409
00:40:29,920 --> 00:40:32,799
This is Robert Darwin's famous communication network.

410
00:40:32,800 --> 00:40:37,450
And the idea here is that information flows in the 18th century through all sorts of different channels.

411
00:40:38,020 --> 00:40:44,380
Oddly enough, correspondence doesn't show up on his communication network, which I have always found a bit odd.

412
00:40:44,740 --> 00:40:50,290
Uh, but again, this goes back to our time here. Uh, my time here in Oxford with Nicholas, one of the,

413
00:40:50,290 --> 00:40:56,260
one of the first ideas we had was electronic enlightenment correspondence in general should be really rich for text for years.

414
00:40:56,270 --> 00:41:04,000
There should be some sort of platform in which text for use, uh, is filtered through this information space that is at one time,

415
00:41:04,360 --> 00:41:11,079
uh, sociable, but also, uh, a way of diffusing information, a way of building literary authority and otherwise.

416
00:41:11,080 --> 00:41:17,680
So we all know electronic enlightenment. And you give you an example of how, uh, this sort of thing can work with poor correspondence.

417
00:41:17,980 --> 00:41:22,090
We can take an event like the the earthquake in Lisbon in 1755.

418
00:41:22,090 --> 00:41:29,950
It happens in November 1st. And we know that Voltaire gets news of this, gets wind of this, and it sort of shakes his faith in.

419
00:41:31,320 --> 00:41:35,810
In, uh. In an optimistic worldview in divine providence.

420
00:41:36,290 --> 00:41:43,429
And he writes upon and. And we find traces of the Pope already in January 1756.

421
00:41:43,430 --> 00:41:51,860
So it's very fast in terms of early modern, and it's just a little bit and it's, uh, uh, someone that's right to run.

422
00:41:51,860 --> 00:41:57,680
Hello. Who says while I was in his presence and he recited this poem and I have these last eight verses,

423
00:41:58,110 --> 00:42:03,830
uh, and it's a very funny letter, uh, because he says he wrote this poem.

424
00:42:04,220 --> 00:42:09,560
I was sent the eight verses which enclose here. I have reason to believe that it is in his own admission.

425
00:42:09,560 --> 00:42:16,340
It seemed to me these were a little materialistic things, and I've replied so.

426
00:42:16,400 --> 00:42:24,740
But there's no indication here. I've replied with a few fragments of my own translation of Van Haller's, uh, work on, uh, the origins you met.

427
00:42:25,190 --> 00:42:27,829
Uh, so this is a really a sort of Frankenstein poem here.

428
00:42:27,830 --> 00:42:37,460
That is eight lines of Voltaire, eight lines of, uh, translation of, uh, of Eisenhower's work, which he does directly into, into the correspondence.

429
00:42:37,850 --> 00:42:39,739
So this is this poetics of text reuse.

430
00:42:39,740 --> 00:42:46,700
That's, that's been and if we compare the eight lines by Voltaire, we can see this is, uh, we talked about this this afternoon,

431
00:42:46,700 --> 00:42:49,909
this idea of at some point you get to the end point of these reuses,

432
00:42:49,910 --> 00:42:53,719
you want to see what the what the, the differences are rather than the similarities.

433
00:42:53,720 --> 00:42:57,410
And in this case, the last two lines don't make it into the final cut.

434
00:42:58,220 --> 00:43:08,440
And because we're using modern, uh, modern text from the Voltaire Foundation, uh, we wouldn't get this end of this, this couplet comfort,

435
00:43:08,450 --> 00:43:16,670
you know, maternal mortality for Soufriere to Sumatra, where the Voltaire thought this was a bit depressing and he doesn't continue on with it.

436
00:43:16,700 --> 00:43:24,650
Uh, but it's a way of recreating this, this sort of notion of his creating through the correspondence that was creating, uh, different ways.

437
00:43:24,820 --> 00:43:28,940
He's retouching his work, and eventually the sequences get longer as we go on.

438
00:43:28,940 --> 00:43:38,629
And so by March, uh, it's much longer. And it's another letter to Von Harlow where he says, I was in the present, and he gave us that road to shake.

439
00:43:38,630 --> 00:43:45,220
So we know it's a retouched piece. And again, here we can show, uh, that more or less it's the same version, right?

440
00:43:45,230 --> 00:43:50,530
There's a few differences. Providence and presence has been changed to prime, uh, live.

441
00:43:50,570 --> 00:43:55,660
But again, this this can tell you how the correspondence is really interesting to see different states of,

442
00:43:55,750 --> 00:44:01,970
of a work which don't quite make it into editions or might not make it into conditions, but good ones, they work.

443
00:44:03,410 --> 00:44:10,760
Uh, finally. Uh oh. And then we know as a sort of, uh, media event, somebody sends a letter to Russo and says,

444
00:44:10,760 --> 00:44:14,750
what, are you going to let this guy question divine providence?

445
00:44:15,110 --> 00:44:21,259
And, of course, Rousseau was always up to the challenge. And he responds with this really long letter, which is about 6000 words, uh,

446
00:44:21,260 --> 00:44:27,890
which gets printed as on optimism and, and, and and sort of, uh, flies all over you.

447
00:44:28,670 --> 00:44:31,520
Voltaire was not very happy about this, that his letter was printed.

448
00:44:31,940 --> 00:44:40,730
Uh, and so we can only assume that one of the responses to this in the same year that on optimism was, was, uh, published is probably content.

449
00:44:40,880 --> 00:44:48,800
So, again, this way of thinking of, uh, reuse as a way of, uh, of augmenting or understanding, uh, cultural dissemination.

450
00:44:49,610 --> 00:44:56,989
And also, uh, I'd be remiss not to mention just the way that Voltaire reads and reads with his correspondence,

451
00:44:56,990 --> 00:45:02,240
which is really he was like a Reader's Digest. Uh, and he loved to exhibit works, and he send it to his friend.

452
00:45:02,240 --> 00:45:10,880
So here is another added distance works. It's published posthumously, 1771, uh, or 1773.

453
00:45:11,000 --> 00:45:17,569
Here Voltaire is taking little snippets out of it, sending it to, uh, that on there and saying, oh, don't you think this is great?

454
00:45:17,570 --> 00:45:25,340
It's too bad. It's actually a pretty bad book. Uh, he does the same thing for Mademoiselle, the font, uh, which is nice, because she's blind.

455
00:45:25,580 --> 00:45:30,200
Uh, so he's sort of saying, here, I'm giving you all the good parts.

456
00:45:30,560 --> 00:45:33,740
Don't waste your time with the book. It's a bad book.

457
00:45:34,040 --> 00:45:35,240
But I've given you the good things.

458
00:45:35,540 --> 00:45:41,149
And he says, I take these little diamonds at random, and there are thousands of this tastes whose brilliance struck me.

459
00:45:41,150 --> 00:45:43,430
That does not prevent the book from being very bad.

460
00:45:43,820 --> 00:45:50,360
I spend my life looking for precious stones in manure, and when I find the sum, I put them aside and make a profit from them.

461
00:45:50,750 --> 00:45:58,129
That's why bad books are sometimes very useful. So here, in a nutshell, is this sort of text reuse poetics of the 18th century.

462
00:45:58,130 --> 00:46:06,320
So this is notion of somehow you accrue through this dissemination of, of different of both the act of extracting, uh,

463
00:46:06,320 --> 00:46:12,770
these precious diamonds from books, but also sending them out so we can't pick it up unless somebody sends it out.

464
00:46:12,770 --> 00:46:17,210
So text reuse can only help you in this moment that somebody reuses the text again.

465
00:46:17,660 --> 00:46:20,930
Um, but again, it's a whole question.

466
00:46:20,930 --> 00:46:24,379
So we have writers notebooks, for example. So Voltaire knows he has his notebooks.

467
00:46:24,380 --> 00:46:29,990
So these are the things that he writes down, these these quotations of how what's the moment that the quote,

468
00:46:29,990 --> 00:46:38,899
he decides the quotation is going to move out into, into the world. And so all of this, to move finally to the new, uh, project we have now, which is,

469
00:46:38,900 --> 00:46:45,980
which is to think precisely about, uh, this whole sort of network or this whole sort of realm or cultural system.

470
00:46:46,400 --> 00:46:51,379
Uh, that is the 18th century based on reuses. So we want to think about intertextual networks.

471
00:46:51,380 --> 00:46:55,730
Don't worry too much about this network. They all sort of look like giant hairballs when you do it.

472
00:46:55,730 --> 00:47:03,410
But Voltaire is always in the middle just because, uh, as I said, he was a real texter user.

473
00:47:03,680 --> 00:47:07,790
Uh, and this project, uh, called modern that Nicholas mentioned, uh,

474
00:47:07,790 --> 00:47:13,279
is based on this notion that perhaps we can think about the exchange of text we use.

475
00:47:13,280 --> 00:47:21,470
So not just identifying the reuses, that's the first step, but thinking about how the reuses travel and are used and reused, uh, through networks.

476
00:47:21,680 --> 00:47:29,300
And so borrowing from work from Danielle Stein, who started looking into networks at Stanford with the Mapping the Republic of Letters project,

477
00:47:29,720 --> 00:47:39,860
more recently, thinking about social network analysis as a as a way of understanding cultural exchange, uh, through different, different areas.

478
00:47:39,860 --> 00:47:46,100
And here, Ruth and Sebastian Anand have been instrumental in promoting this approach in, in the humanities.

479
00:47:46,580 --> 00:47:51,530
Uh, they have a wonderful new book out I recommend to everyone called Tutor Networks of Power.

480
00:47:52,310 --> 00:48:02,420
There were also, of course, uh, an Oxford Studies collection by Dan and, uh, Chloe Edmunds and thinking about how to use social and network analysis.

481
00:48:03,900 --> 00:48:11,100
For the humanities. And a good primer. Again, it's by Ruth and Sebastian with with Katherine Coleman and Scott Weingarten at work.

482
00:48:11,100 --> 00:48:13,050
Turn. For my own part.

483
00:48:13,350 --> 00:48:23,610
To keep this notion of theorising, uh, into it, I, uh, I add a sort of extra layer of what Bruno Latour calls the actor network theory.

484
00:48:23,610 --> 00:48:30,780
And here, the way we think about this with the networks that we want to build, uh, the two advantages of actor network theory.

485
00:48:31,110 --> 00:48:35,489
One is that, uh, it's it's, uh, there's no presuppositions.

486
00:48:35,490 --> 00:48:41,879
There's no, uh, a priori, uh, you simply generate these networks and you try to trace where the data takes you.

487
00:48:41,880 --> 00:48:45,120
So it's very online with what we call data driven research.

488
00:48:45,360 --> 00:48:51,809
You build the data sets, you run the algorithms over it, what comes out then you try to make sense of, uh,

489
00:48:51,810 --> 00:49:00,840
and also within this sort of intellectual framework of, uh, actor network theory, actors can be anything they need, not be human beings.

490
00:49:01,290 --> 00:49:05,550
Uh, they can be books, they can be texts, they can be snippets. Uh, they can be speed bumps.

491
00:49:05,570 --> 00:49:11,040
There's another Tor tells us it's anything that affects the behaviour or the predicted behaviour of a network.

492
00:49:11,430 --> 00:49:17,190
Uh, and another important takeaway from this is that sometimes it's not the originators nor the receivers,

493
00:49:17,460 --> 00:49:22,320
but it's the mediators that are in fact the most important nodes in a network.

494
00:49:23,610 --> 00:49:26,040
Uh, but to do this, of course, we have to have a lot of data.

495
00:49:27,070 --> 00:49:34,209
And this is what we've been doing for the past couple years, compiling lots and lots of data as much as we can get in the 18th century realm,

496
00:49:34,210 --> 00:49:36,760
with the only caveat being that we don't digitise anything.

497
00:49:37,300 --> 00:49:45,010
So the idea is that digitised collections are significantly sufficient now that we should use it,

498
00:49:45,010 --> 00:49:50,810
and that because, yes, there are lots of errors and because, yes, they come from OCR, uh, there are errors.

499
00:49:50,830 --> 00:49:57,430
We should nonetheless use them. So we've taken uh, a mixture of, uh, corrected works, which we call canon.

500
00:49:57,430 --> 00:50:02,290
So the, the distinction is only qualitative and archive, which are OCR.

501
00:50:02,860 --> 00:50:12,040
So we're about 12,400 books. Uh, in this, in this main collection, which we will then compare iteratively through pamphlets,

502
00:50:12,040 --> 00:50:16,359
letters, dictionaries and finally the press which we're working on with the BNF.

503
00:50:16,360 --> 00:50:23,829
Uh, and the notion here is to think about to that communication network of Darnton, uh, that it's not just books, uh,

504
00:50:23,830 --> 00:50:29,650
that are important in the 18th century, but how they move through things like pamphlets, letters, and especially the press.

505
00:50:30,070 --> 00:50:35,110
Very interesting to have a think about, uh, just scale.

506
00:50:35,380 --> 00:50:40,990
I think now we have some notion that the press will be about 300,000, 300 million words.

507
00:50:41,380 --> 00:50:44,800
So we're up to a billion words, which is completely arbitrary.

508
00:50:44,800 --> 00:50:48,670
But it's what I said in my, uh, grant proposal that we get a billion words.

509
00:50:49,180 --> 00:50:56,890
But the idea being that with a billion words, you can build your, uh, your own data, uh, your own large language model for the 18th century.

510
00:50:57,220 --> 00:51:01,870
So we can think about either fine tuned one of them, but maybe we can ask Voltaire some questions,

511
00:51:02,410 --> 00:51:06,250
uh, through the language model that we built nonetheless.

512
00:51:06,280 --> 00:51:11,079
Uh, so what? Who cares? This is the question I tell my students.

513
00:51:11,080 --> 00:51:15,310
Just put it in. So what? So what? Uh, what can we do then?

514
00:51:15,580 --> 00:51:19,569
We have all this data. We're going to run these alignments. We're going to we're going to have to filter them.

515
00:51:19,570 --> 00:51:22,840
But that's a whole that's a question. That's a story for another time.

516
00:51:23,140 --> 00:51:26,469
Uh, the first thing we want to do is just simply supplement traditional methods.

517
00:51:26,470 --> 00:51:30,910
It's just new ways of, uh, of doing what historians have always done,

518
00:51:31,750 --> 00:51:38,250
but on steroids and in some sense, in a way that's, uh, both faster and perhaps more comprehensive.

519
00:51:38,260 --> 00:51:43,390
So it will contain a the modern database will contain a great number of these intertextual reuses,

520
00:51:43,810 --> 00:51:50,800
make it possible to sort of follow the fortunes of inter text or concepts as they move through this information space.

521
00:51:50,920 --> 00:51:55,959
This is very much the sort of conservative side of intertextuality come from Harold Bloom.

522
00:51:55,960 --> 00:52:03,820
So to take one example, we have, uh, Barlow, uh, who writes, uh, reading a book would have laid over all of them.

523
00:52:04,630 --> 00:52:08,290
But this is very much an aesthetic judgement. It's talking about verisimilitude.

524
00:52:08,680 --> 00:52:20,440
It's about how, uh, how a text, a fiction, uh, can has to be, has to be sort of true for people to embrace it.

525
00:52:20,950 --> 00:52:24,520
Uh, and you can sort of follow it as it moves through the 18th century.

526
00:52:24,530 --> 00:52:28,359
Uh, and so cognac brings it up. But in speaking about imagination.

527
00:52:28,360 --> 00:52:32,110
So it's still very much the aesthetic, this notion that it's still verisimilitude,

528
00:52:32,110 --> 00:52:36,480
but he's talking about the imagination and how the imagination forms and fables and fictions, etc.

529
00:52:36,700 --> 00:52:40,749
Uh, by the time you get to Samuel Johnson, the Rebecca, he repeats it.

530
00:52:40,750 --> 00:52:43,900
He has quotation marks. It's good, but he undermines completely.

531
00:52:43,910 --> 00:52:49,170
Barlow and he says, ran a book on a railway racer, the demon quokka, Cedric Campbell.

532
00:52:49,180 --> 00:52:52,930
Very true. So even though the person that wrote it never wrote a word of truth.

533
00:52:53,440 --> 00:52:59,899
Uh, and then by the time we get to the revolution. This is an interesting text, which I'll talk about here shortly, called The Impulse Electric.

534
00:52:59,900 --> 00:53:07,380
You can, uh, you can see that. There's no indication at all that it's Bollo, so no idea that's what the text is there.

535
00:53:07,620 --> 00:53:11,070
But just by its context, you can see that it's changed into a moral category.

536
00:53:11,490 --> 00:53:15,570
Uh, this true? This is very different from the beginning.

537
00:53:16,020 --> 00:53:20,400
From what? And so which by this time it's changed completely in its valence.

538
00:53:21,340 --> 00:53:24,370
So that's just one example of what you could do with our data set.

539
00:53:24,700 --> 00:53:28,840
Secondly, this gets to what Nicholas was mentioning early earlier.

540
00:53:29,200 --> 00:53:35,500
In a sense of distant reading, we hope we can really sort of remap literary the literary history of the 18th century

541
00:53:36,190 --> 00:53:40,239
using these sort of social network analysis and how they move through the different thing.

542
00:53:40,240 --> 00:53:48,220
And this sort of presupposes a sort of radical, uh, notion of intertextuality, where everything is in an inter text.

543
00:53:48,580 --> 00:53:55,930
And in fact, it's more it's more about this process of dissemination that tells us, uh, not of reproduction, but of productivity.

544
00:53:55,930 --> 00:54:04,780
So an inter text or a cluster of inter texts is a productive force, uh, sometimes without an author, without any reference to intentionality.

545
00:54:04,780 --> 00:54:08,229
It's just these reuses that sort of move through these texts.

546
00:54:08,230 --> 00:54:17,680
And so we have to sort of think about different ways of thinking about authorship, how we understand authors in this, in this, in this text.

547
00:54:17,680 --> 00:54:21,990
And for us, it's just one attribute, uh, of a text object.

548
00:54:22,000 --> 00:54:26,080
So the text is primordial for us, the text that circulates, that moves its way through.

549
00:54:26,590 --> 00:54:32,860
If we know the author, great. If we don't, it's all right. Uh, we hope to find it at some point.

550
00:54:33,460 --> 00:54:40,120
And these are attributes that can be derived from other relationships between texts, objects, and then an author object.

551
00:54:40,120 --> 00:54:43,620
If we talk about author objects in our networks, uh,

552
00:54:43,630 --> 00:54:48,670
it's the result of the combination of all these different characteristics attributed to it by the relations between the text objects.

553
00:54:48,670 --> 00:54:56,200
So this is a sort of, uh, an author's important, uh, not because of what we read about literary history,

554
00:54:56,200 --> 00:55:01,840
but it's in their place in the network and their sort of lines of force as we users.

555
00:55:03,180 --> 00:55:08,309
This also causes us to think about what the text, uh, very prosaically, for us,

556
00:55:08,310 --> 00:55:14,250
the text is the sort of concordance of at least four similar trigrams, uh, that follow within a defined interval.

557
00:55:14,640 --> 00:55:17,880
Uh, but this poses all sorts of questions. And, uh.

558
00:55:19,600 --> 00:55:27,850
All reuses are not equal. So for example and this comes from Dario Nicolas, you found this great example of Marat during the revolution,

559
00:55:28,270 --> 00:55:32,310
uh, reuses whether consciously, whether on purpose or not, we don't know.

560
00:55:32,320 --> 00:55:36,970
It doesn't really matter. Uh, fentanyl. So you would never join these two together?

561
00:55:37,330 --> 00:55:40,900
Uh, he changes it a bit. New becomes Diego.

562
00:55:41,380 --> 00:55:46,320
Uh, but the context is completely different. I mean, it just couldn't be clearer.

563
00:55:46,330 --> 00:55:49,690
And so here we are, like, this is what we want. This is exactly what we want.

564
00:55:50,290 --> 00:55:54,280
But nonetheless, in the bottom here and the the titles are the same.

565
00:55:54,790 --> 00:55:58,510
And we'd say this is really interesting, but if you look at the text, it's not interesting at all.

566
00:55:59,380 --> 00:56:04,850
It's reuse. It's an inter text. You can't say it has an author.

567
00:56:05,330 --> 00:56:09,590
It circulates. It has some value, but we're not quite sure.

568
00:56:10,040 --> 00:56:12,200
Uh, so this sort of gets us into this notion.

569
00:56:12,620 --> 00:56:21,620
And this is exactly why we always we've been trotting out this, this notion of bath for 15 years, but we never quite got to the end of the quotation.

570
00:56:22,460 --> 00:56:26,690
And then he says, in fact, intertextuality is a condition of any text whatsoever.

571
00:56:27,050 --> 00:56:29,810
It's obviously not reducible to a problem of source and influences,

572
00:56:29,810 --> 00:56:34,879
which is what we've always thought into text is a general field of anonymous formulas,

573
00:56:34,880 --> 00:56:41,420
the origin of which is rarely identifiable of unconscious or automatic quotations given without inverted commas.

574
00:56:41,450 --> 00:56:45,710
So this is how do you find that? How do you understand that? What does that mean?

575
00:56:46,220 --> 00:56:52,760
Uh, and how can we, uh, implement that into our into our networks and then find a come back to, isn't it?

576
00:56:53,240 --> 00:56:56,690
He says, in fact, some texts are really, really hyper textual.

577
00:56:56,690 --> 00:57:00,980
And this is before Tim Berners-Lee. So it's a it's a theoretical hypertext.

578
00:57:01,400 --> 00:57:05,450
Uh, there's no literary work that to some degree in depending on the reading does not evoke another.

579
00:57:05,840 --> 00:57:08,240
Uh, and in this sense, all works are hypertension.

580
00:57:08,630 --> 00:57:14,690
But like Orwell's and Animal Farm, some are more hyper textual, more obviously massively in explicitly than others.

581
00:57:15,050 --> 00:57:20,660
You know, CBD is one of these, uh, Voltaire's questions on security is one of these hyper textual works.

582
00:57:21,050 --> 00:57:25,550
Uh, and we find them again and again in popping up the Republican.

583
00:57:25,730 --> 00:57:32,360
This is a, uh, a beginning, a sort of very preliminary, uh, look at network, uh, measures.

584
00:57:32,660 --> 00:57:36,620
It kept popping up. We didn't even know what it was. It was just in our it was in our corpora.

585
00:57:36,950 --> 00:57:42,410
Uh, and then we looked at it and we said, oh, this makes perfect sense why it would be there and why would have these high scores?

586
00:57:42,740 --> 00:57:51,469
Uh, because it it's a text reuse machine. Uh, and he writes it in at the very beginning of, of the revolution, and he says,

587
00:57:51,470 --> 00:57:54,740
I'm offering it to these students, all these little maxims, all these little princes.

588
00:57:54,740 --> 00:57:58,760
And you have one a day, uh, that you're meant to read and think about and ruminate on.

589
00:57:59,390 --> 00:58:05,600
And it's for the entire Republican calendar. Uh, and so again, we thought, well, this is really interesting.

590
00:58:05,600 --> 00:58:09,499
We should be able to find more than we thought. But in fact, it was a lot of work.

591
00:58:09,500 --> 00:58:13,309
And Dario had to do a lot of hand checking for these different things.

592
00:58:13,310 --> 00:58:15,530
But we can find the sort of modern sources.

593
00:58:15,530 --> 00:58:20,690
There's a lot of Rousseau, there's some Benjamin Franklin, there's some Claude, there's a history of China.

594
00:58:20,900 --> 00:58:25,070
Uh, so it's very syncretic and it's sort of, uh, where these come from.

595
00:58:25,160 --> 00:58:33,020
But there's a whole mass of maxims, uh, of reasons that we, we can't identify and, and that remain unidentifiable.

596
00:58:33,020 --> 00:58:34,400
And so if you just take one day.

597
00:58:35,150 --> 00:58:43,010
So this is the first decade of, uh, one decade of pretty major, uh, you can see the, the different sort of pathways that go through here.

598
00:58:43,010 --> 00:58:44,480
We can't some we haven't found.

599
00:58:44,870 --> 00:58:53,360
It goes from Benjamin Franklin to Voltaire to lawyer, uh, back to Franklin's, uh, Sinek, uh, Benjamin Franklin Duplo, etc.

600
00:58:53,780 --> 00:59:03,139
Uh, so it's a really fascinating work. Uh, it causes you to go in and really get into the text which Dario is proclaimed, as do close reading.

601
00:59:03,140 --> 00:59:08,990
So that's sort of a lovely pun on close reading. Uh, but it's interesting to see how these transformations happen.

602
00:59:09,020 --> 00:59:13,400
Again, there's no indication of the author in this text. There's just the reuse.

603
00:59:13,910 --> 00:59:17,870
Uh, but you can see that there are certain changes that happen for whatever reason in this case,

604
00:59:17,870 --> 00:59:25,489
in the first case with this history of Louis the 12th and Adam is changed into in the Atlantic, a comma changes completely.

605
00:59:25,490 --> 00:59:32,000
The text, it's a different text. So it's it's still the clue is it is a different who does it belong to?

606
00:59:32,570 --> 00:59:35,840
What does it mean? Uh, secondly, Virginia Seaborne.

607
00:59:35,840 --> 00:59:39,680
So this goes back to this notion of originality, less privacy by less pressing.

608
00:59:39,680 --> 00:59:44,690
What is it? So it's more about rationalism. Uh, less than originality.

609
00:59:44,690 --> 00:59:54,380
So this is this sort of hyper textual, um, text, and it gets really wild if you, if you, if you move it out here we have two lines from lucid.

610
00:59:54,920 --> 00:59:58,129
Uh, they appear to be near patterns. Increase their sources.

611
00:59:58,130 --> 01:00:01,280
Seymour, tell who they are and then mark two lines from the. I had you.

612
01:00:01,280 --> 01:00:11,750
Come on. Uh, and also with a little note about you, I had the book from Butler as a feature in, uh, the sixth decade of, uh, number 30.

613
01:00:11,930 --> 01:00:18,410
You have both these lines that have been melded together. Multiple identities, two sources similar to point, but double.

614
01:00:18,770 --> 01:00:24,200
So, Paul, Paul Butler, uh, shows up out of nowhere, just thrown in there.

615
01:00:24,260 --> 01:00:27,860
John. Don't go I don't I won't do it in the future.

616
01:00:28,310 --> 01:00:32,210
Uh, so it's a completely who does this play? Is this corner? Is this Voltaire?

617
01:00:32,510 --> 01:00:36,080
What is Ballard doing there? Uh, what's going on?

618
01:00:36,410 --> 01:00:40,400
Uh, so again, this is this the computer takes you only so far.

619
01:00:40,700 --> 01:00:43,940
Uh, and then you have to just sort of think about what's going on.

620
01:00:44,000 --> 01:00:48,020
And then finally another text that's popped up, which is completely crazy.

621
01:00:48,410 --> 01:00:53,750
And I'll end on this is this text by Jean-Marie Chassagne, who was completely unknown.

622
01:00:53,750 --> 01:00:56,239
It's too bad. Uh, Katrina, Seth isn't here today.

623
01:00:56,240 --> 01:01:02,360
She might be one of the few people with Michelle DeLong who actually knows who this person is, because it showed up in her anthology of.

624
01:01:02,740 --> 01:01:06,280
French poetry from 2008. And he's this sort of illuminator.

625
01:01:06,550 --> 01:01:12,700
He's one of the illuminists or. And Mesmer comes from Lyon and he writes this insane book called The Cataracts of the imagination.

626
01:01:13,030 --> 01:01:17,770
Flood of screwball mania, literary vomit, small encyclopaedic haemorrhage.

627
01:01:18,040 --> 01:01:25,230
Monster of monsters by a minute is the inspired who publishes The Lock of the coffin for years,

628
01:01:25,270 --> 01:01:28,629
so I don't know how you would critically edit that text.

629
01:01:28,630 --> 01:01:31,950
Where do you put it? Where do you map that up?

630
01:01:31,960 --> 01:01:35,230
The vision really gets a little more specific, and it's an insane book,

631
01:01:35,230 --> 01:01:42,520
but it's really interesting because I think it's the culmination of this really open poetics of reuse for the 18th century.

632
01:01:42,520 --> 01:01:46,749
And he has this preface position. So he says, I'm not going to write a preface.

633
01:01:46,750 --> 01:01:52,800
And then he writes 100 pages of preface where it says, I'm just weaving together here prose and verse.

634
01:01:52,810 --> 01:01:56,440
I call on all the annals to my aid. I call on all the authors are review them.

635
01:01:56,440 --> 01:02:02,770
As the new dictator of the literary republic. I indulge in the factions who make this names, etc. so you get sort of where he's going.

636
01:02:02,980 --> 01:02:04,360
My pamphlet fleshed out.

637
01:02:04,570 --> 01:02:11,740
It's four volumes, by the way, not quite a pamphlet fleshed out in this way with historical traits swollen with the simulations of the mind,

638
01:02:11,760 --> 01:02:17,530
studded with quotations of all kinds, becomes a literary mosaic. And this goes back to what I said at the beginning, and you can see it.

639
01:02:17,830 --> 01:02:20,709
He takes great inspiration from one Montana because he said,

640
01:02:20,710 --> 01:02:24,640
Montana is great because he who put a chapter title and then [INAUDIBLE] give you a chapter.

641
01:02:24,640 --> 01:02:27,130
It has nothing to do with the title. And he thinks that's amazing.

642
01:02:27,520 --> 01:02:34,389
Uh, and then so there so you can see the sort of inspiration from Bell in all of his notes, notes to notes.

643
01:02:34,390 --> 01:02:40,090
So he has one for one sentence here. Revolt. The uncle used to keep too young.

644
01:02:40,240 --> 01:02:43,900
Some more. There's a there's there's an asterix note.

645
01:02:44,470 --> 01:02:48,820
There's another note that goes on for two more pages. Just commentary himself.

646
01:02:48,820 --> 01:02:53,800
And he's both. He's a that's a fantastic work. Sometimes he cites where it comes from.

647
01:02:54,250 --> 01:02:58,180
Other times there's nothing. Uh, so again, it's this machine.

648
01:02:58,450 --> 01:03:00,550
It's a, it's a sort of text reuse machine.

649
01:03:00,970 --> 01:03:08,350
Um, I won't read all of his quote, but he gets really into this notion of how he's building, uh, with all the different,

650
01:03:08,740 --> 01:03:14,889
uh, all these different types of texts, what he calls, uh, we have to build a type of encyclopaedia.

651
01:03:14,890 --> 01:03:19,180
So he's taking up this notion of the on site as we come back to it, uh,

652
01:03:19,270 --> 01:03:24,489
we have to use this idea of universality, uh, with a multiplicity of their talents as the poet,

653
01:03:24,490 --> 01:03:28,540
to throw into the same more of the sort of the epigram, the roses, the vegetation, the nettles,

654
01:03:28,540 --> 01:03:34,690
etc. every genre, every author, the entire 18th century should come to bear on this text.

655
01:03:35,080 --> 01:03:37,239
And he continues on this bizarre, overflowing.

656
01:03:37,240 --> 01:03:43,930
So this sort of expansion of wandering thoughts and hidden passages does not require a very strange effort of the imagination.

657
01:03:45,350 --> 01:03:53,480
Uh, imaginative, which I wasn't quite sure what the word is in French, with dictionaries, collections, patient copyists and time.

658
01:03:53,750 --> 01:04:00,979
The most limited scribe is capable of populating the literary universe with encyclopaedic monsters, so to echo hard times,

659
01:04:00,980 --> 01:04:08,660
with the right tools, the right large language model, the right resources, we too could build these sorts of poetic model monsters.

660
01:04:09,050 --> 01:04:14,030
And perhaps that's precisely what language models like GPT are encyclopaedic monsters.

661
01:04:14,910 --> 01:04:21,270
Ditto, of course, already for saw this monstrosity in the making of his own secret belly when he

662
01:04:21,270 --> 01:04:26,580
took stock of their universal universe and sort of universal text in 1755,

663
01:04:27,030 --> 01:04:32,430
far from the perfect system of human understanding that was proposed by Dan on Bear in the in the preface,

664
01:04:32,970 --> 01:04:39,060
in the discord plenum in there, what they had was uneven, unwieldy, and ultimately uncontrollable.

665
01:04:40,110 --> 01:04:45,660
The proof can be found in 100 places in this work. Here we are bloated and exorbitantly fat.

666
01:04:45,900 --> 01:04:48,900
There we are, lean, petty, pathetic, dry and scrawny.

667
01:04:49,230 --> 01:04:53,010
In one place we are skeletal and another we rather seem drop sickle.

668
01:04:53,460 --> 01:05:00,930
We are by turns dwarves and giants, classes and pygmies, standing straight, nicely built and well-proportioned, hunchbacked, lame and deformed.

669
01:05:01,680 --> 01:05:08,579
Add to all these quirks that of a text which is sometimes abstract, recondite and mannered, more often careless,

670
01:05:08,580 --> 01:05:16,110
drawn out and diffuse, and you will compare the work to a monster of the poetic art, or even to something more hideous.

671
01:05:16,110 --> 01:05:22,500
Yet, as the predecessor to Wikipedia and perhaps even the internet, or at the very least the World Wide Web,

672
01:05:23,880 --> 01:05:32,460
and in direct lineage with the other imaginary machines of Charles Babbage and Ada Lovelace, of Alan Turing, of Vannevar Bush.

673
01:05:32,790 --> 01:05:36,809
Uh, the poetics of re-use enacted by the Encyclopaedia, and indeed,

674
01:05:36,810 --> 01:05:42,300
as we've seen over the entirety of the 18th century, is a relatively modern, if not to say contemporary phenomenon.

675
01:05:42,870 --> 01:05:49,170
And we who live now amongst the encyclopaedic monsters of AI, would then do well not to forget our history.

676
01:05:50,350 --> 01:05:53,960
To return very briefly to our illuminated poet.

677
01:05:53,980 --> 01:05:58,210
I tried to find more information of the 19th century biographies of the great French literature.

678
01:05:58,870 --> 01:06:02,050
It just as it was born in died. And it gives a list of his work.

679
01:06:02,680 --> 01:06:08,290
But at the very end there's a very curious note where it says, Mr. Qassim, you left behind a many,

680
01:06:08,410 --> 01:06:13,480
many a manuscript which his brother, a grocer in Leon, used to wrap the medicines in his shop.

681
01:06:14,500 --> 01:06:21,670
Clearly, this is yet another example. Perhaps not so much intractable in computational terms of the poetics of text reuse.

682
01:06:22,150 --> 01:06:22,930
Thank you very much.