0:00:00.285,0:00:03.178
This is an image of the planet Earth.

0:00:03.178,0:00:06.271
It looks very much like the Apollo pictures

0:00:06.271,0:00:07.882
that are very well known.

0:00:07.882,0:00:09.952
There is something different;

0:00:09.952,0:00:11.399
you can click on it,

0:00:11.399,0:00:12.597
and if you click on it,

0:00:12.597,0:00:15.669
you can zoom in on almost any place on the Earth.

0:00:15.669,0:00:17.668
For instance, this is a bird's-eye view

0:00:17.668,0:00:20.334
of the EPFL campus.

0:00:20.334,0:00:22.442
In many cases, you can also see

0:00:22.442,0:00:26.182
how a building looks from a nearby street.

0:00:26.182,0:00:27.604
This is pretty amazing.

0:00:27.604,0:00:31.031
But there's something missing in this wonderful tour:

0:00:31.031,0:00:33.219
It's time.

0:00:33.219,0:00:36.289
i'm not really sure when this picture was taken.

0:00:36.289,0:00:37.701
I'm not even sure it was taken

0:00:37.701,0:00:43.784
at the same moment as the bird's-eye view.

0:00:43.784,0:00:45.993
In my lab, we develop tools

0:00:45.993,0:00:47.757
to travel not only in space

0:00:47.757,0:00:50.315
but also through time.

0:00:50.315,0:00:52.185
The kind of question we're asking is

0:00:52.185,0:00:53.578
Is it possible to build something

0:00:53.578,0:00:55.756
like Google Maps of the past?

0:00:55.756,0:00:59.066
Can I add a slider on top of Google Maps

0:00:59.066,0:01:00.869
and just change the year,

0:01:00.869,0:01:02.660
seeing how it was 100 years before,

0:01:02.660,0:01:04.329
1,000 years before?

0:01:04.329,0:01:06.452
Is that possible?

0:01:06.452,0:01:08.704
Can I reconstruct social networks of the past?

0:01:08.704,0:01:11.753
Can I make a Facebook of the Middle Ages?

0:01:11.753,0:01:15.529
So, can I build time machines?

0:01:15.529,0:01:18.094
Maybe we can just say, "No, it's not possible."

0:01:18.094,0:01:21.904
Or, maybe, we can think of it from an information point of view.

0:01:21.904,0:01:25.094
This is what I call the information mushroom.

0:01:25.094,0:01:26.677
Vertically, you have the time.

0:01:26.677,0:01:29.417
and horizontally, the amount of digital information available.

0:01:29.417,0:01:32.899
Obviously, in the last 10 years, we have much information.

0:01:32.899,0:01:36.447
And obviously the more we go in the past, the less information we have.

0:01:36.447,0:01:38.765
If we want to build something like Google Maps of the past,

0:01:38.765,0:01:40.259
or Facebook of the past,

0:01:40.259,0:01:41.833
we need to enlarge this space,

0:01:41.833,0:01:43.771
we need to make that like a rectangle.

0:01:43.771,0:01:45.281
How do we do that?

0:01:45.281,0:01:47.379
One way is digitization.

0:01:47.395,0:01:49.174
There's a lot of material available --

0:01:49.190,0:01:55.460
newspaper, printed books, thousands of printed books.

0:01:55.460,0:01:57.228
I can digitize all these.

0:01:57.228,0:01:59.965
I can extract information from these.

0:01:59.965,0:02:03.508
Of course, the more you go in the past,[br]the less information you will have.

0:02:03.508,0:02:06.154
So, it might not be enough.

0:02:06.154,0:02:08.562
So, I can do what historians do.

0:02:08.562,0:02:10.086
I can extrapolate.

0:02:10.086,0:02:14.556
This is what we call, in computer science, simulation.

0:02:14.556,0:02:16.307
If I take a log book,

0:02:16.307,0:02:18.711
I can consider, it's not just a log book

0:02:18.711,0:02:21.683
of a Venetian captain going to a particular journey.

0:02:21.683,0:02:23.326
I can consider it is actually a log book

0:02:23.326,0:02:25.908
which is representative of[br]many journeys of that period.

0:02:25.908,0:02:28.153
I'm extrapolating.

0:02:28.153,0:02:30.191
If I have a painting of a facade,

0:02:30.191,0:02:32.942
I can consider it's not just that particular building,

0:02:32.942,0:02:36.874
but probably it also shares the same grammar

0:02:36.874,0:02:40.915
of buildings where we lost any information.

0:02:40.915,0:02:43.773
So if we want to construct a time machine,

0:02:43.773,0:02:45.112
we need two things.

0:02:45.112,0:02:47.346
We need very large archives,

0:02:47.346,0:02:50.088
and we need excellent specialists.

0:02:50.088,0:02:51.962
The Venice Time Machine,

0:02:51.962,0:02:53.767
the project I'm going to talk to you about,

0:02:53.767,0:02:56.787
is a joint project between the EPFL

0:02:56.787,0:02:59.765
and the University of Venice Ca'Foscari.

0:02:59.765,0:03:01.930
There's something very peculiar about Venice,

0:03:01.930,0:03:04.604
that its administration has been

0:03:04.604,0:03:06.798
very, very bureaucratic.

0:03:06.798,0:03:08.991
They've been keeping track of everything,

0:03:08.991,0:03:11.906
almost like Google today.

0:03:11.906,0:03:13.420
At the Archivio di Stato,

0:03:13.420,0:03:15.184
you have 80 kilometers of archives

0:03:15.184,0:03:17.193
documenting every aspect

0:03:17.193,0:03:19.439
of the life of Venice over[br]more than 1,000 years.

0:03:19.439,0:03:21.359
You have every boat that goes out,

0:03:21.359,0:03:22.435
every boat that comes in.

0:03:22.435,0:03:25.232
You have every change that was made in the city.

0:03:25.232,0:03:28.523
This is all there.

0:03:28.523,0:03:32.431
We are setting up a 10-year digitization program

0:03:32.431,0:03:34.108
which has the objective of transforming

0:03:34.108,0:03:35.492
this immense archive

0:03:35.492,0:03:37.918
into a giant information system.

0:03:37.918,0:03:39.775
The type of objective we want to reach

0:03:39.775,0:03:44.501
is 450 books a day that can be digitized.

0:03:44.501,0:03:46.748
Of course, when you digitize, that's not enough,

0:03:46.748,0:03:48.035
because these documents,

0:03:48.035,0:03:50.674
most of them are in Latin, in Tuscan,

0:03:50.689,0:03:52.204
in Venetian dialect,

0:03:52.204,0:03:53.879
so you need to transcribe them,

0:03:53.879,0:03:55.560
to translate them in some cases,

0:03:55.560,0:03:56.680
to index them,

0:03:56.680,0:03:58.844
and this is obviously not easy.

0:03:58.844,0:04:02.688
In particular, traditional optical[br]character recognition method

0:04:02.688,0:04:04.112
that can be used for printed manuscripts,

0:04:04.112,0:04:08.116
they do not work well on the handwritten document.

0:04:08.116,0:04:10.246
So the solution is actually to take inspiration

0:04:10.246,0:04:13.147
from another domain: speech recognition.

0:04:13.147,0:04:15.202
This is a domain of something [br]that seems impossible,

0:04:15.202,0:04:17.739
which can actually be done,

0:04:17.739,0:04:19.933
simply by putting additional constraints.

0:04:19.933,0:04:21.519
If you have a very good model

0:04:21.519,0:04:23.045
of a language which is used,

0:04:23.045,0:04:25.131
if you have a very good model of a document,

0:04:25.131,0:04:26.563
how well they are structured.

0:04:26.563,0:04:27.916
And these are administrative documents.

0:04:27.931,0:04:30.063
They are well structured in many cases.

0:04:30.063,0:04:33.371
If you divide this huge archive into smaller subsets

0:04:33.371,0:04:36.248
where a smaller subset[br]actually shares similar features,

0:04:36.248,0:04:40.279
then there's a chance of success.

0:04:42.761,0:04:45.196
If we reach that stage, then there's something else:

0:04:45.196,0:04:48.718
we can extract from this document events.

0:04:48.718,0:04:51.016
Actually probably 10 billion events

0:04:51.016,0:04:52.947
can be extracted from this archive.

0:04:52.947,0:04:54.671
And this giant information system

0:04:54.671,0:04:56.487
can be searched in many ways.

0:04:56.487,0:04:57.855
You can ask questions like,

0:04:57.855,0:05:00.615
"Who lived in this palazzo in 1323?"

0:05:00.615,0:05:02.837
"How much cost a sea bream at the Realto market

0:05:02.837,0:05:04.561
in 1434?"

0:05:04.561,0:05:06.021
"What was the salary

0:05:06.021,0:05:08.066
of a glass maker in Murano

0:05:08.066,0:05:09.472
maybe over a decade?"

0:05:09.472,0:05:10.894
You can ask even bigger questions

0:05:10.894,0:05:13.632
because it will be semantically coded.

0:05:13.632,0:05:15.772
And then what you can do is put that in space,

0:05:15.772,0:05:17.945
because much of this information is spatial.

0:05:17.945,0:05:19.880
And from that, you can do things like

0:05:19.880,0:05:21.993
reconstructing this extraordinary journey

0:05:21.993,0:05:25.349
of that city that managed to[br]have a sustainable development

0:05:25.349,0:05:27.475
over a thousand years,

0:05:27.475,0:05:29.095
managing to have all the time

0:05:29.095,0:05:31.956
a form of equilibrium with its environment.

0:05:31.956,0:05:33.204
You can reconstruct that journey,

0:05:33.204,0:05:36.100
visualize it in many different ways.

0:05:36.100,0:05:38.799
But of course, you cannot understand [br]Venice if you just look at the city.

0:05:38.799,0:05:41.195
You have to put it in a larger European context.

0:05:41.195,0:05:44.016
So the idea is also to document all the things

0:05:44.016,0:05:46.439
that worked at the European level.

0:05:46.439,0:05:48.403
We can reconstruct also the journey

0:05:48.403,0:05:50.393
of the Venetian maritime empire,

0:05:50.393,0:05:53.559
how it progressively controlled the Adriatic Sea,

0:05:53.559,0:05:57.305
how it became the most powerful medieval empire

0:05:57.305,0:05:58.866
of its time,

0:05:58.866,0:06:01.038
controlling most of the sea routes

0:06:01.038,0:06:03.971
from the east to the south.

0:06:05.305,0:06:07.621
But you can even do other things,

0:06:07.621,0:06:09.898
because in these maritime routes,

0:06:09.898,0:06:11.873
there are regular patterns.

0:06:11.889,0:06:14.382
You can go one step beyond

0:06:14.382,0:06:16.502
and actually create a simulation system,

0:06:16.502,0:06:19.317
create a Mediterranean simulator

0:06:19.317,0:06:21.910
which is capable actually of reconstructing

0:06:21.910,0:06:24.112
even the information we are missing,

0:06:24.112,0:06:27.100
which would enable us to have [br]questions you could ask

0:06:27.100,0:06:30.088
like if you were using a route planner.

0:06:30.088,0:06:33.159
"If I am in Corfu in June 1323

0:06:33.159,0:06:35.685
and want to go to Constantinople,

0:06:35.685,0:06:37.828
where can I take a boat?"

0:06:37.828,0:06:39.195
Probably we can answer this question

0:06:39.195,0:06:43.668
with one or two or three days' precision.

0:06:43.668,0:06:45.275
"How much will it cost?"

0:06:45.275,0:06:48.867
"What are the chance of encountering pirates?"

0:06:48.867,0:06:50.678
Of course, you understand,

0:06:50.678,0:06:53.287
the central scientific challenge[br]of a project like this one

0:06:53.287,0:06:57.016
is qualifying, quantifying and representing

0:06:57.016,0:07:00.346
uncertainty and inconsistency[br]at each step of this process.

0:07:00.346,0:07:03.058
There are errors everywhere,

0:07:03.058,0:07:05.547
errors in the document, it's[br]the wrong name of the captain,

0:07:05.547,0:07:08.760
some of the boats never actually took to sea.

0:07:08.760,0:07:13.617
There are errors in translation, interpretative biases,

0:07:13.624,0:07:17.090
and on top of that, if you add algorithmic processes,

0:07:17.090,0:07:20.039
you're going to have errors in recognition,

0:07:20.039,0:07:22.000
errors in extraction,

0:07:22.000,0:07:26.481
so you have very, very uncertain data.

0:07:26.481,0:07:30.238
So how can we detect and[br]correct these inconsistencies?

0:07:30.238,0:07:33.898
How can we represent that form of uncertainty?

0:07:33.898,0:07:35.995
It's difficult. One thing you can do

0:07:35.995,0:07:38.221
is document each step of the process,

0:07:38.221,0:07:40.669
not only coding the historical information

0:07:40.669,0:07:43.348
but what we call the meta-historical information,

0:07:43.348,0:07:46.011
how is historical knowledge constructed,

0:07:46.011,0:07:48.009
documenting each step.

0:07:48.009,0:07:49.654
That will not guarantee that we actually converge

0:07:49.654,0:07:52.104
toward a single story of Venice,

0:07:52.104,0:07:54.242
but probably we can actually reconstruct

0:07:54.242,0:07:57.290
a fully documented potential story of Venice.

0:07:57.290,0:07:58.749
Maybe there's not a single map.

0:07:58.749,0:08:00.869
Maybe there are several maps.

0:08:00.869,0:08:03.085
The system should allow for that,

0:08:03.085,0:08:05.944
because we have to deal with[br]a new form of uncertainty,

0:08:05.944,0:08:10.585
which is really new for this type of giant databases.

0:08:10.585,0:08:12.775
And how should we communicate

0:08:12.790,0:08:16.769
this new research to a large audience?

0:08:16.769,0:08:19.432
Again, Venice is extraordinary for that.

0:08:19.432,0:08:21.603
With the millions of visitors that come every year,

0:08:21.603,0:08:23.366
it's actually one of the best places

0:08:23.366,0:08:26.354
to try to invent the museum of the future.

0:08:26.354,0:08:29.658
Imagine, horizontally you see the reconstructed map

0:08:29.658,0:08:30.944
of a given year,

0:08:30.944,0:08:33.902
and vertically, you see the document

0:08:33.902,0:08:35.413
that served the reconstruction,

0:08:35.413,0:08:38.813
paintings, for instance.

0:08:38.813,0:08:41.393
Imagine an immersive system that permits

0:08:41.393,0:08:44.895
to go and dive and reconstruct[br]the Venice of a given year,

0:08:44.895,0:08:47.610
some experience you could share within a group.

0:08:47.610,0:08:49.856
On the contrary, imagine actually that you start

0:08:49.856,0:08:52.063
from a document, a Venetian manuscript,

0:08:52.063,0:08:55.112
and you show, actually, what[br]you can construct out of it,

0:08:55.112,0:08:56.884
how it is decoded,

0:08:56.884,0:08:59.299
how the context of that document can be recreated.

0:08:59.299,0:09:01.184
This is an image from an exhibit

0:09:01.184,0:09:03.460
which is currently conducted in Geneva

0:09:03.460,0:09:05.814
with that type of system.

0:09:05.814,0:09:07.989
So to conclude, we can say that

0:09:07.989,0:09:11.068
research in the humanities is about to undergo

0:09:11.068,0:09:12.870
an evolution which is maybe similar

0:09:12.870,0:09:17.452
to what happened to life sciences 30 years ago.

0:09:17.452,0:09:22.128
It's really a question of scale.

0:09:22.130,0:09:25.433
We see projects which are

0:09:25.433,0:09:29.276
much beyond any single research team can do,

0:09:29.276,0:09:31.519
and this is really new for the humanities,

0:09:31.519,0:09:35.388
which very often take the habit of working

0:09:35.388,0:09:39.396
in small groups or only with a couple of researchers.

0:09:39.396,0:09:41.514
When you visit the Archivio di Stato,

0:09:41.514,0:09:44.336
you feel this is beyond what any single team can do,

0:09:44.336,0:09:48.170
and that should be a joint and common effort.

0:09:48.170,0:09:51.276
So what we must do for this paradigm shift

0:09:51.276,0:09:53.178
is actually foster a new generation

0:09:53.178,0:09:54.715
of "digital humanists"

0:09:54.715,0:09:56.805
that are going to be ready for this shift.

0:09:56.805,0:09:58.764
I thank you very much.

0:09:58.764,0:10:02.764
(Applause)