WEBVTT

00:00:26.335 --> 00:00:27.831
I'm here to tell you

00:00:28.021 --> 00:00:31.441
why I don't tell the truth about castles.

00:00:33.211 --> 00:00:34.913
You might think it's my job.

00:00:34.913 --> 00:00:36.664
After all, we expect professionals

00:00:36.664 --> 00:00:39.890
to speak with authority
and give us clear-cut solutions,

00:00:40.110 --> 00:00:42.110
and that makes us very, very nervous

00:00:42.110 --> 00:00:45.476
because there's so much
we simply don't know about history.

00:00:45.926 --> 00:00:46.932
And as a result,

00:00:46.932 --> 00:00:51.055
a lot of things have become established
in our collective memory as the truth

00:00:51.055 --> 00:00:53.342
simply because someone said it once,

00:00:53.342 --> 00:00:55.101
it sounded convincing,

00:00:55.101 --> 00:00:56.866
and nobody since has stood up to say,

00:00:56.866 --> 00:00:59.627
"Well, we don't know exactly
what it was like,

00:00:59.627 --> 00:01:01.404
but it wasn't like that."

00:01:03.124 --> 00:01:05.169
Take Greek temples.

00:01:05.169 --> 00:01:08.664
Everyone knows they're made
of beautiful shining white marble.

00:01:08.664 --> 00:01:12.383
We've seen them that way for centuries,
from postcards to museums,

00:01:12.863 --> 00:01:16.060
and that establishes certain
seeing habits in our heads,

00:01:16.650 --> 00:01:21.724
where we've seen them this way so much
that anything different just looks wrong.

00:01:22.094 --> 00:01:24.982
And yet today we know for a fact

00:01:24.982 --> 00:01:27.342
that they were painted
in bright garish colors;

00:01:27.342 --> 00:01:30.049
we're just a little unclear
on some of the details.

00:01:32.059 --> 00:01:35.856
I've colored this one in myself
in about five minutes of research,

00:01:35.856 --> 00:01:38.771
so it's likely to be wrong
in all the relevant places,

00:01:38.791 --> 00:01:41.772
and it's still more correct
than the white one.

00:01:42.722 --> 00:01:45.821
So why do we continue
to show them in white?

00:01:45.821 --> 00:01:47.497
Well, there's two reasons for that:

00:01:47.497 --> 00:01:50.907
One is that we as humans like certainty.

00:01:51.187 --> 00:01:55.070
And so we would prefer
to be absolutely certain

00:01:55.070 --> 00:01:58.069
even if it's the absolute certainty
that we are absolutely wrong

00:01:58.069 --> 00:01:59.069
(Laughter)

00:01:59.069 --> 00:02:01.118
than to say, "Well,

00:02:01.118 --> 00:02:04.033
maybe it could have been approximately,

00:02:04.033 --> 00:02:05.974
I think, something like …"

00:02:05.974 --> 00:02:07.170
And then there's the fact

00:02:07.170 --> 00:02:10.466
that when we're trying to establish
a new truth in people's heads,

00:02:10.466 --> 00:02:13.228
we want it to be
the correct truth this time.

00:02:13.988 --> 00:02:16.607
But even if we're not
entirely clear on all the details,

00:02:16.607 --> 00:02:18.754
that doesn't mean
we can't make a statement.

00:02:19.384 --> 00:02:21.416
If you ask me right now what time it is,

00:02:21.416 --> 00:02:22.878
I can't tell you,

00:02:23.248 --> 00:02:27.525
but I don't have to shrug my shoulders
and just say, "I have no idea."

00:02:28.045 --> 00:02:31.078
I know this event is on
from 12:00 till 6:00,

00:02:31.078 --> 00:02:33.394
so that eliminates
half the clock right there.

00:02:34.394 --> 00:02:37.368
We've had our first coffee break,
we've not had the second,

00:02:37.368 --> 00:02:39.260
so it's between 2:00 and 5:00.

00:02:40.070 --> 00:02:44.710
I know there were people ahead of me,
and I'm not being told I'm out of time,

00:02:45.190 --> 00:02:47.762
so it must be around 4:30.

00:02:49.362 --> 00:02:51.039
Is that correct?

00:02:51.039 --> 00:02:51.988
I don't know.

00:02:51.988 --> 00:02:55.420
It might not be the truth,
but I don't have to tell you the truth.

00:02:55.420 --> 00:02:58.317
I just have to know
how correct I'm likely to be

00:02:58.587 --> 00:03:01.941
because how correct I am
can be very, very important.

00:03:02.210 --> 00:03:04.587
Me telling you it's about 4:30
is pretty useless

00:03:04.587 --> 00:03:07.249
if you want to know
whether you can still catch your bus;

00:03:07.249 --> 00:03:10.018
and in that case,
we might have to ask more people,

00:03:10.018 --> 00:03:12.541
we might have to fill it in
with more clues and so on;

00:03:12.541 --> 00:03:13.767
and that's science.

00:03:13.767 --> 00:03:15.217
We ask a question,

00:03:15.217 --> 00:03:18.744
and then we fill in the unknown
and get more and more precise.

00:03:19.134 --> 00:03:21.940
So the scientific method
is pretty well-established:

00:03:22.090 --> 00:03:24.563
You ask a question
about the world around you,

00:03:24.563 --> 00:03:27.147
you research what
you already know about it,

00:03:27.147 --> 00:03:31.228
you design an experiment
to test what you don't know about it,

00:03:31.438 --> 00:03:34.571
you gather the data, you analyze them,
and you reach a conclusion;

00:03:34.761 --> 00:03:37.241
and that conclusion could be,
"I need more data."

00:03:37.411 --> 00:03:39.694
Then you go back,
design another experiment,

00:03:39.694 --> 00:03:41.459
run it again, gather more data,

00:03:41.459 --> 00:03:43.810
and you get more data
and more data and more data,

00:03:43.810 --> 00:03:45.482
and suddenly you're buried in data,

00:03:45.482 --> 00:03:47.224
and you're dealing with big data,

00:03:47.224 --> 00:03:49.423
where scientists now have this problem

00:03:49.423 --> 00:03:51.541
that there's so many data

00:03:51.541 --> 00:03:54.056
they can never read them all
in one lifetime.

00:03:54.056 --> 00:03:56.230
They have to find new ways
to deal with that.

00:03:56.890 --> 00:03:58.241
And then there's me.

00:03:58.971 --> 00:04:00.290
This is me.

00:04:00.290 --> 00:04:03.353
You can tell I'm not
the kind of scientist with a lab coat,

00:04:03.353 --> 00:04:06.337
and my data problem is slightly different.

00:04:06.557 --> 00:04:10.120
Basically I'm dealing
with one student's lab report

00:04:10.120 --> 00:04:11.940
that they dropped on the floor,

00:04:11.940 --> 00:04:14.163
lost half the pages
and then shuffled the rest,

00:04:14.163 --> 00:04:16.854
and there's probably a coffee stain
on the relevant bit.

00:04:16.934 --> 00:04:19.900
So what I've got is I've got
half a broken castle,

00:04:20.483 --> 00:04:22.143
slightly burned,

00:04:23.066 --> 00:04:26.968
I've got a legal contract from 1388
that was written by a guy

00:04:26.968 --> 00:04:30.872
who managed to spell the name "Arnold"
four different ways in three pages,

00:04:31.672 --> 00:04:33.387
I've got some rocks from the village

00:04:33.387 --> 00:04:36.433
that may or may not have
belonged to this castle,

00:04:37.413 --> 00:04:39.278
I have got a map
that was done by a guy

00:04:39.278 --> 00:04:42.775
for whom this was a 10-minute squiggle
in an eighth-year campaign,

00:04:43.345 --> 00:04:45.689
a painting that was drawn

00:04:45.689 --> 00:04:49.026
about 300 years
after the castle burnt down,

00:04:49.356 --> 00:04:51.822
and a book that was probably propaganda.

00:04:52.252 --> 00:04:54.402
And of course I could go
look in the archives,

00:04:54.402 --> 00:04:57.261
I can get another archaeological
excavation going, and so on,

00:04:57.261 --> 00:05:00.213
but at some point, there's simply
no way to gather more data.

00:05:00.213 --> 00:05:02.507
And then you expect me to take that

00:05:02.507 --> 00:05:05.864
and mash it all up
into the truth about castles?

00:05:09.304 --> 00:05:11.607
You want a reconstruction
that's so realistic

00:05:11.607 --> 00:05:13.232
it feels like you're really there,

00:05:13.232 --> 00:05:16.826
like every little pebble
in the courtyard is just right.

00:05:17.686 --> 00:05:19.969
There's a reason that a lot
of sites and museums

00:05:19.969 --> 00:05:22.214
don't use the word "reconstruction";

00:05:22.214 --> 00:05:23.764
instead, you find a picture,

00:05:23.764 --> 00:05:27.534
and next to it, it has the disclaimer
"Artist's impression."

00:05:28.324 --> 00:05:30.657
And that doesn't mean
they didn't do any research;

00:05:30.657 --> 00:05:33.276
it just means they didn't document
what they researched.

00:05:33.276 --> 00:05:35.895
We don't know who they talked to,
which books they read,

00:05:35.895 --> 00:05:37.648
which conclusions they drew,

00:05:37.648 --> 00:05:40.330
and which other theories they discarded.

00:05:40.330 --> 00:05:42.046
Now, imagine for a moment

00:05:42.046 --> 00:05:44.990
that we would treat a text the same way.

00:05:44.990 --> 00:05:46.092
You go into the museum.

00:05:46.092 --> 00:05:49.602
There's a plaque, and it says,
"Author's impression."

00:05:49.602 --> 00:05:52.528
The author thinks there might
have been a castle here.

00:05:52.738 --> 00:05:55.161
You wouldn't take that very seriously.

00:05:55.161 --> 00:05:59.127
So why do we treat text
so differently from models?

00:05:59.127 --> 00:06:03.742
It's because we've come to a consensus
on what makes a scientific text,

00:06:03.742 --> 00:06:05.636
and it's quite simply this.

00:06:05.956 --> 00:06:08.110
When you're writing
a scientific document,

00:06:08.110 --> 00:06:09.937
you put in footnotes,

00:06:09.937 --> 00:06:12.482
you cite works by previous scholars,

00:06:12.507 --> 00:06:14.405
you show your argumentation -

00:06:15.675 --> 00:06:18.329
you simply give your document provenance -

00:06:19.139 --> 00:06:21.783
because showing you a picture of the truth

00:06:22.453 --> 00:06:25.773
isn't going to help you
without me explaining why it's true.

00:06:26.423 --> 00:06:30.088
The truth is, all of these
are correct at the same time.

00:06:30.668 --> 00:06:33.295
That's the truth,
but it's not a very helpful truth,

00:06:34.585 --> 00:06:38.560
because without context,
data are not information.

00:06:38.560 --> 00:06:40.669
So I'll give you a little context.

00:06:43.707 --> 00:06:45.305
So for a little context,

00:06:45.305 --> 00:06:48.497
this first clock shows
the time in Luxembourg,

00:06:48.497 --> 00:06:50.766
and the second one
has the time in Tokyo,

00:06:51.116 --> 00:06:53.215
the third one is one
of those annoying clocks

00:06:53.215 --> 00:06:55.501
everyone had in their kitchens
about 10 years ago

00:06:55.501 --> 00:06:57.164
that actually run counterclockwise,

00:06:57.164 --> 00:06:59.644
and the fourth one is not a clock,
it's a barometer -

00:06:59.644 --> 00:07:01.777
you just wouldn't know that
by looking at it.

00:07:03.577 --> 00:07:04.776
So in historic research,

00:07:04.776 --> 00:07:07.424
when we deal with images,
we know what to do:

00:07:07.424 --> 00:07:10.977
We give those provenance
through metadata and paradata.

00:07:11.457 --> 00:07:13.487
Metadata you've probably heard.

00:07:13.507 --> 00:07:16.044
Metadata are data about the data.

00:07:16.044 --> 00:07:18.523
You can see those when
you're browsing your computer,

00:07:18.523 --> 00:07:20.859
and you can see who made a file,
when it was made,

00:07:20.859 --> 00:07:22.704
when it was last opened, and so on.

00:07:23.104 --> 00:07:25.070
Paradata are slightly more complex.

00:07:25.070 --> 00:07:28.137
Paradata are data
that give context for the data,

00:07:28.327 --> 00:07:30.943
so like how they were gathered,
how they were processed,

00:07:30.943 --> 00:07:33.306
which decisions were made
about them, and so on.

00:07:34.406 --> 00:07:35.953
The metadata for this image

00:07:35.953 --> 00:07:40.300
would be that it was taken by me
on the first of June, 2017

00:07:40.300 --> 00:07:42.568
on a Sony compact camera.

00:07:42.568 --> 00:07:47.362
The paradata are that it was
picture 111 in a series of 128

00:07:47.362 --> 00:07:50.564
and I took it on my first
research trip to this castle.

00:07:51.314 --> 00:07:53.232
And I love to show this picture

00:07:53.232 --> 00:07:57.826
because this picture has everything in it
that is wrong with models.

00:07:58.921 --> 00:08:01.786
You walk up the stairs in this castle,
you come to the attic,

00:08:01.786 --> 00:08:04.502
and there's a big glass box
with this model sitting in it.

00:08:04.502 --> 00:08:05.645
And what I love about it

00:08:05.645 --> 00:08:09.262
is that there are no data
attached to it whatsoever.

00:08:09.262 --> 00:08:10.918
You don't have a scale bar.

00:08:11.958 --> 00:08:15.025
You don't have a date
it was made or who made it.

00:08:15.055 --> 00:08:17.355
You don't have a date
it's supposed to represent.

00:08:17.355 --> 00:08:18.647
There's nothing even to say

00:08:18.647 --> 00:08:21.603
that it's supposed to be this castle
that you're standing in.

00:08:22.483 --> 00:08:24.959
And if you're talking
about decision-making processes

00:08:24.959 --> 00:08:26.018
in the reconstruction,

00:08:26.018 --> 00:08:28.565
if you take a closer look
at that center tower there,

00:08:28.565 --> 00:08:30.326
it becomes very, very obvious

00:08:30.326 --> 00:08:33.896
the size of that tower was not based
on an archeological excavation

00:08:33.896 --> 00:08:36.386
or because there was
a foundation there or something.

00:08:36.386 --> 00:08:38.923
No, that's the size
of the toilet paper roll they had.

00:08:38.923 --> 00:08:41.138
(Laughter)

00:08:41.138 --> 00:08:43.393
And so this model makes me happy

00:08:43.393 --> 00:08:45.851
because it's everything
I'm trying to avoid.

00:08:48.151 --> 00:08:51.113
And I'm not the only person
trying to avoid this kind of thing.

00:08:51.113 --> 00:08:54.283
A lot of intelligent people
are working and avoiding this.

00:08:54.593 --> 00:08:57.540
There are some hugely
complex systems these days

00:08:57.540 --> 00:08:59.891
that go into great detail on data,

00:08:59.891 --> 00:09:02.775
metadata, paradata,
how they all relate, and so forth;

00:09:02.775 --> 00:09:05.954
and my favorite one
takes about six months to learn.

00:09:06.514 --> 00:09:08.724
Now that's bad enough for me
as a researcher,

00:09:08.724 --> 00:09:10.790
but imagine that you,
as a museum visitor,

00:09:10.790 --> 00:09:12.839
have to go on a six-month
training course

00:09:12.839 --> 00:09:14.852
to understand what you're seeing.

00:09:15.522 --> 00:09:19.772
So, instead, I have a system
that's just good enough for me.

00:09:19.772 --> 00:09:21.686
I simply take my model,

00:09:21.686 --> 00:09:26.792
and I tell you which parts are true
and which ones are not.

00:09:26.792 --> 00:09:28.893
So probably true is the easiest.

00:09:28.893 --> 00:09:31.826
That's the category of things
that I think are true

00:09:31.826 --> 00:09:33.858
because they're still there,

00:09:33.858 --> 00:09:36.567
so that could be things
like the castle ruins.

00:09:37.757 --> 00:09:39.696
Next, pretty close to true,

00:09:40.981 --> 00:09:42.696
we have a lot of evidence for those.

00:09:42.696 --> 00:09:43.699
So for example,

00:09:43.699 --> 00:09:46.621
I was saying foundations,
towers on foundations -

00:09:46.621 --> 00:09:49.844
we fill in the gaps
what we have good evidence.

00:09:50.254 --> 00:09:53.962
Third stage, extrapolation,
could be true - maybe not.

00:09:53.962 --> 00:09:56.583
That's where I'm working
on secondary and tertiary data,

00:09:56.583 --> 00:09:58.253
like the maps and images.

00:09:58.253 --> 00:10:00.360
And then there's my favorite category -

00:10:01.015 --> 00:10:03.495
the stuff that's not really true.

00:10:04.900 --> 00:10:07.399
Now, these things I need
to put in my model

00:10:07.709 --> 00:10:10.328
because the model would be
missing something without it.

00:10:10.328 --> 00:10:13.094
If I didn't put these in,
I would be telling you a lie,

00:10:14.014 --> 00:10:16.142
but I have no idea what to really put in.

00:10:16.482 --> 00:10:17.816
It's an interesting problem.

00:10:17.816 --> 00:10:21.882
So that's things like I know
the great hall had paintings on the walls,

00:10:21.882 --> 00:10:24.259
I will never know what exactly
was painted on them,

00:10:24.259 --> 00:10:25.829
so I have to make something up,

00:10:25.829 --> 00:10:28.628
but if I left them as a blank stone
the way they are now,

00:10:28.628 --> 00:10:30.668
that would be making a statement.

00:10:31.398 --> 00:10:34.773
And then, of course, I need to attach
my metadata and my paradata,

00:10:34.773 --> 00:10:37.738
and tell you why it's in that category.

00:10:38.058 --> 00:10:41.174
And finally, I need
to make very, very sure

00:10:41.174 --> 00:10:43.531
that you don't only know
why it's in that category

00:10:43.531 --> 00:10:45.892
but which part exactly
I'm talking about.

00:10:46.492 --> 00:10:49.142
If you remember that clock from earlier,

00:10:49.472 --> 00:10:52.769
well, I can tell you for a fact
that it's Friday afternoon.

00:10:53.089 --> 00:10:55.788
I can also tell you
with absolute certainty

00:10:55.788 --> 00:10:59.842
that sometime in the last two millennia,
we had a castle on this hill.

00:11:00.542 --> 00:11:02.537
What I cannot tell you

00:11:02.537 --> 00:11:06.119
is whether in that window, in 1548,

00:11:06.119 --> 00:11:09.719
we had an archway
and that archway had a stone

00:11:09.719 --> 00:11:12.864
and that stone was exactly
312 millimeters wide.

00:11:13.234 --> 00:11:15.101
It could have been 317,

00:11:15.101 --> 00:11:17.800
but my drawing is going to say
one way or the other.

00:11:19.600 --> 00:11:23.669
And that is the really, really
interesting point for future researchers

00:11:23.949 --> 00:11:26.531
because if I've told you
I have no idea what was here,

00:11:26.531 --> 00:11:29.530
they can use that point to research,
and then they can say,

00:11:29.530 --> 00:11:32.540
"Look, we found more data,
and actually you're completely wrong.

00:11:32.540 --> 00:11:35.025
It was 483 millimeters."

00:11:35.025 --> 00:11:36.569
And I can say, "Hooray!"

00:11:36.569 --> 00:11:39.615
because that advances
our state of collective knowledge.

00:11:40.025 --> 00:11:42.127
So if I'm doing science properly,

00:11:42.127 --> 00:11:44.751
I want people to be able
to prove me wrong.

00:11:46.390 --> 00:11:49.887
So that's why I'm not going to tell you
the truth about castles,

00:11:51.454 --> 00:11:54.344
and why I make it
very, very clear to you

00:11:54.804 --> 00:11:56.336
when I'm just making it up.

00:11:56.336 --> 00:11:57.707
(Laughter)

00:11:58.487 --> 00:11:59.740
Thank you.

00:11:59.740 --> 00:12:02.454
(Applause)