Frederic Kaplan: How I built an information time machine

翻译人员: Li Yang 校对人员: Neo Liu

This is an image of the planet Earth.

这是一张地球的图像。

It looks very much like the Apollo pictures

看上去很像那非常有名的

that are very well known.

从阿波罗号上发回的地球图像。

There is something different;

但这里的图像又有一些不同的地方；

you can click on it,

你可以用鼠标点击它，

and if you click on it,

如果你点击了这个图像，

you can zoom in on almost any place on the Earth.

你也就可以放大图像中地球上的几乎任何一个地方。

For instance, this is a bird's-eye view

举个例子来看，这是一张俯瞰

of the EPFL campus.

洛桑联邦理工学院校园的图像。

In many cases, you can also see

通常，你也可以像从附近的街道上一样，

how a building looks from a nearby street.

看看这里的一座建筑是什么样子的。

This is pretty amazing.

这非常令人惊叹。

But there's something missing in this wonderful tour:

但这场绝妙的旅程中似乎又忽略了什么：

It's time.

那便是时间。

i'm not really sure when this picture was taken.

我不确定这张图片是什么时候拍的；

I'm not even sure it was taken

我甚至不确定它是不是

at the same moment as the bird's-eye view.

和前面俯瞰学院的那张照片一起拍的。

In my lab, we develop tools

在我的实验室里，我们在开发一种

to travel not only in space

能让我们不仅在空间里旅行的，

but also through time.

而且在时间中旅行的工具。

The kind of question we're asking is

我们想探讨的问题是

Is it possible to build something

我们有没有可能做出一种像

like Google Maps of the past?

关于过去的谷歌地图一样的东西？

Can I add a slider on top of Google Maps

我能不能在谷歌地图的顶端添加

and just change the year,

一行可以滑动变化年份的时间条，

seeing how it was 100 years before,

来看看这里一百年前是什么样子的，

1,000 years before?

一千年前又是什么样子的？

Is that possible?

这可能吗？

Can I reconstruct social networks of the past?

我能不能重新构建出过去的社交网络？

Can I make a Facebook of the Middle Ages?

我能不能做出中世纪的脸书（Facebook）？

So, can I build time machines?

我，能不能做出时间机器？

Maybe we can just say, "No, it's not possible."

或许我们可以直接说，“不，这不可能的。”

Or, maybe, we can think of it from an information point of view.

又或许我们可以从信息学的角度来思考这个问题。

This is what I call the information mushroom.

我把这个东西叫做信息蘑菇。

Vertically, you have the time.

数轴上是年代，

and horizontally, the amount of digital information available.

横轴上是我们能获得的关于那个年代的数字信息。

Obviously, in the last 10 years, we have much information.

很显然，我们有非常多的关于过去十年的信息，

And obviously the more we go in the past, the less information we have.

但在越来越久远的年代，我们能获得的信息也越来越少。

If we want to build something like Google Maps of the past,

如果我们想做类似过去的谷歌地图、

or Facebook of the past,

过去的脸书（Facebook）这样的东西，

we need to enlarge this space,

我们需要扩大图片中获得的数字信息的区域（橙色部分），

we need to make that like a rectangle.

我们需要让这块区域变成矩形的形状。

How do we do that?

那我们该怎么做呢？

One way is digitization.

一种方法是把所有我们能得到的资料数字化，

There's a lot of material available --

我们手边有很多过去的资料，

newspaper, printed books, thousands of printed books.

从报纸到数千的纸质图书。

I can digitize all these.

我可以把所有这些资料数字化。

I can extract information from these.

我可以从它们之中提取信息。

Of course, the more you go in the past, the less information you will have.

当然，在越来越久远的年代，我们有的资料也越来越少。

So, it might not be enough.

所以，仅仅这样可能还是不够。

So, I can do what historians do.

但我还可以做一些历史学家做的事情。

I can extrapolate.

我可以从拥有的资料信息对未知的那些事实进行推断。

This is what we call, in computer science, simulation.

在计算机学中我们把这个叫做模拟。

If I take a log book,

如果我有一本航海日志，

I can consider, it's not just a log book

我可以这么想，这不仅仅是一本

of a Venetian captain going to a particular journey.

关于一个威尼斯船长某次特定航程的航海日志。

I can consider it is actually a log book

它可以是一本

which is representative of many journeys of that period.

代表了那个年代很多类似航程的航海日志。

I'm extrapolating.

我在做的便是一个推断的过程。

If I have a painting of a facade,

如果我有一张建筑的照片，

I can consider it's not just that particular building,

我可以认为这不仅仅反映的是那一座特定的建筑的特征，

but probably it also shares the same grammar

它可能也反映了同时代那些建筑的特征，

of buildings where we lost any information.

而那些建筑可能正是我们所知甚少的。

So if we want to construct a time machine,

如果我们想建成一个时间机器，

we need two things.

我们需要两样东西。

We need very large archives,

一方面，我们需要大量的档案，

and we need excellent specialists.

另一方面，我们需要一批杰出的专家。

The Venice Time Machine,

我将和你们介绍的便是

the project I'm going to talk to you about,

威尼斯时间机器项目，

is a joint project between the EPFL

这是一个由洛桑联邦理工学院

and the University of Venice Ca'Foscari.

和威尼斯大学合作的项目。

There's something very peculiar about Venice,

关于威尼斯很特别的一点便是，

that its administration has been

它的管理模式

very, very bureaucratic.

非常的官僚主义。

They've been keeping track of everything,

他们几乎记录下这里发生的一切，

almost like Google today.

而这很像今天的谷歌。

At the Archivio di Stato,

在威尼斯的国家档案馆，

you have 80 kilometers of archives

你可以找到80千米长的档案资料，

documenting every aspect

它们记录下了过去一千多年中威尼斯人们

of the life of Venice over more than 1,000 years.

生活的方方面面。

You have every boat that goes out,

你可以了解到每一艘

every boat that comes in.

出入的小船的信息。

You have every change that was made in the city.

你可以了解到这座城市发生的每一丁点儿变化。

This is all there.

它们都被记录在那里。

We are setting up a 10-year digitization program

我们正在开展一个长达十年的数字化项目，

which has the objective of transforming

它的目标就是

this immense archive

把这些海量的档案信息

into a giant information system.

全部转化成一个巨大的信息系统。

The type of objective we want to reach

要达成这个目标，

is 450 books a day that can be digitized.

我们每天要把450本书数字化。

Of course, when you digitize, that's not enough,

当然，仅仅数字化是不够的，

because these documents,

这些档案中，

most of them are in Latin, in Tuscan,

很多是用拉丁语、托斯卡纳语、

in Venetian dialect,

威尼斯方言记录下的。

so you need to transcribe them,

所以你需要转写它们，

to translate them in some cases,

一些情况下你需要翻译它们，

to index them,

你需要将它们编入索引，

and this is obviously not easy.

而这显然不是件容易的事情。

In particular, traditional optical character recognition method

尤其要指出的是，传统的光学字符识别方法

that can be used for printed manuscripts,

虽然对于印刷本可以使用且非常有效，

they do not work well on the handwritten document.

但对于这些手写的档案似乎并不太行之有效。

So the solution is actually to take inspiration

我们的解决方案是从语音识别这个领域

from another domain: speech recognition.

寻找一些启发。

This is a domain of something that seems impossible,

这个领域看上去做的是一些不可能完成的事情，

which can actually be done,

但其实只要加上一些限制条件，

simply by putting additional constraints.

它们完全是可以做到的。

If you have a very good model

如果你有一个关于被使用的语言的

of a language which is used,

很好的模型；

if you have a very good model of a document,

如果你有一个关于一份条理清晰的

how well they are structured.

档案的很好的模型，

And these are administrative documents.

这便是那些行政管理的档案文献，

They are well structured in many cases.

它们通常都有很好的条理；

If you divide this huge archive into smaller subsets

如果你把这些海量的档案划分成一些小的部分，

where a smaller subset actually shares similar features,

其中每一个部分都和其他部分有相近的特征，

then there's a chance of success.

那么我们便有成功的机会。

If we reach that stage, then there's something else:

如果我们到了那个阶段，我们便可以做一些别的事情：

we can extract from this document events.

我们可以从这些档案文献中提取事件。

Actually probably 10 billion events

实际上大概一百亿件事件

can be extracted from this archive.

可以从这些档案中提取出来。

And this giant information system

而这巨大的信息系统

can be searched in many ways.

又可以被很多种方法搜索。

You can ask questions like,

你可以问这样的问题，

"Who lived in this palazzo in 1323?"

“1323年的时候，谁住在这个宫殿里？”

"How much cost a sea bream at the Realto market

“1434年的时候，里亚托的一个集市里，

in 1434?"

海鲷卖多少钱？

"What was the salary

“穆拉诺岛的一个玻璃工人

of a glass maker in Murano

大约十多年前的工资

maybe over a decade?"

是多少？”

You can ask even bigger questions

因为这些信息会被用语义编码，

because it will be semantically coded.

你又可以问一些更宏大的问题。

And then what you can do is put that in space,

然后我们便需要把这些存在于空间中的信息

because much of this information is spatial.

放回到它们原来的空间中去。

And from that, you can do things like

这样，我们便可以

reconstructing this extraordinary journey

重新构建出这场令人惊叹的关于这座城市的旅程，

of that city that managed to have a sustainable development

让它能有一个持续的、

over a thousand years,

超过数千年的发展过程，

managing to have all the time

能让所有的时间

a form of equilibrium with its environment.

和它所在的空间环境达到一种平衡状态。

You can reconstruct that journey,

我们可以重新建构这场旅行，

visualize it in many different ways.

用不同的方法将它图像化。

But of course, you cannot understand Venice if you just look at the city.

但当然，如果我们仅仅考察威尼斯这一座城市，我们便不能做到完全理解它

You have to put it in a larger European context.

我们需要把它放到更大的欧洲的背景下去观察研究。

So the idea is also to document all the things

这便意味着我们需要记录下

that worked at the European level.

在欧洲层面上发生的所有事情。

We can reconstruct also the journey

我们也可以重新建构

of the Venetian maritime empire,

威尼斯海上帝国时期的旅程，

how it progressively controlled the Adriatic Sea,

看它如何一步步控制了亚得里亚海，

how it became the most powerful medieval empire

看它如何变成那个时代

of its time,

最强大的中世纪帝国，

controlling most of the sea routes

它如何几乎控制了从东到南的

from the east to the south.

所有海上航线。

But you can even do other things,

同时，因为这些海上航线

because in these maritime routes,

有着自己的模式和规律，

there are regular patterns.

我们甚至可以做一些别的事情。

You can go one step beyond

我们可以更进一步，

and actually create a simulation system,

创造出一个模拟系统，

create a Mediterranean simulator

模拟出地中海区域的历史，

which is capable actually of reconstructing

这能让我们甚至重建出

even the information we are missing,

我们丢失的信息，

which would enable us to have questions you could ask

能让我们回答出一些别的问题。

like if you were using a route planner.

比如如果你在进行路线规划，你想问，

"If I am in Corfu in June 1323

“如果我在1323年6月科孚岛，

and want to go to Constantinople,

想前往君士坦丁堡，

where can I take a boat?"

我能在哪里找到船？”

Probably we can answer this question

或许我们可以在当时的两三天的精确度内

with one or two or three days' precision.

回答这个问题。

"How much will it cost?"

“它需要多少钱？”

"What are the chance of encountering pirates?"

“遇到海盗的几率有多少？”

Of course, you understand,

当然，你也应该理解，

the central scientific challenge of a project like this one

对于这样一个项目，最核心的科学性质疑便是

is qualifying, quantifying and representing

能否量化出

uncertainty and inconsistency at each step of this process.

它每一步中的不确定性和不一致性。

There are errors everywhere,

因为到处都有错误，

errors in the document, it's the wrong name of the captain,

档案中有错误，或许是船长的名字错了，

some of the boats never actually took to sea.

或许是有一些小船从来没有出过海，

There are errors in translation, interpretative biases,

翻译中也有错误，我们的解释可能有偏差，

and on top of that, if you add algorithmic processes,

最关键的是，如果我们加上算法的过程，

you're going to have errors in recognition,

我们将会在信息识别、

errors in extraction,

信息提取中都存在错误，

so you have very, very uncertain data.

这样我们拥有的便是非常不确定的信息资料。

So how can we detect and correct these inconsistencies?

那我们如何发现并纠正这些偏差呢？

How can we represent that form of uncertainty?

我们如何表示出这种不确定性呢？

It's difficult. One thing you can do

这是非常困难的，我们能做的

is document each step of the process,

便是记录下我们过程中的每一步，

not only coding the historical information

这不仅仅是翻译出那些历史信息，

but what we call the meta-historical information,

而且是翻译出那些我们叫做元历史的信息，

how is historical knowledge constructed,

关于那些历史是如何构建的，

documenting each step.

我们要记录下每一步。

That will not guarantee that we actually converge

这当然不会保证我们真的能汇聚出

toward a single story of Venice,

关于威尼斯的最可靠的过去，

but probably we can actually reconstruct

但或许我们真的能重建出

a fully documented potential story of Venice.

一个可能的威尼斯的过去。

Maybe there's not a single map.

也许不仅仅有一张地图，

Maybe there are several maps.

也许有很多张地图。

The system should allow for that,

这个系统应该承认并接受这些事实，

because we have to deal with a new form of uncertainty,

因为我们必须要面对并处理这种新的不确定性，

which is really new for this type of giant databases.

它对于我们这种巨大的数据库而言确实是非常新的。

And how should we communicate

然后，我们应该怎样

this new research to a large audience?

和更多的人交流我们这项全新的研究呢？

Again, Venice is extraordinary for that.

再一次地，威尼斯在这里非常地特别，它有自己的优势。

With the millions of visitors that come every year,

在威尼斯每年有数百万的游客前来观光，

it's actually one of the best places

这使它实际上变成了

to try to invent the museum of the future.

构建出未来博物馆的最佳的选择之一。

Imagine, horizontally you see the reconstructed map

想象一下，在横轴上你看到某个特定年份的

of a given year,

重新组织建构好的地图，

and vertically, you see the document

在竖轴上，你看到完成这一重建的

that served the reconstruction,

档案资料，

paintings, for instance.

比如说绘画作品。

Imagine an immersive system that permits

想象一下，这样一个身临其境的系统

to go and dive and reconstruct the Venice of a given year,

能让你深入到威尼斯的每一个特定年份去体验，

some experience you could share within a group.

这显然是你应该和他人分享的经历。

On the contrary, imagine actually that you start

另一方面，实际上你体验的这一切

from a document, a Venetian manuscript,

都是从一份档案、一份威尼斯的手稿出发构建的，

and you show, actually, what you can construct out of it,

你看到你能从那些档案资料中得到什么，

how it is decoded,

它们是如何被解读出来的，

how the context of that document can be recreated.

那些档案中的内容又是如何被重现的。

This is an image from an exhibit

这便是这样一件展览品的概念，

which is currently conducted in Geneva

而它现在正在和这种信息系统一起

with that type of system.

在日内瓦进行着。

So to conclude, we can say that

总而言之，我们可以说

research in the humanities is about to undergo

现在研究人类相关的人文学很像

an evolution which is maybe similar

30多年前在生命科学领域

to what happened to life sciences 30 years ago.

发生的一场革命性的变化。

It's really a question of scale.

这真的是个规模的问题。

We see projects which are

我们看到很多项目，

much beyond any single research team can do,

它们在做的远远超过任何一个单一的研究小组，

and this is really new for the humanities,

这对人文学者来说确实是非常新颖的，

which very often take the habit of working

因为他们通常适应于

in small groups or only with a couple of researchers.

在小的团队里工作或仅仅和一些研究者一起工作。

When you visit the Archivio di Stato,

当你参观威尼斯国家档案馆的时候，

you feel this is beyond what any single team can do,

你会觉得这远远超过了任何一个团队能做的事情，

and that should be a joint and common effort.

那应该是共同努力的结果。

So what we must do for this paradigm shift

所以应对这种模式的转换

is actually foster a new generation

我们应该培养出新的一代人，

of "digital humanists"

他们便是“数字人文主义者”，

that are going to be ready for this shift.

他们应该能准备好迎接这种转变。

I thank you very much.

非常感谢。

(Applause)

（鼓掌）