Dan Berkenstock: The world is one big dataset. Now, how to photograph it ...

翻译人员: RUI FAN 校对人员: Shiwen He

Five years ago, I was a Ph.D. student

五年前，我是个博士生

living two lives.

过着两种生活

In one, I used NASA supercomputers

在一种生活里 我用NASA（美国国家航空航天局）的超级电脑

to design next-generation spacecraft,

设计新一代的航天器

and in the other I was a data scientist

在另一种生活里，我是一个数据科学家

looking for potential smugglers

寻找那些敏感核技术的

of sensitive nuclear technologies.

潜在走私者

As a data scientist, I did a lot of analyses,

作为一个数据科学家，我做了很多的分析

mostly of facilities,

大部分是对设施的分析

industrial facilities around the world.

世界各地的工业设施

And I was always looking for a better canvas

而我一直在寻找一个可以更好的框架

to tie these all together.

可以将这些都联结起来

And one day, I was thinking about how

有一天，我正在思考

all data has a location,

所有的数据是如何定位的

and I realized that the answer

然后我意识到

had been staring me in the face.

答案一直就在我的眼前

Although I was a satellite engineer,

虽然我是一个卫星工程师

I hadn't thought about using satellite imagery

我却从来没有想过将卫星图像

in my work.

运用到我的工作中

Now, like most of us, I'd been online,

现在，我会像大多数人一样

I'd see my house, so I thought,

上网看看自己的房子，于是我想

I'll hop in there and I'll start looking up

我也可以开始研究研究

some of these facilities.

其中的一些设施

And what I found really surprised me.

结果我被自己看到的震惊了

The pictures that I was finding

那些我看到的图像

were years out of date,

已经过时好多年了

and because of that,

也因为如此

it had relatively little relevance

它跟我当时的工作

to the work that I was doing today.

几乎没什么关联

But I was intrigued.

不过我很着迷

I mean, satellite imagery is pretty amazing stuff.

大家都知道，卫星图像是非常了不起的东西

There are millions and millions of sensors

今天我们身边有成千上百万的

surrounding us today,

传感器

but there's still so much we don't know on a daily basis.

然而每天都还有那么多 我们不知道的东西

How much oil is stored in all of China?

中国藏有多少石油？

How much corn is being produced?

有多少玉米被生产出来？

How many ships are in all of our world's ports?

全世界的所有港口有多少船只？

Now, in theory, all of these questions

理论上，所有的这些问题

could be answered by imagery,

都可以用图像来回答

but not if it's old.

不过如果是老的图像就不行了

And if this data was so valuable,

并且如果这些数据真的那么重要

then how come I couldn't get my hands

那为什么我不可以获得

on more recent pictures?

更新的图像呢？

So the story begins over 50 years ago

故事始于50多年前

with the launch of the first generation

美国政府发射的

of U.S. government photo reconnaissance satellites.

第一代照片侦查卫星

And today, there's a handful

如今，还有少数

of the great, great grandchildren

这些早期冷战时期的机器的

of these early Cold War machines

曾子孙们

which are now operated by private companies

它们现在多由私营公司运作

and from which the vast majority of satellite imagery

也是我们今天日常看到的

that you and I see on a daily basis comes.

绝大部分的卫星图像的来源

During this period, launching things into space,

在此期间，把东西发射到太空

just the rocket to get the satellite up there,

仅仅是用火箭将一个卫星送上去

has cost hundreds of millions of dollars each,

一次就要花掉数亿美元，十亿啊

and that's created tremendous pressure

这种强大的压力下 不能过于频繁地

to launch things infrequently

发射东西

and to make sure that when you do,

而当你发射东西的时候

you cram as much functionality in there as possible.

就要保证塞进去的功能越多越好

All of this has only made satellites

所有的这一切都导致了卫星

bigger and bigger and bigger

变得越来越大

and more expensive,

也越来越贵

now nearly a billion, with a b, dollars per copy.

如今每个要花掉近10亿美元

Because they are so expensive,

因为它们如此昂贵

there aren't very many of them.

所以数量就不是很多

Because there aren't very many of them,

因为数量不是很多

the pictures that we see on a daily basis

我们日常看到的图像

tend to be old.

就偏陈旧

I think a lot of people actually understand this anecdotally,

我想很多人可以理解这个故事

but in order to visualize just how sparsely

不过为了能让大家更好地想象

our planet is collected,

这些卫星图像有多么稀疏

some friends and I put together a dataset

我和我的一些朋友制作了一个数据库

of the 30 million pictures that have been gathered

这个数据库里有从2000年到2010年

by these satellites between 2000 and 2010.

这些卫星收集的3000万张图像

As you can see in blue, huge areas of our world

你可以看到蓝色的区域 全球有大面积的地方

are barely seen, less than once a year,

几乎看不到 图像少于一年一次

and even the areas that are seen most frequently,

即使是那些最经常看到的区域

those in red, are seen at best once a quarter.

红色的区域 也只是最多每个季度看到一次

Now as aerospace engineering grad students,

这时，作为航空航天工程专业的研究生

this chart cried out to us as a challenge.

这个图给我们发起了一个挑战

Why do these things have to be so expensive?

为什么这些东西要如此昂贵呢？

Does a single satellite really have to cost

单单一个卫星真的必须花费

the equivalent of three 747 jumbo jets?

等同于3架747大型喷气式客机的费用么？

Wasn't there a way to build a smaller,

难道没有办法造一个更小的

simpler, new satellite design that could enable

更简单的、更新的卫星

more timely imaging?

来提供即时的图像么？

I realize that it does sound a little bit crazy

我知道这听上去有些疯狂

that we were going to go out and just

我们准备自己动手

begin designing satellites,

开始设计卫星

but fortunately we had help.

不过幸运的是，我们有帮手

In the late 1990s, a couple of professors

在90年代后期，几个教授

proposed a concept for radically reducing the price

提出了一个概念，可以大幅降低

of putting things in space.

将东西发射到太空的成本

This was hitchhiking small satellites

即在发射大卫星的时候

alongside much larger satellites.

顺带稍上小卫星

This dropped the cost of putting objects up there

这让送东西上太空的成本

by over a factor of 100,

降低了100倍

and suddenly we could afford to experiment,

于是突然之间，我们可以做更多的实验

to take a little bit of risk,

冒更多的风险

and to realize a lot of innovation.

追求更多的创新

And a new generation of engineers and scientists,

新一代的工程师和科学家们

mostly out of universities,

他们大部分来自大学

began launching these very small,

开始打造这些很小的

breadbox-sized satellites called CubeSats.

面包盒大小的“立方体卫星”

And these were built with electronics obtained

这些卫星是用从 RadioShack公司（一家美国电子产品连锁店）

from RadioShack instead of Lockheed Martin.

而不是洛克希德·马丁公司（一家美国航空航天制造商） 买的电子部件打造的

Now it was using the lessons learned from these early missions

利用从这些早期发射中学到的经验

that my friends and I began a series of sketches

我和我的朋友开始设计

of our own satellite design.

我们自己的卫星

And I can't remember a specific day

我不记得具体是哪一天

where we made a conscious decision

我们突然做了这样一个有意识的决定

that we were actually going to go out and build these things,

就是我们真的要打造这些东西

but once we got that idea in our minds

不过一旦我们有了这些想法——

of the world as a dataset,

这个世界是一个数据库

of being able to capture millions of data points

每天我们都能捕捉数百万个数据点

on a daily basis describing the global economy,

来描绘全球经济

of being able to unearth billions of connections

我们能够发掘它们之间

between them that had never before been found,

从来没被发现过的、无数的关联——

it just seemed boring

其它一切的工作

to go work on anything else.

似乎都显得无趣

And so we moved into a cramped,

于是我们搬到了帕洛阿尔托 （美国加州的城市，科技创业公司的天堂）

windowless office in Palo Alto,

一个拥挤的、没有窗户的办公室

and began working to take our design

并开始将我们的设计

from the drawing board into the lab.

从图版转到实验室

The first major question we had to tackle

我们首先要对付的一大问题就是

was just how big to build this thing.

我们要造一个多大的东西

In space, size drives cost,

在太空，体积决定成本

and we had worked with these very small,

在学校，我们使用这些很小的、

breadbox-sized satellites in school,

面包盒大小的卫星

but as we began to better understand the laws of physics,

不过当我们开始更好地理解了物理定律

we found that the quality of pictures

我们发现这些卫星能够获取的

those satellites could take was very limited,

图像的质量非常局限

because the laws of physics dictate

因为物理定律决定了

that the best picture you can take through a telescope

通过望远镜拍的最好的照片

is a function of the diameter of that telescope,

是这个望远镜直径的函数

and these satellites had a very small,

而这些卫星只有很小的

very constrained volume.

非常有限的空间

And we found that the best picture we would

我们发现我们能够捕捉到的

have been able to get looked something like this.

最好的照片，看上去是这样的

Although this was the low-cost option,

虽然这是个低成本的选择

quite frankly it was just too blurry

但老实说，这实在太模糊了

to see the things that make satellite imagery valuable.

这样的卫星图像也没有价值

So about three or four weeks later,

大约三四个星期以后

we met a group of engineers randomly

我们巧遇到了一群工程师

who had worked on the first

他们曾经参与了第一架

private imaging satellite ever developed,

商业成像卫星的开发工作

and they told us that back in the 1970s,

他们告诉我们，过去在1970年代

the U.S. government had found a powerful

美国政府找到了一个强大的

optimal tradeoff --

最佳的权衡——

that in taking pictures at right about one meter resolution,

就是以刚好一米的分辨率进行拍照

being able to see objects one meter in size,

能够看到物体一米大小

they had found that they could not just get very high-quality images,

他们发现，他们不仅可以获得高质量的图像

but get a lot of them at an affordable price.

而且可以用低廉的价格获取很多

From our own computer simulations,

从我们自己的计算机模拟结果中

we quickly found that one meter really was

我们很快发现，一米确实是

the minimum viable product

最小的、可行的，能够看到

to be able to see the drivers of our global economy,

全球经济的驱动因素的分辨率

for the first time, being able to count

第一次，我们可以数

the ships and cars and shipping containers and trucks

那些在地球上每天在移动的

that move around our world on a daily basis,

船只、车辆、集装箱和卡车的数量

while conveniently still not being able to see individuals.

虽然还看不到清楚人物

We had found our compromise.

我们找到了我们的最佳权衡点

We would have to build something larger

我们需要造一个

than the original breadbox,

比原有的面包盒更大的东西

now more like a mini-fridge,

更像个迷你冰箱

but we still wouldn't have to build a pickup truck.

但是我们却不需要造一个货车

So now we had our constraint.

现在我们有了限制

The laws of physics dictated

物理定律决定了

the absolute minimum-sized telescope that we could build.

我们能够打造的最小的望远镜

What came next was making the rest of the satellite

接下来要做的，就是把这个卫星的其它部分

as small and as simple as possible,

造得尽可能地小而简单

basically a flying telescope with four walls

基本上就是一个会飞的望远镜，它有四面墙

and a set of electronics smaller than a phone book

和一些体积比通讯录还小的

that used less power than a 100 watt lightbulb.

用的电比100瓦灯泡还少的电子部件

The big challenge became actually taking

真正最大的挑战是

the pictures through that telescope.

通过那个望远镜拍照

Traditional imaging satellites use a line scanner,

传统的成像卫星采用行扫描仪

similar to a Xerox machine,

类似于复印机

and as they traverse the Earth, they take pictures,

随着它们穿越地球，它们进行拍照

scanning row by row by row

逐行地进行扫描

to build the complete image.

从而建立完整的图像

Now people use these because they get a lot of light,

如今人们用这个，因为它们可以获得很多光线

which means less of the noise you see

也就意味着少一些

in a low-cost cell phone image.

你在便宜的手机照片上会看到的噪点

The problem with them is they require

这个方法的问题是，它们需要

very sophisticated pointing.

非常复杂的定位

You have to stay focused on a 50-centimeter target

你需要从600英里以外

from over 600 miles away

聚焦一个50厘米的对象

while moving at more than seven kilometers a second,

同时以每秒7公里的速度移动

which requires an awesome degree of complexity.

这些都需要非常高的准确度

So instead, we turned to a new generation of video sensors,

因此，我们采用了一种新的视频传感器

originally created for use in night vision goggles.

它最早是用于夜视镜的

Instead of taking a single, high quality image,

我们不拍单一的高质量图像

we could take a videostream

而是拍一段

of individually noisier frames,

由很多帧噪点较高的图像组成的视频

but then we could recombine

然后我们在地面上

all of those frames together

运用复杂的像素处理技术

into very high-quality images

把这些图像重新组合在一起

using sophisticated pixel processing techniques

把这些图像重新组合在一起

here on the ground,

组成高质量的图像

at a cost of one one hundredth a traditional system.

而花费只是传统办法的百分之一

And we applied this technique

我们还把这个技术

to many of the other systems on the satellite as well,

运用到卫星的很多其它系统上

and day by day, our design evolved

一天天过去，我们的设计已经从

from CAD to prototypes

CAD（计算机辅助设计）变成了样机

to production units.

从样机变成了实际产品

A few short weeks ago,

短短几个星期前

we packed up SkySat 1,

我们装好了Skysat1

put our signatures on it,

我们在上面签了名

and waved goodbye for the last time on Earth.

最后一次在地球上跟它挥手告别

Today, it's sitting in its final launch configuration

今天，它正坐在它的最终发射装置里

ready to blast off in a few short weeks.

准备在几个星期后升空

And soon, we'll turn our attention to launching

很快，我们会转向发射

a constellation of 24 or more of these satellites

一系列的卫星，24个或者更多

and beginning to build the scalable analytics

并开始建立可扩展的分析

that will allow us to unearth the insights

使得我们能够从我们收集的

in the petabytes of data we will collect.

拍字节（10的15次方）的数据中发掘信息

So why do all of this? Why build these satellites?

那么为什么要做这些呢？为什么要造这些卫星？

Well, it turns out imaging satellites

因为成像卫星

have a unique ability to provide global transparency,

有提供全球透明度的独特能力

and providing that transparency on a timely basis

而即时提供

is simply an idea whose time has come.

全球透明度的时代已经到来

We see ourselves as pioneers of a new frontier,

我们把自己看作是新领域的先驱者

and beyond economic data,

在经济数据以外

unlocking the human story, moment by moment.

慢慢解开人类之谜

For a data scientist

对一个碰巧

that just happened to go to space camp as a kid,

孩提时代去了太空营的数据科学家而言

it just doesn't get much better than that.

没有什么比这更好的了

Thank you.

谢谢大家！

(Applause)

（掌声）