How algorithms shape our world | Kevin Slavin

翻译人员: Felix Chen 校对人员: Chunxiang Qian

This is a photograph

这是一张由艺术家

by the artist Michael Najjar,

迈克尔·纳贾尔拍摄的照片，

and it's real,

这是真实的，

in the sense that he went there to Argentina

他在阿根廷

to take the photo.

拍摄的这张照片。

But it's also a fiction. There's a lot of work that went into it after that.

但其中也有虚构的成分。在拍摄后还做了许多工作。

And what he's done

他所做的是，

is he's actually reshaped, digitally,

他实际上数字化地重塑了

all of the contours of the mountains

所有山峰的轮廓

to follow the vicissitudes of the Dow Jones index.

以遵循道琼斯指数的变化。

So what you see,

因此各位所看到的，

that precipice, that high precipice with the valley,

这悬崖，深沟险壑，

is the 2008 financial crisis.

是2008年的金融危机。

The photo was made

这张照片是在我们

when we were deep in the valley over there.

陷入深谷时制作的。

I don't know where we are now.

我不知道我们现在在哪儿。

This is the Hang Seng index

这是为香港恒生指数

for Hong Kong.

制作的。

And similar topography.

类似的形状。

I wonder why.

我不知道为什么。

And this is art. This is metaphor.

这是艺术。这是隐喻。

But I think the point is

但我认为这是个

that this is metaphor with teeth,

带着牙齿会咬人的隐喻。

and it's with those teeth that I want to propose today

带着这些齿状的线条，今天我建议

that we rethink a little bit

我们重新思考一下

about the role of contemporary math --

当代数学的角色 --

not just financial math, but math in general.

不仅是金融数学，还有一般数学。

That its transition

它由

from being something that we extract and derive from the world

我们从世界中提炼出的某种事物

to something that actually starts to shape it --

转变为事实上开始塑造世界的事物 --

the world around us and the world inside us.

我们周围的世界和我们内心的世界。

And it's specifically algorithms,

特别是算法，

which are basically the math

它基本上是

that computers use to decide stuff.

计算机用于决策的数学。

They acquire the sensibility of truth

它们具有了真理的敏感性，

because they repeat over and over again,

因为它们会不断地重复。

and they ossify and calcify,

它们固化下来，

and they become real.

变得真实。

And I was thinking about this, of all places,

我随时随地都在思考这些，

on a transatlantic flight a couple of years ago,

数年前在一次跨越大西洋的航班上，

because I happened to be seated

因为我恰好坐在一名

next to a Hungarian physicist about my age

与我年纪相仿的匈牙利物理学家旁边，

and we were talking

我们谈论

about what life was like during the Cold War

冷战期间在匈牙利的

for physicists in Hungary.

物理学家生活是什么样的。

And I said, "So what were you doing?"

我说道，“你们都做些什么？”

And he said, "Well we were mostly breaking stealth."

他回答道，“嗯，我们主要是在破解隐形飞机。”

And I said, "That's a good job. That's interesting.

我说，“不错。很有趣。

How does that work?"

怎么做呢？”

And to understand that,

要理解这个，

you have to understand a little bit about how stealth works.

要先对隐形飞机如何工作有点了解。

And so -- this is an over-simplification --

因此 -- 有点过于简化 --

but basically, it's not like

但基本上，不是

you can just pass a radar signal

仅仅让156吨的钢铁

right through 156 tons of steel in the sky.

在天空中穿过雷达信号就完事了。

It's not just going to disappear.

它不会就这么消失了。

But if you can take this big, massive thing,

但如果能把这巨大的东西

and you could turn it into

变成

a million little things --

上百万个小东西 --

something like a flock of birds --

有点像一群鸟 --

well then the radar that's looking for that

那么寻找目标的雷达

has to be able to see

能看到天空中的

every flock of birds in the sky.

每个鸟群。

And if you're a radar, that's a really bad job.

如果你是雷达，这真是个糟糕的工作。

And he said, "Yeah." He said, "But that's if you're a radar.

他说道，“是的。”他说，“如果你是雷达的话。

So we didn't use a radar;

所以我们不用雷达；

we built a black box that was looking for electrical signals,

我们造了个黑盒子来探测电子信号，

electronic communication.

电子通讯。

And whenever we saw a flock of birds that had electronic communication,

当我们看到有电子通讯的一群鸟时，

we thought, 'Probably has something to do with the Americans.'"

我们就认为这可能与美国有关。

And I said, "Yeah.

我说，”是的。

That's good.

这很不错。

So you've effectively negated

你们有效地让60年的

60 years of aeronautic research.

航空研究无效了。

What's your act two?

你的下一步是什么？

What do you do when you grow up?"

你长大后，你想要做什么？“

And he said,

他说，

"Well, financial services."

”嗯，金融服务。“

And I said, "Oh."

我说，”哦。“

Because those had been in the news lately.

因为这已经在最近的新闻里了。

And I said, "How does that work?"

我说，”这工作怎么样？“

And he said, "Well there's 2,000 physicists on Wall Street now,

他说，”嗯，现在有2000名物理学家在华尔街工作，

and I'm one of them."

我是其中之一。“

And I said, "What's the black box for Wall Street?"

我说，”华尔街的黑盒子是什么？“

And he said, "It's funny you ask that,

他说，”你问的这个很有趣，

because it's actually called black box trading.

因为这实际上被称为暗箱交易。

And it's also sometimes called algo trading,

也被称为算法交易，

algorithmic trading."

算法交易。“

And algorithmic trading evolved in part

算法交易的演化某种程度上

because institutional traders have the same problems

是因为机构交易员碰到了

that the United States Air Force had,

与美国空军一样的问题，

which is that they're moving these positions --

他们要移动这些点 --

whether it's Proctor & Gamble or Accenture, whatever --

不管是宝洁还是埃森哲，不管是什么 --

they're moving a million shares of something

他们在市场上交易上百万的

through the market.

某公司的股票。

And if they do that all at once,

如果他们一次移动全部，

it's like playing poker and going all in right away.

有点像象玩扑克室，所有筹码全部下注。

You just tip your hand.

你就露了底牌。

And so they have to find a way --

因此他们不得不找一个方法 --

and they use algorithms to do this --

他们用算法来完成这项工作 --

to break up that big thing

把巨大的交易

into a million little transactions.

转化为上百万次小的交易。

And the magic and the horror of that

其中的神奇和可怕之处是

is that the same math

你用于把庞然大物分解成

that you use to break up the big thing

上百万份的数学方法

into a million little things

也可以用于

can be used to find a million little things

找到上百万个小东西，

and sew them back together

重新拼接起来

and figure out what's actually happening in the market.

并算出市场上到底发生了什么。

So if you need to have some image

因此如果你需要一些

of what's happening in the stock market right now,

描绘了当前市场中的情景的图像，

what you can picture is a bunch of algorithms

你能呈现出的是一组

that are basically programmed to hide,

被设定为隐藏的算法，

and a bunch of algorithms that are programmed to go find them and act.

一组被设定为可找到并执行的算法。

And all of that's great, and it's fine.

这一切太伟大了，太棒了。

And that's 70 percent

美国股票市场

of the United States stock market,

中的百分之70，

70 percent of the operating system

操作系统的百分之70

formerly known as your pension,

前身为退休金，

your mortgage.

按揭。

And what could go wrong?

什么可能出问题？

What could go wrong

一年前

is that a year ago,

出的问题是

nine percent of the entire market just disappears in five minutes,

整个市场的百分之九消失了五分钟，

and they called it the Flash Crash of 2:45.

这被称为“2：45的瞬间崩溃”。

All of a sudden, nine percent just goes away,

突然之间，百分之九就消失了，

and nobody to this day

直到今天大家

can even agree on what happened

对发生了什么还不能达成一致，

because nobody ordered it, nobody asked for it.

因为没人下命令，没人要这么做。

Nobody had any control over what was actually happening.

对那天所发生的大家束手无策。

All they had

他们就是

was just a monitor in front of them

看着面前的屏幕

that had the numbers on it

上的数字

and just a red button

和一个红色按钮

that said, "Stop."

上面写着，“停。”

And that's the thing,

事情就是这样

is that we're writing things,

这就是我们正在编写的东西，

we're writing these things that we can no longer read.

我们在编写我们读不懂的东西。

And we've rendered something

我们把一些事情变得

illegible,

难以理解。

and we've lost the sense

我们已经对

of what's actually happening

这个我们创造的世界中

in this world that we've made.

正在发生的事情失去理解能力。

And we're starting to make our way.

我们开始前进。

There's a company in Boston called Nanex,

在波士顿有个名为Nanex的公司，

and they use math and magic

他们运用数学和魔法

and I don't know what,

和我不知道是什么的东西，

and they reach into all the market data

他们深入研究所有他们能找到的

and they find, actually sometimes, some of these algorithms.

市场数据，实际上有时候是一些算法。

And when they find them they pull them out

当他们找到这些数据时，就把数据抽取出来

and they pin them to the wall like butterflies.

像蝴蝶似的把它们钉在墙上。

And they do what we've always done

他们所做的也是我们在

when confronted with huge amounts of data that we don't understand --

面对大量我们无法理解的数据时所做的 --

which is that they give them a name

给它们一个名字

and a story.

和一个故事。

So this is one that they found,

这就是他们找的一个，

they called the Knife,

他们称之为‘小刀’，

the Carnival,

‘嘉年华’，

the Boston Shuffler,

‘波士顿洗牌者’，

Twilight.

暮光。

And the gag is

有意思的是

that, of course, these aren't just running through the market.

这不仅存在于股票市场上。

You can find these kinds of things wherever you look,

一旦你知道如何寻找它们，

once you learn how to look for them.

无论在哪儿你都能找到这类东西，

You can find it here: this book about flies

你能在这儿找到它：这本关于苍蝇的书

that you may have been looking at on Amazon.

你可能在亚马逊上看到过这本书。

You may have noticed it

你或许已经注意到

when its price started at 1.7 million dollars.

它的价格是一百七十万美元。

It's out of print -- still ...

绝版 -- 仍然是绝版...

(Laughter)

（笑声）

If you had bought it at 1.7, it would have been a bargain.

如果你在一百七十万美元是购买了它，那还算便宜的。

A few hours later, it had gone up

数小时后，它涨到了

to 23.6 million dollars,

两千三百六十万美元，

plus shipping and handling.

含运费和手续费。

And the question is:

问题是：

Nobody was buying or selling anything; what was happening?

没有人购买或销售任何东西；发生了什么？

And you see this behavior on Amazon

你在亚马逊看到的这一行为

as surely as you see it on Wall Street.

毫无疑问与在华尔街看到的一样。

And when you see this kind of behavior,

当你看到这类行为时，

what you see is the evidence

你所看到的就是

of algorithms in conflict,

算法冲突的证据，

algorithms locked in loops with each other,

算法相互锁定，

without any human oversight,

没有人类的监管，

without any adult supervision

没有任何成熟的监督

to say, "Actually, 1.7 million is plenty."

说，“实际上，一百七十万美元是很大一笔了。”

(Laughter)

（笑声）

And as with Amazon, so it is with Netflix.

与亚马逊一样，Netflix也有这样的问题。

And so Netflix has gone through

因此Netflix多年来已经

several different algorithms over the years.

经历了若干不同算法。

They started with Cinematch, and they've tried a bunch of others --

他们开始用的是Cinematch，后来又尝试了一些其他的。

there's Dinosaur Planet; there's Gravity.

有Dinosaur Planet，Gravity。

They're using Pragmatic Chaos now.

现在他们在使用Pragmatic Chaos。

Pragmatic Chaos is, like all of Netflix algorithms,

Pragmatic Chaos，与所有Netflix算法相同，

trying to do the same thing.

试着做同样的事情。

It's trying to get a grasp on you,

它试图把握住你，

on the firmware inside the human skull,

掌控人类头骨内的固件，

so that it can recommend what movie

这样它就能向你推荐

you might want to watch next --

你可能想看的电影 --

which is a very, very difficult problem.

这是个非常非常困难的事情。

But the difficulty of the problem

但问题和事实的难点

and the fact that we don't really quite have it down,

在于我们没有真的掌握它，

it doesn't take away

它没有消除

from the effects Pragmatic Chaos has.

Pragmatic Chaos的影响。

Pragmatic Chaos, like all Netflix algorithms,

Pragmatic Chaos，如同Netflix的所有算法，

determines, in the end,

最后决定了

60 percent

百分之60

of what movies end up being rented.

最终被租用的电影。

So one piece of code

因此一段带有

with one idea about you

你的看法的代码

is responsible for 60 percent of those movies.

对百分之60的电影负责。

But what if you could rate those movies

但如果你在这些电影制作之前

before they get made?

对它们进行评价会怎样？

Wouldn't that be handy?

这样岂不是很方便？

Well, a few data scientists from the U.K. are in Hollywood,

嗯，一些来自英国的数据科学家在好莱坞，

and they have "story algorithms" --

他们有故事算法 --

a company called Epagogix.

一家名为Epagogix的公司。

And you can run your script through there,

你可以向他们提供你的剧本，

and they can tell you, quantifiably,

他们能量化地告诉你

that that's a 30 million dollar movie

这是个三千万美元票房的电影

or a 200 million dollar movie.

或是个两亿美元票房的电影。

And the thing is, is that this isn't Google.

这不是Google。

This isn't information.

不是信息。

These aren't financial stats; this is culture.

不是金融统计；这是文化。

And what you see here,

你在这儿看到的

or what you don't really see normally,

或你没有真正察觉的，

is that these are the physics of culture.

是文化的物理学。

And if these algorithms,

如果这些算法，

like the algorithms on Wall Street,

象华尔街中的算法，

just crashed one day and went awry,

某天崩溃了出错了，

how would we know?

我们怎么知道，

What would it look like?

那会是什么样子？

And they're in your house. They're in your house.

它们在你的屋子里，它们在你的屋子里。

These are two algorithms competing for your living room.

有两种算法在争夺你的客厅。

These are two different cleaning robots

有两种不同的清洁机器人

that have very different ideas about what clean means.

它们对清洁的含义有着非常不同的理解。

And you can see it

如果你让它慢下来，在它上面放上灯光

if you slow it down and attach lights to them,

你就能够看到。

and they're sort of like secret architects in your bedroom.

有点像你卧室里的秘密建筑师。

And the idea that architecture itself

建筑本身

is somehow subject to algorithmic optimization

某种程度上服从算法优化的想法

is not far-fetched.

并非牵强。

It's super-real and it's happening around you.

这是超现实，它就发生在你周围。

You feel it most

当你在一个密封的金属盒子里时，

when you're in a sealed metal box,

一种被称为目标控制电梯的

a new-style elevator;

新式电梯，

they're called destination-control elevators.

最能感受到它。

These are the ones where you have to press what floor you're going to go to

在你进入电梯之前你要按下

before you get in the elevator.

你所要去的楼层的按钮。

And it uses what's called a bin-packing algorithm.

它使用装箱算法。

So none of this mishegas

因此让每个人进入

of letting everybody go into whatever car they want.

他们想进的电梯一点也不混乱。

Everybody who wants to go to the 10th floor goes into car two,

想去10楼的人进入二号电梯，

and everybody who wants to go to the third floor goes into car five.

想去三层的人进入五号电梯。

And the problem with that

问题是

is that people freak out.

人们吓坏了。

People panic.

人们抓狂了。

And you see why. You see why.

你知道为什么。你知道为什么。

It's because the elevator

因为电梯

is missing some important instrumentation, like the buttons.

缺少了些重要的东西，比如按钮。

(Laughter)

（笑声）

Like the things that people use.

正如人们使用的电梯。

All it has

都有

is just the number that moves up or down

标明向上或向下的数字

and that red button that says, "Stop."

还有一个红色按钮，上写着，“停。”

And this is what we're designing for.

这就是我们正在设计的。

We're designing

我们正在设计

for this machine dialect.

这种机器方言。

And how far can you take that? How far can you take it?

能做到什么程度？能用它做到何种境界？

You can take it really, really far.

用它可以走得很远很远。

So let me take it back to Wall Street.

让我们回到华尔街。

Because the algorithms of Wall Street

因为华尔街的算法

are dependent on one quality above all else,

依赖于一个高于一切的特质，

which is speed.

速度。

And they operate on milliseconds and microseconds.

它们的运行时间以毫秒和微妙计算。

And just to give you a sense of what microseconds are,

让你们对微秒有点感觉，

it takes you 500,000 microseconds

点击一下鼠标

just to click a mouse.

要花50万微秒的时间。

But if you're a Wall Street algorithm

但如果你是一个华尔街的算法

and you're five microseconds behind,

落后5微秒，

you're a loser.

你就是失败者。

So if you were an algorithm,

因此，如果你是一个算法，

you'd look for an architect like the one that I met in Frankfurt

你得寻找一个像我在法兰克福所遇的那样的建筑师

who was hollowing out a skyscraper --

把整个摩天大楼掏空 --

throwing out all the furniture, all the infrastructure for human use,

扔掉所有的家具和人类使用的基础设施，

and just running steel on the floors

仅用刚才铺至地面，

to get ready for the stacks of servers to go in --

准备好大批的服务器入驻 --

all so an algorithm

整个算法

could get close to the Internet.

都能快速连入互联网。

And you think of the Internet as this kind of distributed system.

把互联网看成一种分布式系统。

And of course, it is, but it's distributed from places.

当然，它就是，但分布于不同地点。

In New York, this is where it's distributed from:

在纽约，它分布在：

the Carrier Hotel

位于哈德逊大街的

located on Hudson Street.

电信酒店。

And this is really where the wires come right up into the city.

这是线缆真正进入这座城市的地方。

And the reality is that the further away you are from that,

事实上你距离这地方越远，

you're a few microseconds behind every time.

每次都会落后几微秒。

These guys down on Wall Street,

在华尔街上的这些家伙，

Marco Polo and Cherokee Nation,

Marco Polo和Cherokee Nation，

they're eight microseconds

他们比这些

behind all these guys

在电信酒店周围的

going into the empty buildings being hollowed out

被掏空了的大厦里的家伙

up around the Carrier Hotel.

要落后八微秒。

And that's going to keep happening.

这在不断发生。

We're going to keep hollowing them out,

我们要把它们不断掏空，

because you, inch for inch

因为你，每一英寸

and pound for pound and dollar for dollar,

每一磅，每一美元，

none of you could squeeze revenue out of that space

没人能像‘波士顿洗牌者’那样

like the Boston Shuffler could.

从中榨取收益。

But if you zoom out,

但如果你缩小地图，

if you zoom out,

如果你缩小地图，

you would see an 825-mile trench

你会看到一条长达825英里的

between New York City and Chicago

位于纽约城和芝加哥之间的沟渠，

that's been built over the last few years

它在过去几年中

by a company called Spread Networks.

由一家名为Spread Networks的公司建造。

This is a fiber optic cable

这条

that was laid between those two cities

两座城市间的光缆

to just be able to traffic one signal

就是为了以比你点击鼠标

37 times faster than you can click a mouse --

快37倍的速度传输信号 --

just for these algorithms,

就是为了这些算法，

just for the Carnival and the Knife.

就是为了‘嘉年华’和‘小刀’。

And when you think about this,

你想一想，

that we're running through the United States

我们正在用炸药和岩石锯

with dynamite and rock saws

穿过美国，

so that an algorithm can close the deal

只是为了一个算法

three microseconds faster,

能快三微秒完成交易，

all for a communications framework

都是为了一个没人会知道的

that no human will ever know,

通信框架，

that's a kind of manifest destiny;

这有点命运天定论

and we'll always look for a new frontier.

并总是在寻找新的领域。

Unfortunately, we have our work cut out for us.

不幸地是，我们面前困难重重。

This is just theoretical.

这仅仅是理论上的。

This is some mathematicians at MIT.

这是MIT的一些数学家制作的。

And the truth is I don't really understand

我并不太明白

a lot of what they're talking about.

他们所谈论的。

It involves light cones and quantum entanglement,

它涉及光锥体和量子纠缠，

and I don't really understand any of that.

这些我真的都不太明白。

But I can read this map,

但我能看明白这张地图。

and what this map says

这张地图表明

is that, if you're trying to make money on the markets where the red dots are,

如果你要在市场上赚钱，那些红点所在位置，

that's where people are, where the cities are,

也是人所在位置，也是城市所在位置，

you're going to have to put the servers where the blue dots are

就要把服务器放到蓝点所在位置

to do that most effectively.

这样最有效率。

And the thing that you might have noticed about those blue dots

各位也许已经注意到这些蓝点

is that a lot of them are in the middle of the ocean.

许多都在大洋中。

So that's what we'll do: we'll build bubbles or something,

那么我们要做的是，建造一些气泡之类的东西，

or platforms.

或者是平台。

We'll actually part the water

我们们确实能分离水，

to pull money out of the air,

从空气中挖掘财富，

because it's a bright future

因为这很有前途，

if you're an algorithm.

如果你是一个算法的话。

(Laughter)

（笑声）

And it's not the money that's so interesting actually.

实际上有意思的不是钱。

It's what the money motivates,

而是钱所激发的东西。

that we're actually terraforming

我们实际上在用

the Earth itself

这种算法的效率

with this kind of algorithmic efficiency.

在改造地球本身。

And in that light,

根据这点，

you go back

各位回去看看

and you look at Michael Najjar's photographs,

迈克尔·纳贾尔的照片，

and you realize that they're not metaphor, they're prophecy.

会领悟到它们不是隐喻，而是预言。

They're prophecy

它们是

for the kind of seismic, terrestrial effects

我们正在数学上掀起的

of the math that we're making.

那种地震效应的预言。

And the landscape was always made

风景总是由

by this sort of weird, uneasy collaboration

自然和人类之间的这种

between nature and man.

怪异不安的协作产生的。

But now there's this third co-evolutionary force: algorithms --

但现在有这些第三方协同进化力量：算法 --

the Boston Shuffler, the Carnival.

‘波士顿洗牌者‘，’嘉年华’。

And we will have to understand those as nature,

我们将不得不将这些视为自然。

and in a way, they are.

某种程度上，它们是的。

Thank you.

谢谢。

(Applause)

（掌声）