ABOUT THE SPEAKER
Danny Hillis - Computer theorist
Inventor, scientist, author, engineer -- over his broad career, Danny Hillis has turned his ever-searching brain on an array of subjects, with surprising results.

Why you should listen

Danny Hillis is an inventor, scientist, author and engineer. While completing his doctorate at MIT, he pioneered the concept of parallel computers that is now the basis for graphics processors and cloud computing. He holds more than 300 US patents, covering parallel computers, disk arrays, forgery prevention methods, various electronic and mechanical devices, and the pinch-to-zoom display interface. He has recently been working on problems in medicine as well. He is also the designer of a 10,000-year mechanical clock, and he gave a TED Talk in 1994 that is practically prophetic. Throughout his career, Hillis has worked at places like Disney, and now MIT and Applied Invention, always looking for the next fascinating problem.

More profile about the speaker
Danny Hillis | Speaker | TED.com
TED2013

Danny Hillis: The Internet could crash. We need a Plan B

Danny Hillis:網路也有可能癱瘓。我們須要 B 計劃

Filmed:
1,375,608 views

在 1970 以及 80 年代,一種寬厚的精神充滿整個網際網路,它的使用者遠比現在的少得多。但在今日,網路非常普級,串聯起數十億使用者、機器、以及各設施的重要零件──這讓我們在面對網路攻擊或是核心錯誤時,變得十分脆弱。網際網路先驅 Danny Hillis 主張網路最初並不是為這樣的使用尺度而設計的,並且為我們吹響改革的號角,來發展 B 計劃:萬一,或是當網路癱瘓時,一個可以依賴的平行系統。
- Computer theorist
Inventor, scientist, author, engineer -- over his broad career, Danny Hillis has turned his ever-searching brain on an array of subjects, with surprising results. Full bio

Double-click the English transcript below to play the video.

00:16
So, this book that I have in my hand
0
603
3066
嗯,我手上的這本書
00:19
is a directory目錄 of everybody每個人 who had an email電子郵件 address地址
1
3669
3934
是 1982 年所有電子信箱的
00:23
in 1982. (Laughter笑聲)
2
7603
3519
地址通訊錄。(笑聲)
00:27
Actually其實, it's deceptively迷惑 large.
3
11122
3685
實際上,它的大小是騙人的。
00:30
There's actually其實 only about 20 people on each page,
4
14807
3437
其中每頁根本就只有將近 20 人的資料,
00:34
because we have the name名稱, address地址
5
18244
1890
因為那包含每個人的
人名、地址、
00:36
and telephone電話 number of every一切 single person.
6
20134
3093
及電話號碼。
00:39
And, in fact事實, everybody's每個人的 listed上市 twice兩次,
7
23227
2183
而實際上,每個人都出現了兩次,
00:41
because it's sorted分類 once一旦 by name名稱 and once一旦 by email電子郵件 address地址.
8
25410
4558
因為一次是用人名排列,
而一次是用信箱位址排列。
00:45
Obviously明顯 a very small community社區.
9
29968
2901
很明顯是一個小族群。
00:48
There were only two other DannysDannys on the Internet互聯網 then.
10
32869
3431
當時的網路中只有另外兩個人也叫 Danny。
(和講者同名)
00:52
I knew知道 them both.
11
36300
1600
兩個人我都認識。
00:53
We didn't all know each other,
12
37900
1756
我們並不全然知道對方,
00:55
but we all kind of trusted信任 each other,
13
39656
3113
但是我們都一定程度地信任對方,
00:58
and that basic基本 feeling感覺 of trust相信
14
42769
3798
而這種基本的信任感,
01:02
permeated滲透 the whole整個 network網絡,
15
46567
2587
瀰漫了整個網路世界,
01:05
and there was a real真實 sense that
16
49154
2357
進而存在一種真實的感覺,
01:07
we could depend依靠 on each other to do things.
17
51511
3006
彷彿我們可以仰賴彼此來做些事情。
01:10
So just to give you an idea理念 of the level水平 of trust相信 in this community社區,
18
54517
3575
為了讓各位對這種信任的程度有個概念,
01:13
let me tell you what it was like
19
58092
1625
讓我來告訴你們
01:15
to register寄存器 a domain name名稱 in the early days.
20
59717
4105
在早些年代註冊一個網域名稱
是什麼樣子。
01:19
Now, it just so happened發生 that I got to register寄存器
21
63822
2630
嗯,意外地我那時註冊的
01:22
the third第三 domain name名稱 on the Internet互聯網.
22
66452
2460
是整個網際網路中
第三個網域名稱。
01:24
So I could have anything I wanted
23
68912
1648
所以我可以用任何我喜歡的名字,
01:26
other than bbnBBN.comCOM and symbolics符號處理.comCOM.
24
70560
4445
只要不是 bbn.com 或是 symbolics.com。
01:30
So I picked採摘的 think.comCOM, but then I thought,
25
75005
3414
所以我選了 think.com,但是我又想,
01:34
you know, there's a lot of really interesting有趣 names out there.
26
78419
2986
你知道的,還有一堆有趣的名字可以用。
01:37
Maybe I should register寄存器 a few少數 extras演員 just in case案件.
27
81405
4209
也許我可以多註冊幾個
以備哪天不時之須。
01:41
And then I thought, "Nah, that wouldn't不會 be very nice不錯."
28
85614
3284
但最後我想:「算了,這樣不太好。」
01:44
(Laughter笑聲)
29
88898
5695
(笑聲)
01:50
That attitude態度 of only taking服用 what you need
30
94593
3629
這種須要多少拿多少
的態度
01:54
was really what everybody每個人 had on the network網絡 in those days,
31
98222
4044
真的就是那年代的網路中
每個人心中所想的,
01:58
and in fact事實, it wasn't just the people on the network網絡,
32
102266
3297
而事實上,並不只有人這麼想,
02:01
but it was actually其實 kind of built內置 into the protocols協議
33
105563
2720
實際上這想法也建立到網路本身
02:04
of the Internet互聯網 itself本身.
34
108283
1944
的通訊協定中。
02:06
So the basic基本 idea理念 of I.P., or Internet互聯網 protocol協議,
35
110227
3859
所以 I.P.,或是所謂的網路協定,
的基本概念,
02:09
and the way that the -- the routing路由 algorithm算法 that used it,
36
114086
3729
還有它基於這協定的路由演算法,
(譯註:即決定由哪臺電腦接手傳輸的演算法。)
02:13
were fundamentally從根本上 "from each according根據 to their ability能力,
37
117815
4129
基本上就是「盡其所能,
02:17
to each according根據 to their need."
38
121944
2118
各取所需。」
02:19
And so, if you had some extra額外 bandwidth帶寬,
39
124062
3013
所以,如果你有多的頻寬,
02:22
you'd deliver交付 a message信息 for someone有人.
40
127075
1443
你會幫別人傳些訊息。
02:24
If they had some extra額外 bandwidth帶寬, they would deliver交付 a message信息 for you.
41
128518
2993
而如果他們有多的頻寬,
他們也會幫你傳遞訊息。
02:27
You'd kind of depend依靠 on people to do that,
42
131511
2014
一定程度上你必須依賴人們這麼做,
02:29
and that was the building建造 block.
43
133525
2664
而這是網際網路的構成要素。
02:32
It was actually其實 interesting有趣 that such這樣 a communist共產 principle原理
44
136189
2627
很有趣的是,這樣一種
共產主義的原則
02:34
was the basis基礎 of a system系統 developed發達 during the Cold War戰爭
45
138816
2436
是冷戰時期國防部所開發的一個系統,
它的基礎,
02:37
by the Defense防禦 Department,
46
141252
2431
是冷戰時期國防部所開發的一個系統,
它的基礎,
02:39
but it obviously明顯 worked工作 really well,
47
143683
3070
但很明顯地它運作得非常好,
02:42
and we all saw what happened發生 with the Internet互聯網.
48
146753
3443
而我們都見證了網際網路的發展。
02:46
It was incredibly令人難以置信 successful成功.
49
150196
2272
簡直是不可思議地成功。
02:48
In fact事實, it was so successful成功 that there's no way
50
152468
3297
事實上,網際網路實在太成功,
以致於你不可能
02:51
that these days you could make a book like this.
51
155765
3588
在現代做出這麼薄的通訊錄。
02:55
My rough calculation計算 is it would be about 25 miles英里 thick.
52
159353
6166
我粗略估計那會有 25 哩長。
03:01
But, of course課程, you couldn't不能 do it,
53
165519
1174
但當然地,你不可能完成它,
03:02
because we don't know the names of all the people
54
166693
1790
因為我們不知道所有
網路或電子信箱擁有者的名字,
03:04
with Internet互聯網 or email電子郵件 addresses地址,
55
168483
3074
因為我們不知道所有
網路或電子信箱擁有者的名字,
03:07
and even if we did know their names,
56
171557
1532
而即使我們知道了名字,
03:08
I'm pretty漂亮 sure that they would not want their name名稱,
57
173089
2433
我很確定大家並不希望他們的名字、
03:11
address地址 and telephone電話 number published發表 to everyone大家.
58
175522
4296
信箱、以及電話號碼
被公開。
03:15
So the fact事實 is that there's a lot of bad guys on the Internet互聯網 these days,
59
179818
3702
所以結果就是現今的網路中
有許多壞人,
03:19
and so we dealt處理 with that by making製造
60
183520
4171
因此我們藉由許多機制來面對,
這包含:
03:23
walled薄壁 communities社區,
61
187691
1899
有防火牆的社群、
03:25
secure安全 subnetworks子網, VPNsVPN的,
62
189590
4737
安全子網路、虛擬私人網路,
03:30
little things that aren't really the Internet互聯網
63
194327
1596
一些小方法,
它們並不真的是網際網路
03:31
but are made製作 out of the same相同 building建造 blocks,
64
195923
1971
卻都建構於
先前所提的構成要素,
03:33
but we're still basically基本上 building建造 it out of those
65
197894
2240
但我們基本上還是用那些構成要素,
03:36
same相同 building建造 blocks with those same相同 assumptions假設 of trust相信.
66
200134
4267
以及那些信任感的前提,
來建構網路。
03:40
And that means手段 that it's vulnerable弱勢
67
204401
2722
而這表示當面對
03:43
to certain某些 kinds of mistakes錯誤 that can happen發生,
68
207123
2396
某些可能發生的錯誤,
03:45
or certain某些 kinds of deliberate商榷 attacks攻擊,
69
209519
1861
或是某些蓄意的攻擊時,
它是脆弱的,
03:47
but even the mistakes錯誤 can be bad.
70
211380
3105
但只是錯誤也可以變得很糟糕。
03:50
So, for instance,
71
214485
2949
所以,舉例來說,
03:53
in all of Asia亞洲 recently最近,
72
217434
1953
最近在整個亞洲,
03:55
it was impossible不可能 to get YouTubeYouTube的 for a little while
73
219387
3433
使用 YouTube 是幾乎不可能的,
03:58
because Pakistan巴基斯坦 made製作 some mistakes錯誤
74
222820
2280
因為巴基斯坦在為內部網路
審查 YouTube 時
04:01
in how it was censoring終檢 YouTubeYouTube的 in its internal內部 network網絡.
75
225100
3735
發生了些錯誤。
04:04
They didn't intend打算 to screw up Asia亞洲, but they did
76
228835
3094
他們並不希望把亞洲網路搞亂,
但是他們還做了,
04:07
because of the way that the protocols協議 work.
77
231929
2904
這歸咎於網路協定的運作方式。
04:10
Another另一個 example that may可能 have affected受影響 many許多 of you in this audience聽眾 is,
78
234833
3224
另一個可能影響在座於多人的是,
04:13
you may可能 remember記得 a couple一對 of years年份 ago,
79
238057
2591
你可能記得幾年之前,
04:16
all the planes飛機 west西 of the Mississippi密西西比州 were grounded接地
80
240648
2813
所有密西西比河以西的飛機
都停飛了,
04:19
because a single routing路由 card in Salt Lake City
81
243461
2892
因為一塊鹽湖城的路由卡中
04:22
had a bug竊聽器 in it.
82
246353
2362
有一個小錯誤。
04:24
Now, you don't really think
83
248715
2273
現在,你並不真的認為
04:26
that our airplane飛機 system系統 depends依靠 on the Internet互聯網,
84
250988
3026
我們的飛航系統必須仰賴網路,
04:29
and in some sense it doesn't.
85
254014
1006
而某些角度來說並不是的。
04:30
I'll come back to that later後來.
86
255020
1616
我等等再回來談這個。
04:32
But the fact事實 is that people couldn't不能 take off
87
256636
2250
但事實就是人們沒辦法讓飛機起飛,
04:34
because something was going wrong錯誤 on the Internet互聯網,
88
258886
2275
因為網路出了錯,
04:37
and the router路由器 card was down.
89
261161
2575
而那張路由卡停止運作。
04:39
And so, there are many許多 of those things that start開始 to happen發生.
90
263736
3779
因此,許多這類的事情開始發生。
04:43
Now, there was an interesting有趣 thing that happened發生 last April四月.
91
267515
3139
嗯,有件有趣的事發生在去年四月。
04:46
All of a sudden突然,
92
270654
1526
剎那之間,
04:48
a very large percentage百分比 of the traffic交通 on the whole整個 Internet互聯網,
93
272180
3332
非常大量的流量,佔滿整個網際網路,
04:51
including包含 a lot of the traffic交通 between之間 U.S. military軍事 installations安裝,
94
275512
4131
還包含了美國軍方各裝置間的網路,
04:55
started開始 getting得到 re-routed改道 through通過 China中國.
95
279643
2768
並開始重新導向到中國。
04:58
So for a few少數 hours小時, it all passed通過 through通過 China中國.
96
282411
2722
所以有幾個小時,這流量在中國流通。
05:01
Now, China中國 Telecom電信 says it was just an honest誠實 mistake錯誤,
97
285133
4574
而現在,中國電信說那只是一個純然的錯誤,
05:05
and it is actually其實 possible可能 that it was, the way things work,
98
289707
3857
而那的確也有可能,由於網路的運作方式,
05:09
but certainly當然 somebody could make
99
293564
2035
但是無礙地,有人可以做出
05:11
a dishonest不誠實 mistake錯誤 of that sort分類 if they wanted to,
100
295599
3273
蓄意的那類錯誤,只要他們願意,
05:14
and it shows節目 you how vulnerable弱勢 the system系統 is even to mistakes錯誤.
101
298872
3358
而這告訴你整個系統即使對於錯誤,
是多麼脆弱。
05:18
Imagine想像 how vulnerable弱勢 the system系統 is to deliberate商榷 attacks攻擊.
102
302230
4921
再想想面對蓄意攻擊時,
會有多脆弱。
05:23
So if somebody really wanted to attack攻擊 the United聯合的 States狀態
103
307151
3398
如果近期真的有人想要攻擊美國、
05:26
or Western西 civilization文明 these days,
104
310549
2092
或是西方文明,
05:28
they're not going to do it with tanks坦克.
105
312641
2189
他們並不會用訴諸坦克。
05:30
That will not succeed成功.
106
314830
2413
那將不會成功。
05:33
What they'll他們會 probably大概 do is something
107
317243
2003
他們可能做的是某些
05:35
very much like the attack攻擊 that happened發生
108
319246
3291
和伊朗核能裝置攻擊事件
非常類似的攻擊。
05:38
on the Iranian伊朗的 nuclear facility設施.
109
322537
2707
和依朗核能裝置攻擊事件
非常類似的攻擊。
05:41
Nobody沒有人 has claimed聲稱 credit信用 for that.
110
325244
2274
沒人承認做了這件事。
05:43
There was basically基本上 a factory of industrial產業 machines.
111
327518
3507
基本上就是有一個工廠的工業機械。
05:46
It didn't think of itself本身 as being存在 on the Internet互聯網.
112
331025
2577
沒人覺得它們會連在網路上。
05:49
It thought of itself本身 as being存在 disconnected斷開的 from the Internet互聯網,
113
333602
2587
大家認為機器是和網際網路沒有通連的,
05:52
but it was possible可能 for somebody to smuggle走私
114
336189
1947
但很有可能某人偷偷帶了
05:54
a USBUSB drive駕駛 in there, or something like that,
115
338136
2532
一個 USB 到那裡,或是類似的東西,
05:56
and software軟件 got in there that causes原因 the centrifuges離心機,
116
340668
3396
而軟體得以侵入並造成廠房離心機
05:59
in that case案件, to actually其實 destroy破壞 themselves他們自己.
117
344064
2993
,在這個例子裡,自我破壞。
06:02
Now that same相同 kind of software軟件 could destroy破壞 an oil refinery煉油廠
118
347057
2666
而同一類的軟體也可以破壞石油精煉場、
06:05
or a pharmaceutical製藥 factory or a semiconductor半導體 plant.
119
349723
4422
或是製藥場、或是半導體工廠。
06:10
And so there's a lot of -- I'm sure you've read a lot in papers文件,
120
354145
3266
因此有很多,我相信你們在文獻中看到很多,
06:13
about worries about cyberattacks網絡攻擊
121
357411
2191
關於網路攻擊的擔心、
06:15
and defenses防禦 against反對 those.
122
359602
2610
或是如何去防衛。
06:18
But the fact事實 is, people are mostly大多 focused重點 on
123
362212
1980
但事實上,人們主要關注於
06:20
defending衛冕 the computers電腦 on the Internet互聯網,
124
364192
2283
防衛有連上網的電腦,
06:22
and there's been surprisingly出奇 little attention注意
125
366475
2386
而只有少到令人訝異的心力
06:24
to defending衛冕 the Internet互聯網 itself本身 as a communications通訊 medium.
126
368861
4365
是用於防衛
當作通訊媒介的網路本身。
06:29
And I think we probably大概 do need to pay工資
127
373226
1863
而我認為我們可能真的須要
06:30
some more attention注意 to that, because it's actually其實 kind of fragile脆弱.
128
375089
3088
更加關注這件事,因為
網路確實有些脆弱。
06:34
So actually其實, in the early days,
129
378177
2927
所以,事實上,前些日子,
06:37
back when it was the ARPANETARPANET,
130
381104
1724
回到 ARPANET 的時代,
(譯註:ARPANET 為美國網際網路的前身。)
06:38
there were actually其實 times -- there was a particular特定 time it failed失敗 completely全然
131
382828
3589
實際上有好幾次,其中還有一次網路完全停擺,
06:42
because one single message信息 processor處理器
132
386417
3285
就因為一個訊息處理器
06:45
actually其實 got a bug竊聽器 in it.
133
389702
2476
有些內部錯誤。
06:48
And the way the Internet互聯網 works作品 is
134
392178
2272
而網路運作的方式是
06:50
the routers路由器 are basically基本上 exchanging交換 information信息
135
394450
3567
不同路由器基本上會交換資訊,
06:53
about how they can get messages消息 to places地方,
136
398017
2590
這資訊是關於
如何將訊息傳到目的地,
06:56
and this one processor處理器, because of a broken破碎 card,
137
400607
3898
而這一個出錯的處理器,
因為一張損壞的電路板
07:00
decided決定 it could actually其實 get a message信息
138
404505
2009
而決定它會在一個時間為負數的時刻
07:02
to some place地點 in negative time.
139
406514
2688
來傳遞訊息。
07:05
So, in other words, it claimed聲稱 it could deliver交付 a message信息 before you sent發送 it.
140
409202
3983
所以,也就是說,它宣聲它可以
在你傳出訊息前就把訊息傳出去。
07:09
So of course課程, the fastest最快的 way to get a message信息 anywhere隨地
141
413185
3150
當然地,最快傳播訊息的方法
07:12
was to send發送 it to this guy,
142
416335
1876
就是把訊息傳給這個人,
07:14
who would send發送 it back in time and get it there super early,
143
418211
3547
他可以準時回傳訊息並在
「超早」的時間點送達訊息,
07:17
so every一切 message信息 in the Internet互聯網
144
421758
2856
所以網路上的每個訊息
07:20
started開始 getting得到 switched交換的 through通過 this one node節點,
145
424614
3242
漸漸開始想藉由這個節點
來交換訊息,
07:23
and of course課程 that clogged堵塞 everything up.
146
427856
1516
而當然網路就塞爆了。
07:25
Everything started開始 breaking破壞.
147
429372
2155
所有通路開始攤瘓。
07:27
The interesting有趣 thing was, though雖然,
148
431527
2040
而儘管如此,有趣的是
07:29
that the sysadmins系統管理員 were able能夠 to fix固定 it,
149
433567
1840
系統管理員是有辦法修正這個錯誤的,
07:31
but they had to basically基本上 turn every一切 single thing on the Internet互聯網 off.
150
435407
4782
但他們基本上必須暫停
網路上每一個傳輸任務。
07:36
Now, of course課程 you couldn't不能 do that today今天.
151
440189
1398
嗯,當然,在今日你不可能這麼做。
07:37
I mean, everything off, it's like
152
441587
2394
我的意思是,暫停每個傳輸任務
就像是
07:39
the service服務 call you get from the cable電纜 company公司,
153
443981
2232
整個世界裡你只接得到
07:42
except for the whole整個 world世界.
154
446213
3761
電纜公司的服務電話一般。
07:45
Now, in fact事實, they couldn't不能 do it for a lot of reasons原因 today今天.
155
449974
2059
實際上,今日的社會裡,
他們還有很多原因不能這麼做。
07:47
One of the reasons原因 is a lot of their telephones電話
156
452033
2622
其中一個原因是許多人的電話
07:50
use IPIP protocol協議 and use things like SkypeSkype的 and so on
157
454655
2969
是使用 IP 協定並使用 Skype 之類的軟體,
07:53
that go through通過 the Internet互聯網 right now,
158
457624
2069
而這些軟體是藉由網路傳輸訊息的,
07:55
and so in fact事實 we're becoming變得 dependent依賴的 on it
159
459693
3282
所以我們開始在愈來愈多的事務上,
07:58
for more and more different不同 things,
160
462975
1792
漸漸和網路分不開,
08:00
like when you take off from LAXLAX,
161
464767
2938
就像是當你從洛山機國際機場起飛,
08:03
you're really not thinking思維 you're using運用 the Internet互聯網.
162
467705
1881
你不會認為你正在使用網路。
08:05
When you pump gas加油站, you really don't think you're using運用 the Internet互聯網.
163
469586
4116
當你在加油時,你更不認為你用有到網路。
08:09
What's happening事件 increasingly日益, though雖然, is these systems系統
164
473702
2125
然而這些系統中,正加速發生的
08:11
are beginning開始 to use the Internet互聯網.
165
475827
1867
就是開始使用網路的這個事實。
08:13
Most of them aren't based基於 on the Internet互聯網 yet然而,
166
477694
3050
它們大部份還沒架構於網路之上,
08:16
but they're starting開始 to use the Internet互聯網 for service服務 functions功能,
167
480744
2536
但他們正開始將網路用於
各種服務、
08:19
for administrative行政的 functions功能,
168
483280
1905
以及管理功能,
08:21
and so if you take something like the cell細胞 phone電話 system系統,
169
485185
3063
所以如果你以手機系統,
這種多數時候還與網路較無關聯的事物為例,
08:24
which哪一個 is still relatively相對 independent獨立 of the Internet互聯網 for the most part部分,
170
488248
4320
所以如果你以手機系統,
這種多數時候還與網路較無關聯的事物為例,
08:28
Internet互聯網 pieces are beginning開始 to sneak潛行 into it
171
492568
2959
網路已經開始
由某些控制及管理等功能
08:31
in terms條款 of some of the control控制 and administrative行政的 functions功能,
172
495527
3804
潛入其中,
08:35
and it's so tempting誘人的 to use these same相同 building建造 blocks
173
499331
2330
而使用同樣的建構要素,
是那麼誘人,
08:37
because they work so well, they're cheap低廉,
174
501661
2264
因為它們運作得很理想,它們便宜,
08:39
they're repeated重複, and so on.
175
503925
1100
他們可以重覆利用,等等。
08:40
So all of our systems系統, more and more,
176
505025
2772
所以我們所有的系統,愈來愈多,
08:43
are starting開始 to use the same相同 technology技術
177
507797
1780
開始使用這同一項科技
08:45
and starting開始 to depend依靠 on this technology技術.
178
509577
2067
且開始依賴這項科技。
08:47
And so even a modern現代 rocket火箭 ship these days
179
511644
2425
因此就算是現代的火箭
08:49
actually其實 uses使用 Internet互聯網 protocol協議 to talk
180
514069
2726
實際上也使用網際網路來做
08:52
from one end結束 of the rocket火箭 ship to the other.
181
516795
1816
兩火箭之間的通訊。
08:54
That's crazy. It was never designed設計 to do things like that.
182
518611
3186
這很瘋狂。網路並不是為了這個設計的。
08:57
So we've我們已經 built內置 this system系統
183
521797
3117
所以我們已經建構這個系統,
09:00
where we understand理解 all the parts部分 of it,
184
524914
3110
在這系統中我們了解每一個部份,
09:03
but we're using運用 it in a very, very different不同 way than we expected預期 to use it,
185
528024
3698
但我們使用的方法
完完全全超出我們預設的模式,
09:07
and it's gotten得到 a very, very different不同 scale規模
186
531722
2439
而它已經到了和一個原本設計
非常非常不一樣的境界。
09:10
than it was designed設計 for.
187
534161
2171
而它已經到了和一個原本設計
非常非常不一樣的境界。
09:12
And in fact事實, nobody沒有人 really exactly究竟 understands理解
188
536332
2676
而實際上,沒有人真正完全了解
09:14
all the things it's being存在 used for right now.
189
539008
2263
現今用到的所有網路科技。
09:17
It's turning車削 into one of these big emergent應急 systems系統
190
541271
2503
它已經變成眾多巨大不穩定系統
的其中之一,
09:19
like the financial金融 system系統, where we've我們已經 designed設計 all the parts部分
191
543774
3481
好比說是經濟體系,我們設計了它的每一個部份
09:23
but nobody沒有人 really exactly究竟 understands理解
192
547255
2606
但是根本沒有人完全了解
09:25
how it operates操作 and all the little details細節 of it
193
549861
3227
它怎麼運作、
它的所有小細節、
09:28
and what kinds of emergent應急 behaviors行為 it can have.
194
553088
2719
以及可能會有什麼意外事故。
09:31
And so if you hear an expert專家 talking about the Internet互聯網
195
555807
3133
因此如果你聽到一個專家在談論網路
09:34
and saying it can do this, or it does do this, or it will do that,
196
558940
2695
並聲稱它可以做這事、或是它確實做了這事、
又或者它將可以做這事,
09:37
you should treat對待 it with the same相同 skepticism懷疑論
197
561635
2461
你應該要抱持同樣的懷疑,
09:39
that you might威力 treat對待 the comments註釋 of an economist經濟學家 about the economy經濟
198
564096
4415
就是那種面對經濟專家談論經濟時
會有的懷疑
09:44
or a weatherman天氣預報 about the weather天氣, or something like that.
199
568511
2267
或是面對預報員報氣象時的懷疑,
或是類似的事。
09:46
They have an informed通知 opinion意見,
200
570778
2648
他們有有根據的主張,
09:49
but it's changing改變 so quickly很快 that even the experts專家
201
573426
2541
但它瞬息萬變,即使是專家學者
09:51
don't know exactly究竟 what's going on.
202
575967
1818
也不完全了解全盤的運作。
09:53
So if you see one of these maps地圖 of the Internet互聯網,
203
577785
2692
所以如果你見到一些網路地圖,
09:56
it's just somebody's某人的 guess猜測.
204
580477
1666
那不過是某人的猜測而已。
09:58
Nobody沒有人 really knows知道 what the Internet互聯網 is right now
205
582143
2198
沒人真的知道現在的網路是什麼樣子
10:00
because it's different不同 than it was an hour小時 ago.
206
584341
2741
因為它和一小時之前的樣子
就已經不一樣了。
10:02
It's constantly經常 changing改變. It's constantly經常 reconfiguring重新配置.
207
587082
2801
它持續地改變中。
也持續地改變設定。
10:05
And the problem問題 with it is,
208
589883
1514
而問題是,
10:07
I think we are setting設置 ourselves我們自己 up for a kind of disaster災害
209
591397
3341
我想我們正在把自己置於災難之中
10:10
like the disaster災害 we had in the financial金融 system系統,
210
594738
2795
就像是我們在經濟體系裡面臨的災難一般,
10:13
where we take a system系統 that's basically基本上 built內置 on trust相信,
211
597533
5344
那也是一個類似的系統,
它基本上建構於信任之上、
10:18
was basically基本上 built內置 for a smaller-scale規模較小 system系統,
212
602877
2621
它基本上也是為小規模的系統而建立、
10:21
and we've我們已經 kind of expanded擴大 it way beyond the limits範圍
213
605498
2909
而我們同樣有點把它擴張
並超過
10:24
of how it was meant意味著 to operate操作.
214
608407
1996
它原先預定運作的極限。
10:26
And so right now, I think it's literally按照字面 true真正
215
610403
3268
所以現在,我想它是對的,
10:29
that we don't know what the consequences後果
216
613671
3505
就是說我們並不知道
當網路遇到
10:33
of an effective有效 denial-of-service拒絕服務 attack攻擊
217
617176
2434
有效的阻斷服務攻擊時
10:35
on the Internet互聯網 would be,
218
619610
1773
會是什麼樣子,
10:37
and whatever隨你 it would be is going to be worse更差 next下一個 year,
219
621383
1874
而不論怎樣的結果
都將會一年
10:39
and worse更差 next下一個 year, and so on.
220
623257
1408
比一年更糟。
10:40
But so what we need is a plan計劃 B.
221
624665
2549
所以我們須要 B 計劃。
10:43
There is no plan計劃 B right now.
222
627214
1632
現今並沒有這項替代計劃。
10:44
There's no clear明確 backup備用 system系統 that we've我們已經 very carefully小心 kept不停
223
628846
3503
也沒有明顯的備援系統,
讓我們得以小心地
10:48
to be independent獨立 of the Internet互聯網,
224
632349
1946
避免對網路的依賴,
10:50
made製作 out of completely全然 different不同 sets of building建造 blocks.
225
634295
3066
就是那個已經和原本建構理念
完全不一樣的網路系統。
10:53
So what we need is something that doesn't necessarily一定
226
637361
3013
所以我們所須要的是
10:56
have to have the performance性能 of the Internet互聯網,
227
640374
2734
某種不見得須要網路、
10:59
but the police警察 department has to be able能夠
228
643108
1517
而警察局得以在沒有網路的情況下
11:00
to call up the fire department even without the Internet互聯網,
229
644625
2523
也可以和消防局通聯、
11:03
or the hospitals醫院 have to order訂購 fuel汽油 oil.
230
647148
2565
且醫院也可以訂購燃油。
11:05
This doesn't need to be a multi-billion-dollar數十億美元 government政府 project項目.
231
649713
4614
這並不須要一個
數十億美元的政府計劃。
11:10
It's actually其實 relatively相對 simple簡單 to do, technically技術上,
232
654327
2748
這實際上是相對來說簡單的,
技術上來說,
11:12
because it can use existing現有 fibers纖維 that are in the ground地面,
233
657075
3802
因為可以利用
現有那些埋在地裡的光纖、
11:16
existing現有 wireless無線 infrastructure基礎設施.
234
660877
1959
現有的無線網路設施。
11:18
It's basically基本上 a matter of deciding決定 to do it.
235
662836
2765
這基本上只是
有沒有決心要做的問題。
11:21
But people won't慣於 decide決定 to do it
236
665601
2484
但人們在體認到對這計劃的需求前,
11:23
until直到 they recognize認識 the need for it,
237
668085
2379
根本不會下定決心去做,
11:26
and that's the problem問題 that we have right now.
238
670464
1498
而這才是我們現在所面臨的問題。
11:27
So there's been plenty豐富 of people,
239
671962
2783
所以有一群人、
11:30
plenty豐富 of us have been quietly悄悄 arguing爭論
240
674745
3044
一群我們的伙伴
已經默默地主張
11:33
that we should have this independent獨立 system系統 for years年份,
241
677789
2918
「我們應該要有有獨立的系統」許多年,
11:36
but it's very hard to get people focused重點 on plan計劃 B
242
680707
3009
但要人們專注於 B 計劃是非常困難的,
11:39
when plan計劃 A seems似乎 to be working加工 so well.
243
683716
3674
尤其是當 A 計劃看起來是這麼完美的情況下。
11:43
So I think that, if people understand理解
244
687390
3429
所以我在想,如果人們能理解
11:46
how much we're starting開始 to depend依靠 on the Internet互聯網,
245
690819
3054
我們對網路的依賴有多深、
11:49
and how vulnerable弱勢 it is,
246
693873
1977
以及網路有多脆弱,
11:51
we could get focused重點 on
247
695850
2106
我們就可以專注於
11:53
just wanting希望 this other system系統 to exist存在,
248
697956
3024
對其它系統存在的渴望,
11:56
and I think if enough足夠 people say, "Yeah, I would like to use it,
249
700980
3077
而我想如果有足夠的人說:
「對,我想要使用它,
11:59
I'd like to have such這樣 a system系統," then it will get built內置.
250
704057
3010
我想要有這樣的一個系統。」那它就會實現。
12:02
It's not that hard a problem問題.
251
707067
1423
這不是那麼困難的問題。
12:04
It could definitely無疑 be doneDONE by people in this room房間.
252
708490
3235
這當然可以靠在座的各位來完成。
12:07
And so I think that this is actually其實,
253
711725
4359
所以我想這實際上,
12:11
of all the problems問題 you're going to hear about at the conference會議,
254
716084
3179
在所有你在這次會議會聽到的議題之中,
12:15
this is probably大概 one of the very easiest最簡單的 to fix固定.
255
719263
2678
這也許是最容易解決的一個。
12:17
So I'm happy快樂 to get a chance機會 to tell you about it.
256
721941
2767
所以我很高興有這個機會來告訴你們。
12:20
Thank you very much.
257
724708
2611
謝謝。
12:23
(Applause掌聲)
258
727319
3854
(掌聲)
Translated by Jephian Lin
Reviewed by hunghsin sung

▲Back to top

ABOUT THE SPEAKER
Danny Hillis - Computer theorist
Inventor, scientist, author, engineer -- over his broad career, Danny Hillis has turned his ever-searching brain on an array of subjects, with surprising results.

Why you should listen

Danny Hillis is an inventor, scientist, author and engineer. While completing his doctorate at MIT, he pioneered the concept of parallel computers that is now the basis for graphics processors and cloud computing. He holds more than 300 US patents, covering parallel computers, disk arrays, forgery prevention methods, various electronic and mechanical devices, and the pinch-to-zoom display interface. He has recently been working on problems in medicine as well. He is also the designer of a 10,000-year mechanical clock, and he gave a TED Talk in 1994 that is practically prophetic. Throughout his career, Hillis has worked at places like Disney, and now MIT and Applied Invention, always looking for the next fascinating problem.

More profile about the speaker
Danny Hillis | Speaker | TED.com