237 A Strategy for Great Work [English].srt

1
00:00:11,026 --> 00:00:17,111
hello

2
00:00:18,011 --> 00:00:21,011
thank you very much for coming today and
if you're watching later on video thanks

3
00:00:21,011 --> 00:00:22,084
so much for watching i'm can cash and a

4
00:00:22,084 --> 00:00:25,086
and their session is called strategy

5
00:00:26,004 --> 00:00:32,066
for a great work so think we all try to
do great work I certainly

6
00:00:32,066 --> 00:00:36,100
always try to do great work regardless
if the result in the outcome

7
00:00:37,000 --> 00:00:40,052
certainly the effort is directed towards
doing something great

8
00:00:40,052 --> 00:00:42,121
I think that probably the same for all
love you that's why you're here at the

9
00:00:43,021 --> 00:00:43,112
conference today

10
00:00:44,012 --> 00:00:48,051
that may be why you're watching later on
video houses subject thats

11
00:00:48,051 --> 00:00:53,060
very important to me I I like talking
about like thinking about it and I was

12
00:00:53,006 --> 00:00:55,285
left on stage in 2012 with a

13
00:00:55,879 --> 00:00:59,820
I talk all bleak basics plus habits I
talked with him

14
00:00:59,082 --> 00:01:03,501
some tactical concern some real soon not
in both think about how to setup project

15
00:01:04,239 --> 00:01:08,990
and then had run projects once you have
them going in this is very

16
00:01:08,099 --> 00:01:12,108
tactical served any details things like
code review

17
00:01:13,008 --> 00:01:17,075
when you know working with fewer other
people that you might have on a team

18
00:01:17,075 --> 00:01:21,014
so today you know it different I i would
like to

19
00:01:21,689 --> 00:01:25,711
in this session talk about from higher
level ideas kinda move it up to

20
00:01:25,909 --> 00:01:28,940
the level of strategy rather than
tactics

21
00:01:29,219 --> 00:01:32,820
and so I went out and got a helpful
quotation

22
00:01:32,082 --> 00:01:37,100
from Wikipedia to a started thinking
about what the difference between

23
00:01:38,000 --> 00:01:39,045
strategy and tactics

24
00:01:39,045 --> 00:01:44,103
is into tactics to the actual means used
to gain an objective

25
00:01:45,003 --> 00:01:49,081
what strategy the overall campaign plan.
which may involve complex operational

26
00:01:49,081 --> 00:01:50,090
patterns activity in decision-making

27
00:01:51,071 --> 00:01:54,790
that lead to tactical execution
certainly the a complex operational

28
00:01:55,429 --> 00:01:58,472
pattern sounds like programming to me
right I

29
00:01:58,859 --> 00:02:02,310
definitely so you know I think that
really

30
00:02:02,031 --> 00:02:06,580
30 strategy yes you have been a big idea
what you might want to do

31
00:02:06,859 --> 00:02:10,670
and that really should hopefully be
driving what you actually do to

32
00:02:10,067 --> 00:02:11,074
accomplish the goal

33
00:02:11,074 --> 00:02:16,076
right so today I'd like to talk about
some of these ideas and high-level ideas

34
00:02:16,076 --> 00:02:17,153
that inspire and inform

35
00:02:18,053 --> 00:02:21,131
my every day work but really

36
00:02:22,031 --> 00:02:26,054
this session is not you know sir how to
make a great project in 21 days

37
00:02:26,054 --> 00:02:29,128
try two minutes it's not going to be
it's never that simple right every time

38
00:02:30,028 --> 00:02:32,113
you see something like that axe
experience as a 21-day

39
00:02:33,013 --> 00:02:37,026
its never ever ever that simple so

40
00:02:37,026 --> 00:02:40,059
for simple kinda media need a plan for
kinda

41
00:02:40,059 --> 00:02:43,140
attacking the problem here and so today

42
00:02:44,004 --> 00:02:47,004
how many do that is I tell you some of
my stories

43
00:02:47,004 --> 00:02:50,061
project that I've been involved in um

44
00:02:50,097 --> 00:02:53,175
but you know there's kinda like a note
of caution when you hear stories

45
00:02:54,075 --> 00:02:58,076
and maybe try to apply it to your own
life and I have an example here like

46
00:02:58,085 --> 00:02:59,119
failure in Las Vegas

47
00:03:00,019 --> 00:03:04,066
and you see somebody go by was carrying
a big pile of chips already get them

48
00:03:04,066 --> 00:03:04,143
cashed in

49
00:03:05,043 --> 00:03:09,119
you stop the person said we what did you
do to get that big pile of chips

50
00:03:10,019 --> 00:03:13,025
and their to come I put it all on 14

51
00:03:13,079 --> 00:03:16,104
K so

52
00:03:17,004 --> 00:03:21,041
its true but not so useful to you

53
00:03:21,041 --> 00:03:25,055
right you go over and put it on 14 you
may or may not get the same result most

54
00:03:25,055 --> 00:03:25,138
likely not

55
00:03:26,038 --> 00:03:29,134
so hopefully the stories will be
interesting to you

56
00:03:30,034 --> 00:03:34,059
but the real goal before during and
after the story

57
00:03:34,059 --> 00:03:37,153
rates particularly after is to extract
out the lessons from the stories

58
00:03:38,053 --> 00:03:43,054
because I think the lessons if you take
them in the aggregate starts stringing

59
00:03:43,063 --> 00:03:44,101
together some russians

60
00:03:45,001 --> 00:03:49,002
serve points toward the strategy it
certainly does for me

61
00:03:49,002 --> 00:03:52,021
today eight stories

62
00:03:52,021 --> 00:03:56,054
eleven lessons take all those lessons
put them together and

63
00:03:56,054 --> 00:03:59,079
really that that is what informs my work

64
00:03:59,079 --> 00:04:02,094
and again of course I'm always trying to
do great work so hopefully this will

65
00:04:02,094 --> 00:04:04,139
this will all come together a nice pics
ok with the end

66
00:04:05,039 --> 00:04:08,111
soap to begin with stories

67
00:04:09,011 --> 00:04:14,067
number one stories like a crystal ball
in the election

68
00:04:14,067 --> 00:04:19,916
is no a good idea when you see it so

69
00:04:20,519 --> 00:04:24,280
I joined Apple in June of 2001 and if
you

70
00:04:24,028 --> 00:04:27,034
you think back to that time on Mac OS

71
00:04:27,034 --> 00:04:30,703
10 version 10 dotto was released

72
00:04:31,009 --> 00:04:34,240
about three months before now

73
00:04:34,024 --> 00:04:38,091
it's come a long time ago we think about
it the browser on the platform

74
00:04:38,091 --> 00:04:42,183
was Microsoft Internet Explorer it
that's the browser that shipped

75
00:04:43,083 --> 00:04:46,098
with the operating system from Apple

76
00:04:46,098 --> 00:04:49,149
now we've had there was no framework

77
00:04:50,049 --> 00:04:54,101
to load and render web content

78
00:04:55,001 --> 00:04:58,045
right so we've just had a nap but no
framework so my job

79
00:04:58,045 --> 00:05:02,122
long with someone else going Don we
joined same day our job became

80
00:05:03,022 --> 00:05:06,058
make apple a web browser and in
objective-c framework

81
00:05:06,058 --> 00:05:11,071
to again loading render rate web content
and make that available for the PI to

82
00:05:11,071 --> 00:05:13,099
you all to bed developers

83
00:05:13,099 --> 00:05:16,106
so

84
00:05:17,006 --> 00:05:20,054
we start looking at my options were what
were the options available to time in

85
00:05:20,054 --> 00:05:21,080
you see

86
00:05:21,008 --> 00:05:24,046
through the leading candidates there
some of these where

87
00:05:25,018 --> 00:05:28,062
our commercial products Apapa high
option

88
00:05:28,062 --> 00:05:32,123
some of them were open source project
out there in the world going on

89
00:05:33,023 --> 00:05:36,116
arm we could you know go and meet
perhaps adopt that source code for

90
00:05:37,016 --> 00:05:40,033
procter we can also start from scratch
we considered it

91
00:05:40,033 --> 00:05:43,036
um absolutely so my job

92
00:05:43,036 --> 00:05:46,058
you know to begin to do some technical
investigation up in a particularly the

93
00:05:46,058 --> 00:05:47,103
open-source

94
00:05:48,003 --> 00:05:51,004
I options I am

95
00:05:51,004 --> 00:05:55,353
and can i download Mozilla their toes 11
story II took

96
00:05:55,389 --> 00:06:00,180
per week I remember to try to just get
Mozilla to build on the Mac

97
00:06:00,018 --> 00:06:05,059
and I remember at one point I am I
wanted the steps in this long recipe if

98
00:06:05,059 --> 00:06:07,128
steps that need to get done to get
muscle to build

99
00:06:08,028 --> 00:06:12,085
I need to rebuild the the Metro works
the library

100
00:06:12,085 --> 00:06:15,163
was one of the steps so this was not a
straightforward thing but he eventually

101
00:06:16,063 --> 00:06:17,099
got a building in you know

102
00:06:17,099 --> 00:06:21,118
okay um but this disc process went on
for weeks

103
00:06:22,018 --> 00:06:25,022
you know when I i went you know and and
got these you know the server these

104
00:06:25,058 --> 00:06:28,095
programs the commercial prop products
you know that what he did for that was

105
00:06:28,095 --> 00:06:29,099
you know got them running in

106
00:06:30,035 --> 00:06:34,120
30 doing some performance analysis and
say you know unico compatibility in

107
00:06:35,002 --> 00:06:38,056
and so to between the commercial
products though those test results in

108
00:06:38,074 --> 00:06:39,129
the open source project

109
00:06:40,029 --> 00:06:43,113
some sort if I reckoning okay how much
source code was their

110
00:06:44,013 --> 00:06:47,018
how easy it would be to adapted and so
forth her

111
00:06:47,018 --> 00:06:50,079
you know to test results showing I
printed out on paper

112
00:06:50,079 --> 00:06:53,126
but I couldn't have a demo

113
00:06:54,026 --> 00:06:57,038
no damn just test results

114
00:06:57,038 --> 00:07:01,072
to write about that time getting new
engineer at our team the winning Richard

115
00:07:01,072 --> 00:07:05,075
and so I showed him you know described
to him basically a to describe to you

116
00:07:05,075 --> 00:07:06,150
you know what I've been doing

117
00:07:07,005 --> 00:07:11,087
I showed him the test result in he said
really what have you been to a

118
00:07:12,032 --> 00:07:15,032
six weeks in this is all you have so

119
00:07:15,032 --> 00:07:18,094
you know if I quote yeah I just gonna
shrugged and said that I have

120
00:07:18,094 --> 00:07:21,158
for he just didn't think very much of
that and so he went away and two days

121
00:07:22,058 --> 00:07:22,133
later

122
00:07:23,033 --> 00:07:28,092
he had a demo he took one of those
options KHTML

123
00:07:28,389 --> 00:07:33,090
which is the the the web rendering
engine what that was parted the key D E

124
00:07:33,009 --> 00:07:36,046
desktop environment for Linux heat
ground

125
00:07:36,046 --> 00:07:39,123
that source code got it running on a Mac

126
00:07:40,023 --> 00:07:44,098
just did whatever kind of hack said he
needed to do to get it running

127
00:07:44,098 --> 00:07:47,197
and had Conqueror Akii browser

128
00:07:48,097 --> 00:07:52,121
running on a Mac in the next window

129
00:07:53,021 --> 00:07:57,029
this was amazing

130
00:07:57,029 --> 00:08:00,100
absolutely amazing just standing there
in like looking

131
00:08:01,000 --> 00:08:05,056
at a browser running on the matter fi
can we say to what it was like

132
00:08:05,056 --> 00:08:10,062
a crystal ball whose absolutely amazing

133
00:08:10,062 --> 00:08:13,071
and we saw all the people who are
involved in you know not only are on the

134
00:08:13,071 --> 00:08:15,104
project but the decision makers at the
company

135
00:08:16,004 --> 00:08:19,036
saw that KHTML was our future this

136
00:08:19,036 --> 00:08:22,037
this was just like you opened up this
tunnel to

137
00:08:22,046 --> 00:08:25,115
like the future it was absolutely
amazing

138
00:08:26,015 --> 00:08:29,094
and so the lesson from this experience

139
00:08:29,094 --> 00:08:32,543
know a good idea when you see it really

140
00:08:33,389 --> 00:08:37,370
for me I kinda felt like an bumbling
idiot

141
00:08:37,037 --> 00:08:40,996
there I was for six weeks you know
trying to do you figure out what to do

142
00:08:41,329 --> 00:08:45,366
and really had nothing and someone new
came in and in two days

143
00:08:45,699 --> 00:08:48,732
he had the future right there

144
00:08:49,029 --> 00:08:53,034
that we could use now sooted

145
00:08:53,079 --> 00:08:56,040
lesson for this again is really what
does this mean you no good he when you

146
00:08:56,004 --> 00:08:57,293
see it is even if some

147
00:08:57,329 --> 00:08:59,980
something to like this happens to you
when you're made to look like an idiot

148
00:08:59,098 --> 00:08:59,827
don't

149
00:09:00,709 --> 00:09:03,788
doesn't matter I didn't care it
shouldn't matter to you either

150
00:09:04,499 --> 00:09:07,594
because there's actually some a couple
of good things about this experience

151
00:09:08,449 --> 00:09:11,498
number one I had a great new team member
who I was gonna be

152
00:09:11,939 --> 00:09:15,230
now involved it's going to be helping me
to come to make this

153
00:09:15,023 --> 00:09:20,572
project happen an he's really proved in
two days here's a really good

154
00:09:20,779 --> 00:09:24,821
I am and so is just like what it was in
Houston next day well we

155
00:09:25,199 --> 00:09:29,216
kinda know what to do right we're gonna
start you make in the future real

156
00:09:29,369 --> 00:09:32,371
and so

157
00:09:32,569 --> 00:09:35,590
that's kinda the story it like a crystal
ball but you know there's a bonus

158
00:09:35,779 --> 00:09:39,480
lesson here too not only know a good
idea when you see it

159
00:09:39,048 --> 00:09:42,377
but Canada extra leptin up make

160
00:09:42,809 --> 00:09:47,730
demos and prototypes if you're starting
a new idea

161
00:09:47,073 --> 00:09:50,099
don't him test results her at all well
and good

162
00:09:50,099 --> 00:09:54,448
but you should also at the same time
make something

163
00:09:55,339 --> 00:09:58,680
that you can actually begin to use

164
00:09:58,068 --> 00:10:01,577
like you want to use the real thing that
you're gonna be making ultimately in the

165
00:10:02,189 --> 00:10:03,220
future

166
00:10:03,022 --> 00:10:06,055
get that open up your own travel to the
future

167
00:10:06,055 --> 00:10:09,404
right get something in front view and
working

168
00:10:09,899 --> 00:10:12,986
as soon as you can that's like a crystal
ball

169
00:10:13,769 --> 00:10:20,050
no good idea when you see it and make
demos and prototypes

170
00:10:20,005 --> 00:10:23,324
so the next

171
00:10:23,369 --> 00:10:26,371
story number two is the black

172
00:10:26,389 --> 00:10:30,040
abolished and the lesson is don't try to
solve every problem might want

173
00:10:30,004 --> 00:10:33,583
so continuing right on from the end of
that last arc HTML

174
00:10:33,619 --> 00:10:37,632
fantastic we group good looking the more
we look at the code

175
00:10:37,749 --> 00:10:41,240
the more we are confident that this was
really the

176
00:10:41,024 --> 00:10:44,080
the way for us to go but really what do
we do now

177
00:10:44,008 --> 00:10:48,075
rate me when we to right and so you
imagine open source we

178
00:10:49,047 --> 00:10:52,326
going you know party the Katy desktop so
we went

179
00:10:52,749 --> 00:10:57,806
and we downloaded all Katie but the
thing is we only want idk HTML only

180
00:10:58,319 --> 00:11:01,410
each email a rendering engine

181
00:11:01,041 --> 00:11:04,390
and we also want to KJs the JavaScript
engine so

182
00:11:04,759 --> 00:11:08,765
the job became to serb draw a line
around

183
00:11:08,819 --> 00:11:13,844
just third key HTML part have this whole
desktop environment

184
00:11:14,069 --> 00:11:18,105
so that we could just extract that use
it has turned out to be about a hundred

185
00:11:18,429 --> 00:11:21,492
source files seems tractable

186
00:11:22,059 --> 00:11:25,410
so the next step was to bring those
files or the Mac

187
00:11:25,041 --> 00:11:30,074
after compiling them one at a time right

188
00:11:30,074 --> 00:11:33,783
critic didn't compile because we just
took the source code expected this whole

189
00:11:34,449 --> 00:11:37,490
other desktop environment around it

190
00:11:37,049 --> 00:11:40,648
and now we just heard for the whole rug
out from under it

191
00:11:41,089 --> 00:11:45,124
so you know it didn't compile we had a
whole bunch of missing things and so we

192
00:11:45,439 --> 00:11:46,483
just began

193
00:11:46,879 --> 00:11:49,980
stabbing them in stopping in missing
functions one at a time

194
00:11:49,098 --> 00:11:54,197
and these were very very light that
these were only stops in header files

195
00:11:55,097 --> 00:11:59,180
right to it only wanted to it surged
define the problem in terms of all the

196
00:12:00,008 --> 00:12:01,033
missing things

197
00:12:02,005 --> 00:12:06,022
that we needed to ed this was very very
slow work

198
00:12:06,022 --> 00:12:09,171
it took about another six weeks six
weeks is kinda magical number

199
00:12:09,369 --> 00:12:12,376
for this this project this took about
six weeks it just

200
00:12:12,439 --> 00:12:15,444
compiling II it would be like

201
00:12:15,939 --> 00:12:19,470
channel 11 hours of compiling a every
single day

202
00:12:19,047 --> 00:12:22,079
I am so it was pretty I'm

203
00:12:22,079 --> 00:12:26,158
pretty painstaking and then even at this
that nothing link because it makes these

204
00:12:26,869 --> 00:12:27,944
are only header files

205
00:12:28,619 --> 00:12:31,970
through her only running we were tight I
made a custom make faster don't even try

206
00:12:31,097 --> 00:12:33,816
to run the linker just just like you try
to

207
00:12:34,689 --> 00:12:38,730
show us what we were missing I'm

208
00:12:38,073 --> 00:12:41,992
so after we were done with that right we
had another six weeks

209
00:12:42,649 --> 00:12:45,680
weeks to get everything to link actually
going in and

210
00:12:45,959 --> 00:12:49,006
finishing out so did the header file
actually putting in

211
00:12:49,429 --> 00:12:52,510
you know some stub implementation for
everything and the even

212
00:12:53,239 --> 00:12:57,470
then we had lots of crashes because in a
lot of functions that return strings we

213
00:12:57,047 --> 00:13:00,776
returned empty strings and you know it
obviously there to lower not really what

214
00:13:01,199 --> 00:13:03,100
a lot of the code was expecting so

215
00:13:03,001 --> 00:13:07,640
we also added a whole lotta blog
statements to our implementation saying

216
00:13:07,739 --> 00:13:11,746
yeah this is not implemented at all i
mean it returns a value but edgy you

217
00:13:11,809 --> 00:13:11,847
know

218
00:13:12,189 --> 00:13:15,192
should not implement it all partially
implemented yea we think we're pretty

219
00:13:15,489 --> 00:13:15,566
close n

220
00:13:16,259 --> 00:13:19,341
implement that we had a whole bunch a
blog statements that

221
00:13:20,079 --> 00:13:23,097
we which fee will a whole bunch lawyer

222
00:13:23,259 --> 00:13:26,297
message is getting written to the log in
to track to try to use the program

223
00:13:26,639 --> 00:13:29,641
and after this whole process many many
weeks

224
00:13:29,839 --> 00:13:32,842
we eventually wound up with the Mac
program to used

225
00:13:32,869 --> 00:13:35,947
cage email and then one day

226
00:13:36,649 --> 00:13:39,654
the lights came on so my coworker
richard was actually off on vacation he

227
00:13:40,149 --> 00:13:42,241
was really kinda doing this you know
certain leading up the effort to

228
00:13:43,069 --> 00:13:46,550
get things to link in that crash so much
and then I came in and took over the

229
00:13:46,055 --> 00:13:46,129
process

230
00:13:47,029 --> 00:13:50,268
and actually got the first

231
00:13:50,529 --> 00:13:53,626
being to draw on the screen I looted

232
00:13:54,499 --> 00:13:59,920
a URL in this case WWW dot yahoo dot com

233
00:13:59,092 --> 00:14:02,166
and euros reams and reams and reams and
log messages go by

234
00:14:03,066 --> 00:14:06,155
in the program didn't crash then there
was a short pause

235
00:14:06,749 --> 00:14:09,823
and it drew

236
00:14:10,489 --> 00:14:14,572
a black obelisk

237
00:14:15,319 --> 00:14:18,321
so this is the first if the first thing
that WebKit you know

238
00:14:18,339 --> 00:14:22,170
a Safari ever ever you know what became
Safari ever rendered

239
00:14:22,017 --> 00:14:26,026
now to see it little bit of a secret is
actually turned out if

240
00:14:26,179 --> 00:14:29,217
that this was reversed so that the
publisher been on the others are decided

241
00:14:29,559 --> 00:14:30,980
to scream but you know that

242
00:14:30,098 --> 00:14:33,557
we we fix that took a little geometry
problem very soon after

243
00:14:34,439 --> 00:14:37,447
but you can imagine did you know that
once I got this going at Sky Go Go

244
00:14:37,519 --> 00:14:40,536
infection everybody who was disclosed in
the project were all sitting around for

245
00:14:40,689 --> 00:14:42,230
like 2001

246
00:14:42,023 --> 00:14:47,122
face Odyssey with this black obelisk
going who

247
00:14:47,329 --> 00:14:52,415
so that was a a very happy happy moment
now the lesson from this experience

248
00:14:53,189 --> 00:14:56,210
is don't try to solve every problem it
won't

249
00:14:56,399 --> 00:15:00,410
at this point we had almost no knowledge

250
00:15:00,509 --> 00:15:03,552
of how the innards have Keech Gmail
worked

251
00:15:03,939 --> 00:15:06,947
this was simply a mechanical process

252
00:15:07,019 --> 00:15:11,072
of making the compiler happy making the
linker happy

253
00:15:11,549 --> 00:15:15,551
making it so the program didn't crash in
we try to do something with it

254
00:15:15,569 --> 00:15:18,664
it was all very very mechanical the
compiler says

255
00:15:19,519 --> 00:15:22,593
there's something missing go where that
thing so that it no longer missing

256
00:15:23,259 --> 00:15:27,302
right now it turns out of course a a web
rendering engine is a very very

257
00:15:27,689 --> 00:15:28,766
complicated piece of software

258
00:15:29,459 --> 00:15:32,528
an but we just could we didn't have the
band with

259
00:15:33,149 --> 00:15:36,165
to think about it at this point as we
were trying to just get those lights to

260
00:15:36,309 --> 00:15:37,330
come on

261
00:15:37,519 --> 00:15:43,350
that began to happen as soon as we had a
program that would no longer crash right

262
00:15:43,035 --> 00:15:47,544
don't care if you guys trying to attack
in New big problem you've got this big

263
00:15:47,859 --> 00:15:48,922
problem space particularly

264
00:15:49,489 --> 00:15:52,910
if that problem space is new to you to
try to figure out what the most

265
00:15:52,091 --> 00:15:53,710
important thing is to do first

266
00:15:54,529 --> 00:15:58,290
and do that don't think about the whole
prom or try to

267
00:15:58,029 --> 00:16:01,198
operate on the whole solving the whole
problem

268
00:16:01,459 --> 00:16:04,462
at one time right don't

269
00:16:04,759 --> 00:16:10,852
try to solve every problem at once and
that's the plank

270
00:16:11,689 --> 00:16:14,689
so the next story number three is

271
00:16:14,689 --> 00:16:17,697
the hardest problem and the lesson is
find

272
00:16:18,489 --> 00:16:21,504
smart friends and listen to them so

273
00:16:21,639 --> 00:16:24,721
fast-forward to the fall of 2003 in my
job became too ed

274
00:16:25,459 --> 00:16:30,527
editing to WebKit so we wanted to use
WebKit edit HDMI male

275
00:16:31,139 --> 00:16:35,156
in mailed out a app on back with 10 we
want to make

276
00:16:35,309 --> 00:16:38,383
editing in WebKit lookin work like text
editor

277
00:16:39,049 --> 00:16:42,100
K to the problem it is kinda been unit
printed

278
00:16:42,559 --> 00:16:45,593
problem that we try to solve is somewhat
obvious right a lot of

279
00:16:45,899 --> 00:16:48,978
other platforms in particular I window
sends

280
00:16:49,689 --> 00:16:52,723
you know HTML mail from Outlook and we
wanted to

281
00:16:53,029 --> 00:16:57,053
not only render it which we could now
that we had WebKit but we also wanted to

282
00:16:57,269 --> 00:17:01,280
edited because if you think about it
before it WebKit had editing

283
00:17:01,379 --> 00:17:05,384
the web content which open your mail
message as an attachment in RTF talking

284
00:17:05,429 --> 00:17:05,505
with big

285
00:17:06,189 --> 00:17:09,140
breaking insertion point after you
discovered delete the whole thing with

286
00:17:09,014 --> 00:17:09,020
one

287
00:17:09,074 --> 00:17:12,111
to eat an entire web page with one one
backspace

288
00:17:13,011 --> 00:17:16,049
so that's not what we wanted we want to
be able to get in to the details of the

289
00:17:16,049 --> 00:17:17,768
web content to get it right

290
00:17:18,209 --> 00:17:21,266
so now the hardest problem

291
00:17:21,779 --> 00:17:24,796
I have ever ever worked on in my well
many

292
00:17:24,949 --> 00:17:28,370
many years of programming is arrow
navigation

293
00:17:28,037 --> 00:17:31,084
through web content now what do I mean
by that so

294
00:17:31,084 --> 00:17:35,086
picture a a a wet tiny little web
document like this

295
00:17:35,086 --> 00:17:38,169
and what you do then is you hit the
arrow key the right number of times

296
00:17:39,069 --> 00:17:42,101
and the insertion point wind up at the
end that's it

297
00:17:43,001 --> 00:17:46,059
disco character by character have the
insertion point chopin the right place

298
00:17:46,059 --> 00:17:50,008
each time you press the arrow key and
then similarly

299
00:17:50,539 --> 00:17:53,490
press the R key to go back the other way
in San the insertion point back to the

300
00:17:53,049 --> 00:17:53,698
start

301
00:17:54,139 --> 00:17:58,350
that was it this was turned out to be
very very difficult know why is this

302
00:17:58,035 --> 00:17:58,098
because

303
00:17:58,098 --> 00:18:01,122
you know on white thing RTF document
which is just too

304
00:18:02,022 --> 00:18:08,044
series of characters in indexes into
basically nureyev characters

305
00:18:08,044 --> 00:18:11,075
what content is a is a tree structure so

306
00:18:11,075 --> 00:18:14,083
if you see on the top where the
insertion point is

307
00:18:14,083 --> 00:18:17,432
at the n and then if you look at the the
model

308
00:18:18,179 --> 00:18:22,070
the web document the insertion point
would we get the in

309
00:18:22,007 --> 00:18:26,010
if in the web document insertion point
was

310
00:18:26,037 --> 00:18:29,069
their or their or their

311
00:18:29,069 --> 00:18:34,125
or their right so what I had he was a
complex model the relationship with the

312
00:18:35,025 --> 00:18:36,070
model changed

313
00:18:36,007 --> 00:18:39,073
but the view didn't change me

314
00:18:40,036 --> 00:18:44,325
senator begin thinking about this I used
the DOM the the web don the Document

315
00:18:44,649 --> 00:18:47,677
Object Model that the API for dressing
web content

316
00:18:47,929 --> 00:18:50,976
and to use positions in the web

317
00:18:51,399 --> 00:18:54,476
content to try to understand where to
make the insertion point blank

318
00:18:55,169 --> 00:18:58,231
and so i think is illustrated I came up
with this notion have an equivalent

319
00:18:58,789 --> 00:19:02,200
Dom positions compositions

320
00:19:02,002 --> 00:19:05,331
that would make different compositions
but that would make the insertion point

321
00:19:05,529 --> 00:19:07,760
blank in the same spot

322
00:19:07,076 --> 00:19:11,119
K so I started writing some code for
simple right al

323
00:19:12,019 --> 00:19:15,024
arrow algorithm here get the insertion
point position

324
00:19:15,069 --> 00:19:18,161
in a DOM position then fine dysfunction
in use this function

325
00:19:19,061 --> 00:19:23,074
equivalent downstream position to find
the DOM position

326
00:19:23,074 --> 00:19:26,100
that which still make the insertion
point blank in the same spot but was

327
00:19:27,000 --> 00:19:30,008
closer to the end of the document

328
00:19:30,008 --> 00:19:33,217
and then go to the next position and
then set the insertion point there now

329
00:19:33,289 --> 00:19:36,340
that seems reasonable but kinda focus in
on these two

330
00:19:36,799 --> 00:19:41,120
important functions here right will
naturally it's a program right image if

331
00:19:41,012 --> 00:19:41,671
you wanted to

332
00:19:41,779 --> 00:19:45,803
perfect results the functions need to
work perfectly in the problem with that

333
00:19:46,019 --> 00:19:49,096
was the equivalent downstream position
was implemented in terms of get next

334
00:19:49,789 --> 00:19:50,570
composition

335
00:19:50,057 --> 00:19:53,062
sometimes are have this feedback

336
00:19:53,062 --> 00:19:58,097
process going were bizarre things happen
because above wouldn't feed into itself

337
00:19:58,097 --> 00:20:01,123
right and cost something that was very
very difficult to understand and

338
00:20:02,023 --> 00:20:02,094
diagnose

339
00:20:02,094 --> 00:20:06,097
simply by looking at the result in the
view

340
00:20:06,097 --> 00:20:09,416
does complicated right to get right and
then

341
00:20:10,289 --> 00:20:13,470
other issues come up as well what if
you're already at the

342
00:20:13,047 --> 00:20:16,135
ended the document visual and if the
document as you know if you've ever

343
00:20:17,035 --> 00:20:20,244
probably viewed source in an HTML
document there's

344
00:20:20,559 --> 00:20:25,250
of in many many things at the very least
the cassava closing /body tag closing

345
00:20:25,025 --> 00:20:28,694
each email tech water times their script
content there might be

346
00:20:28,919 --> 00:20:32,570
you know time I P style content after
there could be any sorts

347
00:20:32,057 --> 00:20:36,796
sort of information that not visible
after the visible and if the document

348
00:20:37,309 --> 00:20:41,210
and by Tom position it would it would
you know if you just kind of fended off

349
00:20:41,021 --> 00:20:43,032
to position it o'Mara we go there

350
00:20:43,032 --> 00:20:45,074
and there won't be a place for the
insertion point to point to this is a

351
00:20:45,074 --> 00:20:46,083
problem as well

352
00:20:46,083 --> 00:20:50,087
and then what if you wanna move by word
because the ultimate goal was to make a

353
00:20:51,023 --> 00:20:51,095
text editor

354
00:20:51,095 --> 00:20:54,424
right to hold down the option key an
arrow in the new move word by word not

355
00:20:55,279 --> 00:20:57,070
character by character

356
00:20:57,007 --> 00:21:01,020
and so I was in know where I was stock
working on this

357
00:21:01,002 --> 00:21:04,591
for it sir was just three thank you very
good say six weekly

358
00:21:04,789 --> 00:21:07,790
magical number

359
00:21:07,799 --> 00:21:12,870
but for many weeks without really
feeling like I was making progress

360
00:21:12,087 --> 00:21:15,144
so I asked for help and fortunately I
had some very smart people on my team

361
00:21:16,044 --> 00:21:17,023
which course the

362
00:21:17,419 --> 00:21:21,190
two other stories have already
illustrated explain the problem

363
00:21:21,019 --> 00:21:25,348
to them much as I adjusted to you and
their solution came back

364
00:21:25,519 --> 00:21:29,580
very very clear the DOM positions were
too low level

365
00:21:29,058 --> 00:21:32,497
house trying to build a very a complex
edifice but I was using

366
00:21:33,019 --> 00:21:36,450
to you fundamental an abstraction

367
00:21:36,045 --> 00:21:39,654
to get the job done when you do more
powerful attraction so they came up with

368
00:21:40,059 --> 00:21:43,083
when they gave me what they call it
visible position

369
00:21:43,299 --> 00:21:46,347
and all this was this is super sport
coat is a Supersport class

370
00:21:46,779 --> 00:21:51,630
that represented our place in a document
where the insertion point and blink

371
00:21:51,063 --> 00:21:56,108
30 and then secondarily at library
called visible units

372
00:21:57,008 --> 00:22:01,021
they could take a visible position and
moved by words lines and documents

373
00:22:01,021 --> 00:22:04,980
okay so they gave me this design now the
point is

374
00:22:05,169 --> 00:22:09,460
that I still had the hard work of
figuring everything out

375
00:22:09,046 --> 00:22:13,105
but now I had a complete

376
00:22:13,519 --> 00:22:17,960
outline all the code that I needed

377
00:22:17,096 --> 00:22:20,685
everyplace word by word well I knew now
there was actually a place in the

378
00:22:21,549 --> 00:22:23,370
document is a stub function

379
00:22:23,037 --> 00:22:26,058
that was weird the move word by word

380
00:22:26,058 --> 00:22:29,071
code should go where that should live

381
00:22:29,071 --> 00:22:34,122
and so this dist really the set have
abstractions really helped me to just

382
00:22:35,022 --> 00:22:38,571
organize the work and figure out where
the solution should live

383
00:22:38,769 --> 00:22:43,100
it also made a easier to think about the
prom so that I went up with fewer

384
00:22:43,001 --> 00:22:46,580
these kind of functions that fed back on
each other producing

385
00:22:46,679 --> 00:22:50,090
bizarre results so

386
00:22:50,009 --> 00:22:53,558
when you're blocked don't be blocked

387
00:22:53,639 --> 00:22:57,620
find smart friends and listen to them in
this case they said visible position

388
00:22:57,062 --> 00:23:01,113
visible units I said okay

389
00:23:02,013 --> 00:23:05,792
I will do that and it worked

390
00:23:05,909 --> 00:23:09,440
absolutely work I certainly went back
and got some more they're helping

391
00:23:09,044 --> 00:23:11,048
details but this was the big big help

392
00:23:11,084 --> 00:23:14,100
find smart friends and listen to them
particularly when you're stuck

393
00:23:15,000 --> 00:23:18,097
to not stay blocked and this is how I
got through

394
00:23:18,097 --> 00:23:25,097
my hardest programming problem

395
00:23:26,649 --> 00:23:29,750
and so the next story number four is

396
00:23:29,075 --> 00:23:32,094
quality and the left in his work

397
00:23:32,094 --> 00:23:35,097
should explain itself

398
00:23:35,097 --> 00:23:38,182
so fast-forward to 2005 now

399
00:23:39,082 --> 00:23:42,671
and I joined the iPhone development
project pretty

400
00:23:43,409 --> 00:23:47,100
early on and initially I join to work on
WebKit

401
00:23:47,001 --> 00:23:50,700
because it seems like a a good match for
my skills and experience

402
00:23:50,799 --> 00:23:56,833
at Apple but I soon transition to work
on keyboards

403
00:23:57,139 --> 00:24:01,201
so now if you think about it right this
is kind of what we had when we started

404
00:24:01,759 --> 00:24:02,580
with the iPhone

405
00:24:02,058 --> 00:24:05,487
I we had a screen and we had this

406
00:24:06,009 --> 00:24:10,014
notion at the keyboard will be on the
bottom the app content would be on the

407
00:24:10,059 --> 00:24:10,118
top

408
00:24:10,649 --> 00:24:13,658
annie cruz the big question was well
what should the keyboard

409
00:24:14,549 --> 00:24:17,612
do how should it work watched it look
like

410
00:24:18,179 --> 00:24:21,990
and if you have seen you know even from
the very beginning that the

411
00:24:21,099 --> 00:24:23,758
you know that the Steve Jobs
announcement have the iPhone your you

412
00:24:24,649 --> 00:24:26,850
know the original announcement in 2007

413
00:24:26,085 --> 00:24:30,107
he pointed at very very clearly this
virtue of the iPhone that

414
00:24:31,007 --> 00:24:34,016
it didn't have those placed the keys on
the bottom we could draw the keyboard

415
00:24:34,097 --> 00:24:37,162
and make it look like whatever we want
it

416
00:24:38,062 --> 00:24:42,075
some doubt jumping back to the giving up
that story because

417
00:24:42,075 --> 00:24:45,162
even stages the end of that story right

418
00:24:46,062 --> 00:24:50,110
we didn't have placed the key so we need
to experiment what we're gonna do

419
00:24:51,001 --> 00:24:54,680
right how we gonna replace those plastic
keys with software

420
00:24:54,779 --> 00:24:57,816
and we came up with many many ideas

421
00:24:58,149 --> 00:25:01,158
for how we would do this now there are a
lot of challenges

422
00:25:02,049 --> 00:25:05,080
a particularly at the beginning when we
were only with dealing with

423
00:25:05,008 --> 00:25:08,019
touchscreen keypad particularly a i
touch screens

424
00:25:08,019 --> 00:25:11,075
a in general and touchscreen keyboard
specifically

425
00:25:11,075 --> 00:25:14,097
to the one in the challenges that this
small target too hard to tap its

426
00:25:14,097 --> 00:25:17,148
pretty logical I'm sure you you even
experience that now the small the tower

427
00:25:18,048 --> 00:25:20,099
at Target get the more difficult they
are to tap

428
00:25:20,099 --> 00:25:25,538
and not only not only that but as you as
you go for a target as you get close to

429
00:25:26,429 --> 00:25:27,450
it your finger

430
00:25:27,639 --> 00:25:30,830
covers up the thing that you're looking
for so the smaller than it is

431
00:25:30,083 --> 00:25:34,922
the more difficult it is to you know do
sumfin green a judgment as you get close

432
00:25:35,669 --> 00:25:39,870
for these were we we as in our
experiments we we were ex- seeing their

433
00:25:39,087 --> 00:25:42,496
result this was very challenging to make
a keyboard that worked at all

434
00:25:43,279 --> 00:25:47,070
and so in the beginning we we were
pretty clear we will do this we thought

435
00:25:47,007 --> 00:25:47,816
we were clear

436
00:25:47,879 --> 00:25:52,880
the bigger keys would be the solution
this would solve both problems

437
00:25:52,889 --> 00:25:57,570
so I have an example something in the
spirit

438
00:25:57,057 --> 00:26:00,086
apps wanna the experiments we tried not
an exact

439
00:26:00,086 --> 00:26:03,092
example but something very very much in
the spirit of what we try to call it the

440
00:26:04,046 --> 00:26:05,265
keyboard keyboard

441
00:26:05,679 --> 00:26:09,240
but if you think about you know about
other things in the world are you know

442
00:26:09,024 --> 00:26:14,037
accept a lot of touch input right was a
piano keyboard right

443
00:26:14,037 --> 00:26:19,846
so could maybe conceive a keyboard that
looks like this

444
00:26:20,179 --> 00:26:23,610
so how does this work well I'm gonna try
to protect the Tiki and so

445
00:26:23,061 --> 00:26:26,480
how do you get a key well you look on
all the keys

446
00:26:27,029 --> 00:26:30,330
and you do make a chord multi multi
touch motion

447
00:26:30,033 --> 00:26:37,033
to find all the keys are heavy on it

448
00:26:38,086 --> 00:26:42,132
okay so there you go he H age is
actually got to the same keys

449
00:26:43,032 --> 00:26:47,059
he by age is only on all those two keys
and and and Daiichi

450
00:26:47,059 --> 00:26:53,278
she can see how this would work ray
according keyboard

451
00:26:53,809 --> 00:26:57,120
this is good got the keys are back

452
00:26:57,012 --> 00:26:59,671
you can even really think I'd like you
know it's just like home keys are you

453
00:26:59,779 --> 00:27:02,793
just gonna keep your five fingers right
on top of the key is in you

454
00:27:02,919 --> 00:27:06,955
you know you can use the keyboard like
that at it easy to press the correct key

455
00:27:07,279 --> 00:27:10,353
for that good but it's bad because
there's a learning curve right

456
00:27:11,019 --> 00:27:14,830
the first time experience is a mystery

457
00:27:14,083 --> 00:27:17,096
you see this alphabet soup on this thing
that you click piano keyboard how do I

458
00:27:17,096 --> 00:27:17,775
use this

459
00:27:18,639 --> 00:27:23,639
right and so we were thinking that well
the solution would be we have a tutorial

460
00:27:23,639 --> 00:27:27,678
to teach the people how to use the
keyboard not actually in a different you

461
00:27:28,029 --> 00:27:28,082
know not

462
00:27:28,559 --> 00:27:31,564
this is an ideal actually how some
history

463
00:27:32,059 --> 00:27:35,106
may not happen because the original Mac
ship to the tutorial

464
00:27:35,529 --> 00:27:38,614
to teach people how to use the mouse
since that was new

465
00:27:39,379 --> 00:27:43,620
me so we figure we come up with a
tutorial in the interim you think about

466
00:27:43,062 --> 00:27:43,731
it

467
00:27:44,289 --> 00:27:47,820
is that even if we you know we didn't
have that

468
00:27:47,082 --> 00:27:50,097
keyboard keyboard are you mad enough
that we did we could maybe even come up

469
00:27:50,097 --> 00:27:53,656
with a keyboard that use music to try to
keep people that you could baby play

470
00:27:54,529 --> 00:27:55,570
tonight

471
00:27:55,057 --> 00:27:59,026
so it wasn't the craziest idea in the
world but after each of these

472
00:27:59,539 --> 00:28:00,588
experiments

473
00:28:01,029 --> 00:28:04,320
and each the times we thought through
how to make a tutorial

474
00:28:04,032 --> 00:28:08,211
um we always came back to quality

475
00:28:08,499 --> 00:28:11,870
because your already know how to use it

476
00:28:11,087 --> 00:28:15,174
there's no need for tutorial right you
see the thing

477
00:28:16,074 --> 00:28:19,333
and instead of cash what is this

478
00:28:19,999 --> 00:28:23,075
how do I use it right instead of that
you have

479
00:28:23,759 --> 00:28:27,776
this for you look at it and okay it's
different because it's a touch screen

480
00:28:27,929 --> 00:28:28,470
its

481
00:28:28,047 --> 00:28:33,116
small but where weird sir hauling on and
invoking

482
00:28:33,539 --> 00:28:37,480
write a lot have these kind of these
familiar notions that your already

483
00:28:37,048 --> 00:28:41,747
very very comfortable with so that's why
we wound up with this

484
00:28:42,179 --> 00:28:45,228
and ultimately right we have this today

485
00:28:45,669 --> 00:28:48,716
because we thought was more important

486
00:28:49,139 --> 00:28:52,350
that the work should explain itself
right

487
00:28:52,035 --> 00:28:55,069
you already know how to use quality we
don't need a tutorial

488
00:28:55,069 --> 00:28:58,147
and so this was really the key to that
decision

489
00:28:59,047 --> 00:29:03,076
key a schizophrenic funny word but it's
the here are the key keyboard

490
00:29:03,499 --> 00:29:07,501
but the I'm it's the real reason why we
chose to go with QWERTY

491
00:29:07,519 --> 00:29:13,545
because again we thought that work
explain itself

492
00:29:13,779 --> 00:29:17,340
so the next story number five every word

493
00:29:17,034 --> 00:29:21,093
on every keystroke and to the left is
choose the simplest

494
00:29:21,399 --> 00:29:24,405
thing which might work

495
00:29:24,999 --> 00:29:29,090
to continuing on the development of the
iPhone keyboard

496
00:29:29,009 --> 00:29:32,045
became my job to do the work iniquity
possible

497
00:29:32,045 --> 00:29:35,049
right we had this challenge have
smokey's

498
00:29:35,085 --> 00:29:38,264
difficult to type now into course you
all

499
00:29:39,029 --> 00:29:43,200
care about my problems you just wanna
type quickly

500
00:29:43,002 --> 00:29:46,481
right so the software was gonna need to
help

501
00:29:46,679 --> 00:29:50,360
and so the eventual solution for that
was so auto correction

502
00:29:50,036 --> 00:29:54,475
which Chris what you meant and not what
you did

503
00:29:54,799 --> 00:29:58,845
taking what you didn't figure that what
you meant from it

504
00:29:59,259 --> 00:30:02,580
and of course there's I you know quite a
bit have I

505
00:30:02,058 --> 00:30:05,257
your smart people have thought about the
I

506
00:30:05,779 --> 00:30:08,820
you know text processing and in a text

507
00:30:08,082 --> 00:30:11,401
you note extradition prediction sure
your summer the

508
00:30:12,139 --> 00:30:16,142
to terms that are you know in the state
of the art you know then as now

509
00:30:16,169 --> 00:30:19,960
basin statistics and stochastic
programming conditional random field

510
00:30:19,096 --> 00:30:23,935
know a lot of these involve large
statistical models

511
00:30:24,799 --> 00:30:27,862
to make them work that they rely on
language models

512
00:30:28,429 --> 00:30:31,690
and these were not going to be possible
given to

513
00:30:31,069 --> 00:30:34,072
you note the hardware and software
environment to the original iPhone

514
00:30:34,099 --> 00:30:39,187
not only that I less studied math in the
11th grade

515
00:30:40,087 --> 00:30:43,136
which was in the eighties and it was in
the early eighties

516
00:30:43,919 --> 00:30:48,016
so I'm him so I screws can I was looking
at you know so generous to Catholic

517
00:30:48,889 --> 00:30:50,914
programming okay well you know so

518
00:30:51,139 --> 00:30:54,151
really worried if I was I needed
something simple because

519
00:30:54,259 --> 00:31:01,259
math is hard right

520
00:31:01,035 --> 00:31:05,036
and sell here

521
00:31:05,036 --> 00:31:08,123
it's a little bit have an idea little
little better how how the or a

522
00:31:09,023 --> 00:31:09,099
correction

523
00:31:09,099 --> 00:31:12,258
works right certainly in the you know

524
00:31:13,149 --> 00:31:16,950
who you know way back then in the first
time that I try to get this

525
00:31:16,095 --> 00:31:19,146
get this idea together so you typing a
letter RI well

526
00:31:20,046 --> 00:31:24,365
since the keys are small maybe you meant
those keys

527
00:31:24,779 --> 00:31:27,788
as well so I was highlighted but maybe
you missed

528
00:31:28,679 --> 00:31:32,330
in maybe you meant one of the keys
around RI okay

529
00:31:32,033 --> 00:31:36,072
and then you type the key and walked
similarly maybe you meant one of those

530
00:31:36,072 --> 00:31:39,083
letters around it so even just with two
characters

531
00:31:39,083 --> 00:31:43,107
you can see that there are a number of
possibilities

532
00:31:44,007 --> 00:31:48,009
right and so in it one column are words
that are real words inner

533
00:31:48,009 --> 00:31:52,013
going to be in English Dictionary arm
and the other column there are just

534
00:31:52,049 --> 00:31:52,146
those

535
00:31:53,046 --> 00:31:57,245
words that are in there there letter
combinations but they're not words

536
00:31:57,659 --> 00:32:01,100
you know we have words that I'm at the
one column has

537
00:32:01,001 --> 00:32:04,001
words data in the dictionary and the

538
00:32:04,001 --> 00:32:09,076
other column has just letter
combinations that aren't words

539
00:32:09,085 --> 00:32:12,170
as we do with this is now raw material
right and so

540
00:32:13,007 --> 00:32:19,426
or a correction really ultimately view
too big part of it is just speculation

541
00:32:20,119 --> 00:32:22,162
did you mean this one did you mean that
when did you mean the other one did you

542
00:32:22,549 --> 00:32:23,830
mean this one did you mean this one

543
00:32:23,083 --> 00:32:26,085
and this involved in many many many
dictionary lookups

544
00:32:26,085 --> 00:32:30,164
to just check are these water
combinations are these words in the

545
00:32:30,929 --> 00:32:31,026
dictionary that

546
00:32:31,899 --> 00:32:35,903
is in a selected by the by the user so I
needed something that work really really

547
00:32:36,299 --> 00:32:36,980
fast

548
00:32:36,098 --> 00:32:39,136
and a you know so that the idea was was

549
00:32:40,036 --> 00:32:44,055
offer to me that well you know you could
maybe in as user is typing just kinda

550
00:32:44,379 --> 00:32:47,190
send that dictionary lookup stuff on a
background thread

551
00:32:47,019 --> 00:32:50,608
an you know sorry post the results back
when it's done in Crosby been back then

552
00:32:50,779 --> 00:32:51,805
we didn't have GPD

553
00:32:52,039 --> 00:32:55,470
and so it you know DP threats would have
been the solution and it's like I'm

554
00:32:55,047 --> 00:32:57,366
scared a multi-threaded coat it's hard

555
00:32:57,789 --> 00:33:01,090
it's hard to get right at a secret to
say she's like

556
00:33:01,009 --> 00:33:04,057
there's actually I did have one little
bit of multi-threaded code

557
00:33:04,057 --> 00:33:07,134
in the keyboard in the beginning to do
some initialization and that was as

558
00:33:08,034 --> 00:33:11,099
largest source a bug's a tiny little
piece of code

559
00:33:11,099 --> 00:33:14,144
and that was his largest source a bug
dealing with app startup in

560
00:33:15,044 --> 00:33:20,136
this is launching happenin suspended
Williams said huge source a box

561
00:33:21,036 --> 00:33:24,335
even if just a little bit of code and I
want to keep multi-threading

562
00:33:24,659 --> 00:33:27,950
out %uh or a correction which was much
more complex right

563
00:33:27,095 --> 00:33:33,284
because I wanted to keep thing simple
because I can understand simple

564
00:33:34,139 --> 00:33:38,139
multi-threading is hard I can even like
making a good breakfast in the morning

565
00:33:38,139 --> 00:33:39,154
right to make the egg

566
00:33:39,289 --> 00:33:43,387
make coffee then make the a keeper
single-threaded rate

567
00:33:44,269 --> 00:33:49,274
otherwise it went up with the message
kitchen okay to keep it simple

568
00:33:49,769 --> 00:33:52,818
so here actually a little algorithm for
searching a dictionary

569
00:33:53,259 --> 00:33:56,301
word list on one side

570
00:33:56,679 --> 00:34:00,680
and a little algorithm on the other side

571
00:34:00,779 --> 00:34:05,240
and this is actually thought up by the
same fellow who did the caged emailed Mr

572
00:34:05,024 --> 00:34:09,030
and so he came up with a good idea and
as part of some of our other experiments

573
00:34:09,084 --> 00:34:09,150
and i'd

574
00:34:10,005 --> 00:34:14,053
why I I saw that good idea and I used it
to what you do you have a

575
00:34:14,098 --> 00:34:17,104
Lister words in the dictionary a memory
map it eatery

576
00:34:18,004 --> 00:34:22,343
the list wants to find the line endings
you store those addresses in a pointer

577
00:34:22,379 --> 00:34:22,384
array

578
00:34:22,879 --> 00:34:25,887
you can then binary search in you can
find whether word is in the dictionary

579
00:34:26,679 --> 00:34:30,400
really really fast really really fast

580
00:34:30,004 --> 00:34:33,052
and so next after that I wrote a custom
c plus plus train crash that

581
00:34:33,088 --> 00:34:37,140
could wrap any string in that memory
mapped area

582
00:34:38,004 --> 00:34:40,078
are each of those strings are not
terminated in the end so you could just

583
00:34:41,014 --> 00:34:42,036
kinda going to wrap

584
00:34:42,036 --> 00:34:45,095
and an address and you have a string as
long as realign start

585
00:34:45,095 --> 00:34:48,190
I and so this turned out to be really
really fast like a check dictionary

586
00:34:49,009 --> 00:34:51,031
membership in microseconds

587
00:34:52,012 --> 00:34:55,020
even on the original iPhone hardware

588
00:34:55,002 --> 00:34:58,016
so this was when I really really fast
and to what this

589
00:34:58,034 --> 00:35:01,063
did was it made speculation really
really

590
00:35:01,063 --> 00:35:04,101
cheap and that in turn made auto
correction

591
00:35:05,001 --> 00:35:08,074
possible to go from

592
00:35:08,074 --> 00:35:12,135
this and take all love those
permutations even for as long word as

593
00:35:13,035 --> 00:35:14,094
you might find

594
00:35:14,094 --> 00:35:18,150
in a language and actually go and figure
out

595
00:35:19,005 --> 00:35:22,006
whether that word is going to be in the
dictionary or not

596
00:35:22,051 --> 00:35:26,104
and whether similar options would be as
well mister to be so fast

597
00:35:27,004 --> 00:35:32,010
I stored no state for the current word I
didn't store previous results all I did

598
00:35:32,001 --> 00:35:32,029
was

599
00:35:32,038 --> 00:35:35,129
at restore the string for the users
piping if we think think think back here

600
00:35:36,029 --> 00:35:39,061
all I've had stories I T what if you
typed

601
00:35:39,061 --> 00:35:43,125
the actual characters that were lit up
right not the neighboring ones

602
00:35:44,025 --> 00:35:47,071
only thing stored was the characters
that were lit up

603
00:35:47,071 --> 00:35:50,098
K only such store that string so it
actually

604
00:35:50,098 --> 00:35:53,132
career in the entire algorithm on every
keystroke

605
00:35:54,032 --> 00:35:58,109
from scratch which means of course that
search the whole dictionary seventy

606
00:35:59,009 --> 00:36:00,012
thousand plus words

607
00:36:00,039 --> 00:36:04,072
on every keystroke because to me

608
00:36:04,072 --> 00:36:07,077
this was the simplest thing as I was
thinking about it was the simplest thing

609
00:36:08,022 --> 00:36:09,120
which might work

610
00:36:10,002 --> 00:36:14,002
and by coming up with sufficiently fast
implementations

611
00:36:14,002 --> 00:36:17,005
up this it turned out that with the
simplest thing

612
00:36:17,005 --> 00:36:20,011
didn't work simple is

613
00:36:21,001 --> 00:36:24,044
often speedy and

614
00:36:24,053 --> 00:36:27,075
Voice simple something I could
understand and when they're about record

615
00:36:27,075 --> 00:36:33,092
understand where those bugs might have
been

616
00:36:33,092 --> 00:36:37,124
and everywhere and every keystroke
choose the simplest thing which might

617
00:36:38,024 --> 00:36:42,053
work and again if you're kinda looking
from the beginning a project trying to

618
00:36:42,053 --> 00:36:42,122
think

619
00:36:43,022 --> 00:36:46,098
to the endeavor project and you know
what you might have when you're all done

620
00:36:46,098 --> 00:36:49,109
at White term this way choose the
simplest thing which might work

621
00:36:50,009 --> 00:36:52,097
if you did actually have a crystal ball
you could tell whether maybe your idea

622
00:36:52,097 --> 00:36:52,193
would work

623
00:36:53,093 --> 00:36:56,122
I think the even from the beginning cunt
finger

624
00:36:57,022 --> 00:37:00,115
the simplest thing which could possibly
just give you what you need

625
00:37:01,015 --> 00:37:07,106
go with that first not a good idea

626
00:37:08,006 --> 00:37:12,072
and so the next story number six

627
00:37:12,072 --> 00:37:15,087
is we only need one of these right and
the lesson is only

628
00:37:15,087 --> 00:37:19,091
show your best work

629
00:37:20,027 --> 00:37:23,103
and so fast for it if 2009 I joined

630
00:37:24,003 --> 00:37:28,021
efforts to develop the iPad and at the
iPad Mini Keyboard

631
00:37:28,021 --> 00:37:32,023
and authors by this point I had to sign
quote pasted onto my back want to kick

632
00:37:32,041 --> 00:37:32,103
me

633
00:37:33,003 --> 00:37:37,006
and the other one said because keyboards
so I it was to

634
00:37:37,006 --> 00:37:40,041
a here sir if something that seemed to
be

635
00:37:40,041 --> 00:37:45,050
in my my area and so I noticed that in
landscape returned an iPad 2 landscape

636
00:37:46,031 --> 00:37:46,117
that

637
00:37:47,017 --> 00:37:50,116
the a number of key said you could fit
across the top was close to the size of

638
00:37:51,016 --> 00:37:52,102
a full-size desktop keyboard

639
00:37:53,002 --> 00:37:58,043
and so here it is a representation have
a full-size desktop keyboard

640
00:37:58,043 --> 00:38:01,079
and their is the size of the keys that
we could fit

641
00:38:01,079 --> 00:38:06,084
on an iPad ABC for overlay one on top in
the other

642
00:38:06,084 --> 00:38:10,153
it's pretty close not exact the iPad is
smaller

643
00:38:11,053 --> 00:38:14,120
but it's close and so I'm thinking that
it's okay but you know it's not the

644
00:38:15,002 --> 00:38:16,015
exactly the same size but

645
00:38:16,033 --> 00:38:19,081
auto correction could maybe help you to
type comfortably

646
00:38:19,081 --> 00:38:23,177
right from here big keys across the park
that seem

647
00:38:24,077 --> 00:38:27,103
like a good idea but then we also had to
quit

648
00:38:28,003 --> 00:38:31,078
you know the the question in our minds
what if you wanted more keys we now have

649
00:38:31,078 --> 00:38:32,116
space for them

650
00:38:33,016 --> 00:38:36,092
allocating the space in the original
iPhone keyboard was the place tortuous

651
00:38:36,092 --> 00:38:36,187
because

652
00:38:37,087 --> 00:38:40,169
you know we couldn't even fit you know
that period character in the first

653
00:38:41,069 --> 00:38:44,117
you know in the first you know keyboard
that that you saw we had to put that on

654
00:38:45,017 --> 00:38:46,074
to a secondary key point

655
00:38:46,074 --> 00:38:49,165
and so we didn't have that problem that
we could have more keys we had the room

656
00:38:50,065 --> 00:38:50,071
for them

657
00:38:51,025 --> 00:38:54,064
and so you can imagine that we can get a
start getting some more key issue in

658
00:38:54,064 --> 00:38:56,133
this case I put in shifting caps lock in
10 happen

659
00:38:57,033 --> 00:39:00,091
couple character characters of the site
and then taking a direct your attention

660
00:39:00,091 --> 00:39:03,094
to this year so we would have a switch

661
00:39:03,094 --> 00:39:06,160
that would allow you to go very very
quickly between

662
00:39:07,006 --> 00:39:10,045
come

663
00:39:10,099 --> 00:39:13,168
the one with the keys and in the one

664
00:39:14,068 --> 00:39:18,166
with the more keys and tourists are
similar to what we have where you change

665
00:39:19,066 --> 00:39:19,120
to

666
00:39:20,002 --> 00:39:24,014
numbers and punctuation but this was
really keeping the letters the same

667
00:39:24,032 --> 00:39:27,086
and he's bringing in more keys and so

668
00:39:27,086 --> 00:39:30,089
you know a bunch of us who are working
on this we thought this was

669
00:39:30,089 --> 00:39:34,112
the best of both worlds score

670
00:39:35,012 --> 00:39:40,095
so hi demo this first the

671
00:39:40,095 --> 00:39:43,164
which is as a gulp and so I am

672
00:39:44,064 --> 00:39:47,122
so bringing the iPad you know you can
have the demo reel 'em

673
00:39:48,022 --> 00:39:51,104
and go in there and then there to keep
him there and then you could OKC

674
00:39:52,004 --> 00:39:56,072
put the iPad in front of him and I and
fortunately the work

675
00:39:56,072 --> 00:39:59,079
explained itself to the description was
a short at

676
00:40:00,042 --> 00:40:03,135
we're looking at iPad keyboard options
today here two options press the switch

677
00:40:04,035 --> 00:40:04,131
key

678
00:40:05,031 --> 00:40:08,043
done so he did

679
00:40:08,043 --> 00:40:12,126
he looked at the first keyboard
depressed switch key

680
00:40:13,026 --> 00:40:16,084
the animation he looked at the second
keyboard

681
00:40:16,084 --> 00:40:20,133
the press to switch Keegan he saw the
first one come back

682
00:40:21,033 --> 00:40:26,066
a third time presses switch key fees the
second one come in

683
00:40:26,066 --> 00:40:30,074
and he looks at this for a moment and he
looks up at me and that's really what

684
00:40:30,074 --> 00:40:34,133
are these right

685
00:40:35,033 --> 00:40:41,084
and he's looking at me I

686
00:40:41,084 --> 00:40:44,172
and so I got through a couple times and
so I and he says well which one do you

687
00:40:45,072 --> 00:40:46,076
think

688
00:40:47,012 --> 00:40:50,086
we should use and I said well

689
00:40:50,086 --> 00:40:53,170
you know I think we should use the one
with the bigger keys

690
00:40:54,007 --> 00:40:57,060
because it's just easier to type it
close to the size full-size keyboard

691
00:40:58,023 --> 00:41:05,023
he said okay we'll go with that one

692
00:41:10,028 --> 00:41:13,105
and that was it really a record and
really really was it was it was quick is

693
00:41:14,005 --> 00:41:15,017
that it so that's why we

694
00:41:15,017 --> 00:41:18,050
went up with this and then ultimately
today this

695
00:41:18,005 --> 00:41:22,012
and course the you know the lesson is
only sure your best work

696
00:41:22,057 --> 00:41:26,059
perhaps more than anyone else Steve
understood this

697
00:41:26,077 --> 00:41:29,119
go out and think I've all the ideas that
you can

698
00:41:30,019 --> 00:41:34,115
and get now we just pick up the ideas
but get them and make them and use them

699
00:41:35,015 --> 00:41:36,044
in live with them

700
00:41:36,044 --> 00:41:42,055
and think about them and then try to
figure out which is the best one

701
00:41:42,055 --> 00:41:45,132
so it was interesting to me about that
story is that even though I was under a

702
00:41:46,032 --> 00:41:47,107
lot of pressure

703
00:41:48,007 --> 00:41:52,041
when he asked me which 1i thought was
better I knew

704
00:41:52,041 --> 00:41:55,134
but they didn't have the courage to just
go in there and show him

705
00:41:56,034 --> 00:41:59,095
the one that I thought was really the
best the first time

706
00:41:59,095 --> 00:42:02,157
it needed his extra product that little
bit have

707
00:42:03,057 --> 00:42:06,074
extra investigation libitum extra poking
per se

708
00:42:06,074 --> 00:42:09,130
we women it right which is the best one

709
00:42:10,003 --> 00:42:14,003
which is the second best one with not
show the second best one

710
00:42:14,003 --> 00:42:17,011
right only show your best work

711
00:42:17,038 --> 00:42:20,117
you have to go through all the effort

712
00:42:21,017 --> 00:42:28,017
Chenery lot of ideas to come up with one
that really worth it

713
00:42:28,071 --> 00:42:31,085
and that we only need one of these right
homie show

714
00:42:31,085 --> 00:42:36,110
your best work

715
00:42:37,001 --> 00:42:41,065
okay story number seven is let's tried 2
percent darker

716
00:42:41,074 --> 00:42:44,150
and the lesson is iterating quickly lead
to better work

717
00:42:45,005 --> 00:42:48,009
and so 2012 I

718
00:42:48,054 --> 00:42:52,070
jointed effort to develop Iowa 7 and
course as you know

719
00:42:52,007 --> 00:42:55,034
that one of the big goals for Iowa 7 I

720
00:42:55,097 --> 00:42:59,176
was a whole new look and you have to
question how do you make

721
00:43:00,076 --> 00:43:03,090
all of these choices you know in a lot
of times you know coming up with these

722
00:43:03,009 --> 00:43:04,041
new design ideas

723
00:43:05,022 --> 00:43:08,111
it felt like a game of whack-a-mole
because every decision

724
00:43:09,011 --> 00:43:13,042
affect every other decision that you
make you change one thing

725
00:43:13,042 --> 00:43:16,067
and you had the things over there the
kinda family well with the thing the way

726
00:43:16,067 --> 00:43:19,122
it was before and now it doesn't seem
like that now it seems that the thing

727
00:43:20,022 --> 00:43:25,110
that you just changed dozen really work
well with everything else

728
00:43:26,001 --> 00:43:29,096
and so this required a tremendous amana
tuning

729
00:43:30,005 --> 00:43:34,048
iterating iterating iterating you know
version after version

730
00:43:34,048 --> 00:43:38,070
effort after effort try to get things so
that they all look

731
00:43:38,007 --> 00:43:43,040
like they belong together and so if you
can't cut a break that down here from

732
00:43:44,003 --> 00:43:48,071
you know the process of actually trying
to do that work one design iteration

733
00:43:48,071 --> 00:43:52,089
most the time looks like this K

734
00:43:52,089 --> 00:43:55,147
guy like me will bring a demo to the
designer

735
00:43:56,047 --> 00:44:00,053
the designer will have feedback a Pik
notes

736
00:44:00,053 --> 00:44:03,130
how then go away to make those changes

737
00:44:04,003 --> 00:44:09,090
that the designers requested and we
schedule the next demo

738
00:44:10,017 --> 00:44:14,035
we have that next time we go back to
beginning

739
00:44:14,035 --> 00:44:17,132
fell this process can take hours are
baby

740
00:44:18,032 --> 00:44:22,074
because people are busy right and
sometimes you know it's just it's just

741
00:44:22,074 --> 00:44:24,096
hard to get together

742
00:44:24,096 --> 00:44:27,138
and so we wanted a way to go faster

743
00:44:28,038 --> 00:44:33,052
can we go faster and so Am I very very
smart friend to get smart friends

744
00:44:33,052 --> 00:44:36,100
the real key when I might very smartest
friend came up with this idea

745
00:44:37,000 --> 00:44:42,016
upsetting and water settings while
settings are tunable values

746
00:44:42,016 --> 00:44:45,085
are you take all the settings that you
might want to tune

747
00:44:45,085 --> 00:44:49,159
and put them interested object to choose
to submit a backup properties

748
00:44:50,059 --> 00:44:54,059
news Kb 0 so that when you change when
those properties

749
00:44:54,059 --> 00:44:57,130
the the part to the system they're
interested in those values changing

750
00:44:58,003 --> 00:45:03,088
can know about it she also came up this
way save and restore these options

751
00:45:04,015 --> 00:45:07,061
have and then there was also a way that
you have settings which own

752
00:45:07,061 --> 00:45:10,096
settings which on setting so you could
have a tree have settings

753
00:45:10,096 --> 00:45:13,124
and also came up with an editing you I
and if the editing you I wanna show you

754
00:45:14,024 --> 00:45:17,052
so here's just a very simple example
very simple ap

755
00:45:17,052 --> 00:45:21,120
that I let you manipulate you metric
shapes so I've got not one blue circle

756
00:45:22,002 --> 00:45:25,028
and you can go and with the settings you
are I

757
00:45:25,046 --> 00:45:28,144
you can bring up to hide right in the
app

758
00:45:29,044 --> 00:45:33,143
to look up or sliders in change from one
blue triangle 23

759
00:45:34,043 --> 00:45:38,140
red triangles great and then similarly
we can know this other one that actually

760
00:45:39,004 --> 00:45:40,009
support those nested

761
00:45:40,009 --> 00:45:44,014
settings that would use to iPhone style
site to site navigation to show you the

762
00:45:45,004 --> 00:45:45,083
different levels

763
00:45:46,019 --> 00:45:49,028
settings and so if I take the red
triangles and move it back to the

764
00:45:49,028 --> 00:45:50,031
settings for the

765
00:45:50,031 --> 00:45:53,054
the to circle you have this and so what
did this to you

766
00:45:53,054 --> 00:45:57,060
did something very important this
reduced iteration time because once you

767
00:45:57,006 --> 00:45:58,105
had the right settings surfaced

768
00:45:59,059 --> 00:46:02,085
in the apt he gets it with the designer

769
00:46:02,085 --> 00:46:05,094
engineer could sit with the tire in
trying many many

770
00:46:05,094 --> 00:46:08,158
many options and so we got to the point
where

771
00:46:09,058 --> 00:46:11,141
you know you kinda trying to figure out
what control center could look like in

772
00:46:12,041 --> 00:46:12,133
the design a good thing

773
00:46:13,033 --> 00:46:16,066
well what if we meet two percent darker

774
00:46:16,066 --> 00:46:19,131
even know if you can see that up on the
big screen it's the point a bit

775
00:46:20,031 --> 00:46:23,126
is that we were trying to get tuning to
this level

776
00:46:24,026 --> 00:46:28,057
2 percent darker 2 percent darker on
your left

777
00:46:28,057 --> 00:46:31,120
then on your right wanna

778
00:46:32,002 --> 00:46:36,050
that level of polish inattention to
every single value

779
00:46:36,068 --> 00:46:40,082
in the US so when we had a tuning
session like this it was great because

780
00:46:40,082 --> 00:46:43,084
I could do is go back but back to my
desk now copy that

781
00:46:44,002 --> 00:46:47,079
you know that the default that got saved
from the tuning session

782
00:46:47,079 --> 00:46:51,086
copy those to the new for them commit
scrape

783
00:46:51,086 --> 00:46:54,137
and this made cheap and easy to tune we
could just do

784
00:46:55,037 --> 00:46:58,083
over and over and over and over again

785
00:46:58,083 --> 00:47:02,159
this was the way that we went with just
a handful blur styles for Iowa 7 that

786
00:47:03,059 --> 00:47:05,117
are used consistently everywhere

787
00:47:06,017 --> 00:47:09,071
very very difficult to come up with
those with that settler styles

788
00:47:09,071 --> 00:47:12,100
and this is how we did it by just
hearing iterating in making

789
00:47:13,000 --> 00:47:16,072
into reading easy and cheap the lesson
really

790
00:47:16,072 --> 00:47:19,157
rightish iterating quickly leads to
better work I think

791
00:47:20,057 --> 00:47:24,065
my experience from I was seven was there
was a direct correlation

792
00:47:25,037 --> 00:47:29,048
between how easy we made it try out new
things

793
00:47:29,048 --> 00:47:36,048
and what the quality the final work
product wise

794
00:47:37,006 --> 00:47:41,011
that's what strike two percent darker
lesson is iterating quickly

795
00:47:42,001 --> 00:47:48,038
reach the better work

796
00:47:48,047 --> 00:47:51,140
okay three-member 8

797
00:47:52,004 --> 00:47:57,072
three lessons from making this session

798
00:47:58,008 --> 00:48:02,029
and so this is not the original session
I wrote an entirely different set of

799
00:48:02,029 --> 00:48:04,095
lights

800
00:48:04,095 --> 00:48:07,120
and not only that I did three full
rehearsals with them I like to hurt a

801
00:48:08,002 --> 00:48:10,028
lot before I mean it's important that

802
00:48:10,046 --> 00:48:13,049
that this session goes well for you are
y'all came here and if I'm gonna make

803
00:48:13,076 --> 00:48:15,078
sure they give you a good product image
so

804
00:48:15,096 --> 00:48:18,160
put a tremendous man effort into
preparing the fighting rehearsing

805
00:48:19,006 --> 00:48:23,083
arm and so with another set of slides I
used up a big portion of my time

806
00:48:24,037 --> 00:48:27,103
rehearsing that certified I feud good
thing was

807
00:48:28,003 --> 00:48:31,046
done way ahead of time

808
00:48:31,046 --> 00:48:34,089
it was great it felt good to just be
like when preparing for WBC can be

809
00:48:34,089 --> 00:48:36,186
really points just always seems like
everything is down to the wire

810
00:48:37,086 --> 00:48:40,128
down to the wire in this year plan I was
really ahead of time but had one more

811
00:48:41,028 --> 00:48:44,111
review for the content I had in my
flights just go over the material the

812
00:48:45,011 --> 00:48:46,103
content particular because i'm talking
about.

813
00:48:47,003 --> 00:48:51,081
some past Apple work and it turns out
you know that version just got shot n

814
00:48:51,081 --> 00:48:54,147
you can you can show that they're just
issues that could be resolved the

815
00:48:55,047 --> 00:48:56,136
details aren't important

816
00:48:57,036 --> 00:49:03,079
but there's kinda the one big detail
which is I don't have affection anymore

817
00:49:03,079 --> 00:49:06,158
three weeks ago I had something that was
rehearsed and two weeks ago

818
00:49:07,058 --> 00:49:11,083
I didn't know if I was going to be here
today I am

819
00:49:11,083 --> 00:49:14,087
I need to rewrite

820
00:49:14,087 --> 00:49:17,129
I'm and their or three important lessons
they took away

821
00:49:18,029 --> 00:49:22,042
from that and the first was

822
00:49:22,042 --> 00:49:25,101
during this review session where the
person

823
00:49:26,001 --> 00:49:29,013
who I was submitting my flight for
review

824
00:49:29,013 --> 00:49:32,098
to said you can show that

825
00:49:32,098 --> 00:49:36,197
he didn't raise his voice thank you can
show that

826
00:49:37,097 --> 00:49:42,101
here's why many reasons it's very calm

827
00:49:43,001 --> 00:49:47,093
and even after I mean it was clear it
was bad news to me

828
00:49:47,093 --> 00:49:50,099
and even he found me later initially
made sure that I was okay

829
00:49:51,053 --> 00:49:55,090
with this decision that was a very hard
one to cheer with me

830
00:49:55,009 --> 00:49:58,092
and to the Russian from mississippi kind
to people

831
00:49:59,073 --> 00:50:02,164
but be honest about the work

832
00:50:03,064 --> 00:50:06,091
right there's no reason to like be
shouting

833
00:50:06,091 --> 00:50:09,157
into reading people who work try hard
and doing their best

834
00:50:10,057 --> 00:50:13,079
if you have ever found me in a lab or
you know maybe even to

835
00:50:13,079 --> 00:50:16,136
with what little time we have left in
the conference before 10 if you find me

836
00:50:17,036 --> 00:50:18,075
in a lab today

837
00:50:18,075 --> 00:50:22,124
I am I hope I'll do my utmost to be kind
to you

838
00:50:23,024 --> 00:50:26,057
personally because I'm so glad that
you're here it's a pledge contributing

839
00:50:26,057 --> 00:50:27,129
to our platform

840
00:50:28,029 --> 00:50:32,051
and i wanna see you do great work too
but if you show me peace work to be

841
00:50:32,051 --> 00:50:34,087
honest when I think about it

842
00:50:34,087 --> 00:50:37,149
and those things just need to be kept
separate the personal kindness

843
00:50:38,049 --> 00:50:42,088
and honesty about the work

844
00:50:42,088 --> 00:50:45,131
and secondly it's kind of like a you
know through the def did you know

845
00:50:46,031 --> 00:50:52,117
difference I did the same conifer I
didn't peak is feedback personally

846
00:50:53,017 --> 00:50:57,088
in russian history separate yourself
from your work

847
00:50:57,088 --> 00:51:01,111
now in I always invest a tremendous
amount

848
00:51:02,011 --> 00:51:05,063
have care and emotion into my work its

849
00:51:05,063 --> 00:51:08,086
my nature I can help

850
00:51:08,086 --> 00:51:11,117
after I'm done making something

851
00:51:12,017 --> 00:51:15,094
and bring it to the stage the 3d demo

852
00:51:15,094 --> 00:51:18,098
if they just turn it into force myself
from

853
00:51:18,098 --> 00:51:21,124
that thing that I just

854
00:51:22,024 --> 00:51:25,077
sweated over making because

855
00:51:25,077 --> 00:51:28,082
I that's not me that thing that I just
made

856
00:51:29,027 --> 00:51:33,092
right I am ready to accept whatever
feedback anybody had

857
00:51:33,092 --> 00:51:36,155
and ready to go into the next thing

858
00:51:37,055 --> 00:51:40,062
actually need to kind of divorce
yourself when you get feedback

859
00:51:41,025 --> 00:51:45,050
people hopefully are not giving you the
feedback to giving the you the feedback

860
00:51:45,005 --> 00:51:47,011
about the work

861
00:51:47,056 --> 00:51:50,144
separate yourself from the work
regardless of how much time and effort

862
00:51:51,044 --> 00:51:56,099
how much investment you have in it

863
00:51:56,099 --> 00:51:59,163
the third lesson is well obviously I
need to get back to work if I wanted to

864
00:52:00,063 --> 00:52:00,129
have

865
00:52:01,029 --> 00:52:05,121
show right at all I needed to

866
00:52:06,021 --> 00:52:09,104
figure out how to make a set of flights
that would work

867
00:52:10,004 --> 00:52:14,052
and the lesson here is that you never
done for seems like a

868
00:52:14,052 --> 00:52:17,067
you know some you know some sort if you
know nightmare you know if you think

869
00:52:17,067 --> 00:52:18,068
about it one way

870
00:52:18,068 --> 00:52:22,083
you know that you're never done really
but you think about it no one ever done

871
00:52:22,083 --> 00:52:25,154
we all loved the first version

872
00:52:26,054 --> 00:52:30,075
have the iPhone when we shipped it and
then we went right back in we made

873
00:52:30,075 --> 00:52:34,092
version to an earlier this week we
showed I always stayed in Yosemite and a

874
00:52:34,092 --> 00:52:36,096
whole bunch of other great things

875
00:52:36,096 --> 00:52:39,112
you are never done if you're committed
to

876
00:52:40,012 --> 00:52:43,016
you working in this industry working in
high-tech

877
00:52:43,016 --> 00:52:46,069
are you are never done the world keeps
moving you gotta try to keep moving

878
00:52:46,069 --> 00:52:46,136
weather

879
00:52:47,036 --> 00:52:50,099
to do the best that you can do at any
one time basically you're we should

880
00:52:50,099 --> 00:52:52,166
bring your latest demo if you think
about it that way

881
00:52:53,066 --> 00:52:55,165
that these demos get to be very very
polished polish to the point where

882
00:52:56,065 --> 00:52:57,065
you're gonna actually

883
00:52:57,065 --> 00:53:01,098
actually share it with the rest of the
world me

884
00:53:01,098 --> 00:53:05,112
so still on this this idea this is a
quotation that

885
00:53:06,012 --> 00:53:09,104
actually up on the wall arm

886
00:53:10,004 --> 00:53:13,079
very very large in Cupertino a quote
from Steve jobs

887
00:53:13,079 --> 00:53:16,128
and he said if you do something it
turned out pretty good

888
00:53:17,028 --> 00:53:20,040
then you should go do something else
wonderful not dwell on it too long

889
00:53:20,004 --> 00:53:23,012
just figure out watch next

890
00:53:23,048 --> 00:53:27,132
I think this is true only when you do
something that turns out well

891
00:53:28,032 --> 00:53:31,125
but it even has been case for me for the
session I did for me it didn't turn out

892
00:53:32,025 --> 00:53:32,116
so well

893
00:53:33,016 --> 00:53:36,016
I needed to figure out what was next

894
00:53:36,016 --> 00:53:39,043
give up I could try again and I tried
again

895
00:53:39,043 --> 00:53:42,045
I'm very glad that I did a very happy to
be here today and so

896
00:53:42,063 --> 00:53:45,146
the year kind of the Big Sur have joined
up

897
00:53:46,046 --> 00:53:50,144
lesson from the section is that I
stories and lessons never and I thought

898
00:53:51,044 --> 00:53:54,142
this was going to be a central just
about stories in washington had no idea

899
00:53:55,042 --> 00:53:58,128
that I'd started a process that would
teach me more lessons

900
00:53:59,028 --> 00:54:02,064
and so those are my

901
00:54:02,064 --> 00:54:06,143
stories and there they are

902
00:54:07,043 --> 00:54:10,076
again and I think

903
00:54:10,076 --> 00:54:15,149
you know I hope at least that you found
them entertaining

904
00:54:16,049 --> 00:54:19,116
but really I think the important part is

905
00:54:20,016 --> 00:54:23,043
the the lessons you can take out from
them

906
00:54:23,043 --> 00:54:27,070
because you never gonna find yourself in
exactly the same situations that I did

907
00:54:27,007 --> 00:54:32,038
I but I think that you find yourself in
situations where

908
00:54:33,001 --> 00:54:36,062
you can apply some of these ideas useful
E

909
00:54:36,062 --> 00:54:40,133
and productively work and those are my
story

910
00:54:41,033 --> 00:54:44,045
and so now I ask you heard you

911
00:54:44,045 --> 00:54:48,049
to go out in you make your own stories
tell your own story

912
00:54:48,049 --> 00:54:52,111
extract wasn't out from them right and
then figure out how are you going to use

913
00:54:53,011 --> 00:54:53,075
that

914
00:54:53,075 --> 00:54:57,116
experience to go and do some more great
work up your own

915
00:54:58,016 --> 00:54:58,022
thank you very much
