Talk given by Yegor on the topic of XDSD at the Devternity Conference

devternity_yegor_xdsd_talk_captions.srt

1
00:00:00,030 --> 00:00:02,270
Yegor: A few words about myself just quickly.

2
00:00:02,700 --> 00:00:05,600
I'm writing software for 25 years,

3
00:00:05,600 --> 00:00:07,780
starting in school.

4
00:00:08,080 --> 00:00:10,670
For about 15 years, I was managing a team,

5
00:00:10,670 --> 00:00:14,130
different teams, up to 120 people, sometimes of developers,

6
00:00:14,130 --> 00:00:16,700
in Ukraine mostly.

7
00:00:17,540 --> 00:00:19,800
I'm also a Java developer myself, a project manager,

8
00:00:19,800 --> 00:00:22,560
and an owner of a company [which]

9
00:00:22,560 --> 00:00:24,740
was, you know, managing all these people.

10
00:00:24,740 --> 00:00:27,210
You can see my Twitter account.

11
00:00:27,210 --> 00:00:29,820
You can see my blog. Please subscribe.

12
00:00:29,820 --> 00:00:30,990
That will be interesting, because everything

13
00:00:30,990 --> 00:00:32,970
I'm gonna say today is explained in my

14
00:00:32,970 --> 00:00:34,590
blog and many different articles. So,

15
00:00:34,590 --> 00:00:36,660
I'm gonna present [to] you the way we manage

16
00:00:36,660 --> 00:00:38,730
programmers, the way we manage projects,

17
00:00:38,730 --> 00:00:41,040
and it's quite different from everything

18
00:00:41,040 --> 00:00:42,810
you've seen in many - in all - other places.

19
00:00:42,810 --> 00:00:45,030
So, that's why you definitely will

20
00:00:45,030 --> 00:00:47,190
be interested to see more content from me.

21
00:00:47,190 --> 00:00:50,580
Here's the subject. Here's the problem.

22
00:00:50,580 --> 00:00:52,800
So, in all my projects, for many, many

23
00:00:52,800 --> 00:00:55,770
years I had the same [issue] as these people are

24
00:00:55,770 --> 00:00:58,170
saying about the industry in general. So

25
00:00:58,170 --> 00:00:59,850
all our projects, all the software

26
00:00:59,850 --> 00:01:01,730
development we have, is basically chaos.

27
00:01:01,730 --> 00:01:06,210
So, we fail, we slip the schedule,

28
00:01:06,210 --> 00:01:09,030
we miss time, we don't complete

29
00:01:09,030 --> 00:01:11,160
the requirements, we don't deliver,

30
00:01:11,160 --> 00:01:13,260
we spend money, and customers are not happy.

31
00:01:13,260 --> 00:01:15,990
I've had that for 15 years - in all - in most

32
00:01:15,990 --> 00:01:18,090
of our projects. And, I found

33
00:01:18,090 --> 00:01:20,340
statistics in the report which says

34
00:01:20,340 --> 00:01:22,320
we're in chaos. And, they say that about

35
00:01:22,320 --> 00:01:25,050
like 94 - which is almost a hundred -

36
00:01:25,050 --> 00:01:27,510
percent of projects actually

37
00:01:27,510 --> 00:01:30,270
fail and have to restart. That's what the

38
00:01:30,270 --> 00:01:31,830
report says about the entire industry.

39
00:01:31,830 --> 00:01:34,080
And, they say that only seven percent of

40
00:01:34,080 --> 00:01:36,030
[the] projects failed because of

41
00:01:36,030 --> 00:01:38,940
technical incompetence. So,

42
00:01:38,940 --> 00:01:40,740
the problem is not with programmers.

43
00:01:40,740 --> 00:01:43,200
The problem is with the management.

44
00:01:43,200 --> 00:01:45,030
They fail because we miss requirements,

45
00:01:45,030 --> 00:01:46,500
because we don't understand what to do,

46
00:01:46,500 --> 00:01:48,210
because we cannot control what we do,

47
00:01:48,210 --> 00:01:50,460
because we cannot manage ourselves. It's not because

48
00:01:50,460 --> 00:01:52,110
we're bad programmers, not because we

49
00:01:52,110 --> 00:01:54,240
don't know Java or Ruby or whatever.

50
00:01:54,240 --> 00:01:56,100
It's because we don't manage ourselves the

51
00:01:56,100 --> 00:01:57,990
right way. And, I've had the same [issue], so

52
00:01:57,990 --> 00:01:59,760
that's why I tried to solve the problem,

53
00:01:59,760 --> 00:02:02,240
and invented the methodology which

54
00:02:02,240 --> 00:02:04,650
actually, in my case, in my experience,

55
00:02:04,650 --> 00:02:06,390
in our projects, actually helped us

56
00:02:06,390 --> 00:02:09,150
a lot to to resolve that [issue] and

57
00:02:09,150 --> 00:02:12,210
be successful. There are five things that -

58
00:02:12,210 --> 00:02:13,690
the methodology is quite big, but

59
00:02:13,690 --> 00:02:15,310
there are five things - I want to

60
00:02:15,310 --> 00:02:17,590
highlight today. They're quite

61
00:02:17,590 --> 00:02:18,910
controversial. That's why it's

62
00:02:18,910 --> 00:02:21,850
interesting. So, five things. How we are

63
00:02:21,850 --> 00:02:23,410
different from everything you've seen before:

64
00:02:23,410 --> 00:02:29,170
Thing number one: Traditionally, we

65
00:02:29,170 --> 00:02:31,330
believe, traditionally you believe, or

66
00:02:31,330 --> 00:02:33,790
they believe, that the teamwork is so

67
00:02:33,790 --> 00:02:35,710
important for the success of the project.

68
00:02:35,710 --> 00:02:39,400
We think that it's so important to

69
00:02:39,400 --> 00:02:41,230
keep the team together, to make sure that

70
00:02:41,230 --> 00:02:43,390
people stay together, that they are

71
00:02:43,390 --> 00:02:45,970
motivated to work together. And, the job

72
00:02:45,970 --> 00:02:48,340
of the manager is to make sure there's a

73
00:02:48,340 --> 00:02:50,650
team there, [and to] make sure that people match

74
00:02:50,650 --> 00:02:53,080
to each other. And, this is

75
00:02:53,080 --> 00:02:55,600
the key factor for the

76
00:02:55,600 --> 00:02:57,220
success of this team, for the success of

77
00:02:57,220 --> 00:03:01,480
the project. So, we just spend a

78
00:03:01,480 --> 00:03:04,190
lot of time on on the team building, on team spirit,

79
00:03:04,480 --> 00:03:06,740
on making sure the people love each other.

80
00:03:06,740 --> 00:03:08,940
And, that's how we guarantee the success.

81
00:03:09,790 --> 00:03:12,520
I think that this approach is for lazy managers.

82
00:03:13,180 --> 00:03:14,180
This is for people

83
00:03:14,180 --> 00:03:15,470
who cannot understand how to

84
00:03:15,470 --> 00:03:17,560
manage programmers. This is for people

85
00:03:17,560 --> 00:03:19,660
who are not really managers. This is what

86
00:03:19,660 --> 00:03:21,310
causes troubles in projects. This is what

87
00:03:21,310 --> 00:03:24,400
this is what leads to loss of the

88
00:03:24,400 --> 00:03:26,380
"A players". Do you know what "A players,

89
00:03:26,380 --> 00:03:28,570
B players, and C players" is? This is the

90
00:03:28,570 --> 00:03:30,370
terminology from the sales [world] actually, but,

91
00:03:30,370 --> 00:03:32,230
I think it can be applied to programming

92
00:03:32,230 --> 00:03:33,970
as well. So, in any team, we have

93
00:03:33,970 --> 00:03:35,620
A players, B players, and C players.

94
00:03:35,620 --> 00:03:36,910
Like senior programmers, medium

95
00:03:36,910 --> 00:03:38,590
programmers, and junior programmers.

96
00:03:38,590 --> 00:03:40,510
So A players, they're always - you know - aiming

97
00:03:40,510 --> 00:03:41,740
for excellence. They're always

98
00:03:41,740 --> 00:03:43,960
trying to do better. C players are

99
00:03:43,960 --> 00:03:45,910
the people who we kind of have to remove

100
00:03:45,910 --> 00:03:47,950
from the team. So, this approach with the

101
00:03:47,950 --> 00:03:50,500
teamwork in our experience is really

102
00:03:50,500 --> 00:03:52,450
good for C players and its really good

103
00:03:52,450 --> 00:03:55,300
for lazy managers. In our case, we do the

104
00:03:55,300 --> 00:03:58,690
opposite. We aim for individual

105
00:03:58,690 --> 00:04:00,700
responsibility. Not team responsibility,

106
00:04:00,700 --> 00:04:02,650
but individual responsibility, always. So,

107
00:04:02,650 --> 00:04:04,330
we don't like the idea of the team in

108
00:04:04,330 --> 00:04:06,490
general. We always pay attention to, "Who

109
00:04:06,490 --> 00:04:08,590
is doing what?", "What's the personal scope

110
00:04:08,590 --> 00:04:10,630
of the task?", and we scope that task

111
00:04:10,630 --> 00:04:12,850
individually. So, we always assign tasks

112
00:04:12,850 --> 00:04:14,890
individually to programmers. We always

113
00:04:14,890 --> 00:04:17,260
give tasks to [the programmers.] We always control the

114
00:04:17,260 --> 00:04:19,060
results of each programmer individually.

115
00:04:19,060 --> 00:04:20,710
So - we never - we don't do the

116
00:04:20,710 --> 00:04:23,410
teamwork at all. So, we clearly define the

117
00:04:23,410 --> 00:04:25,600
rules of the work. And this is

118
00:04:25,600 --> 00:04:28,460
a lot of job we do on the management side.

119
00:04:28,460 --> 00:04:30,440
We always have an architect on our team

120
00:04:30,440 --> 00:04:32,360
who is the person who is

121
00:04:32,360 --> 00:04:34,070
responsible - personally responsible - for

122
00:04:34,070 --> 00:04:35,210
the success of the project,

123
00:04:35,210 --> 00:04:37,330
and who is making all technical decisions.

124
00:04:37,520 --> 00:04:39,010
So, we don't have any democracy

125
00:04:39,010 --> 00:04:41,310
in our teams. We don't vote for technical decisions.

126
00:04:41,490 --> 00:04:42,860
It's always the architects who

127
00:04:42,860 --> 00:04:44,720
makes the decisions and always the team -

128
00:04:44,720 --> 00:04:46,610
the players or developers - who are

129
00:04:46,610 --> 00:04:48,590
responsible for their own scope of work -

130
00:04:48,590 --> 00:04:52,100
for their own tasks. We always work

131
00:04:52,100 --> 00:04:54,320
remotely. So, that's probably what is

132
00:04:54,320 --> 00:04:56,030
different from other teams. But, we work

133
00:04:56,030 --> 00:04:57,860
remotely. So, all our programmers - they

134
00:04:57,860 --> 00:04:59,690
stay in their own places, in their

135
00:04:59,690 --> 00:05:02,030
own countries, and we work. We don't have

136
00:05:02,030 --> 00:05:04,520
a central office at all. So, all our teams

137
00:05:04,520 --> 00:05:06,500
work remotely. And, they communicate

138
00:05:06,500 --> 00:05:09,140
only electronically. So, that also

139
00:05:09,140 --> 00:05:10,940
helps us to stay away from this "team idea"

140
00:05:10,940 --> 00:05:13,670
and to focus on an "individual results" idea.

141
00:05:13,670 --> 00:05:17,000
That's [the] first thing. The second one

142
00:05:17,000 --> 00:05:21,530
is over time. Traditionally - (Yegor points at slides) - What's on

143
00:05:21,530 --> 00:05:22,730
the left it's "traditionally". What's on

144
00:05:22,730 --> 00:05:23,780
the right is our approach. So,

145
00:05:23,780 --> 00:05:27,320
traditionally, programmers are managed by

146
00:05:27,320 --> 00:05:30,380
stress, in most cases. So, we just know

147
00:05:30,380 --> 00:05:34,130
that programmers [are], in general, I mean

148
00:05:34,130 --> 00:05:35,900
managers assume that programmers, in

149
00:05:35,900 --> 00:05:37,820
general, are kind of lazy. So, the best way

150
00:05:37,820 --> 00:05:39,560
to manage them is to put the deadlines

151
00:05:39,560 --> 00:05:41,750
as close as possible, and then stress

152
00:05:41,750 --> 00:05:43,760
them so they work overtime. They work on

153
00:05:43,760 --> 00:05:45,380
the weekend, and that's how they will

154
00:05:45,380 --> 00:05:46,910
achieve better results. That's how they

155
00:05:46,910 --> 00:05:48,530
will be more or less under control.

156
00:05:48,530 --> 00:05:50,480
That's the only way bad managers can

157
00:05:50,480 --> 00:05:51,830
control people - [it] is just to keep them

158
00:05:51,830 --> 00:05:54,020
under stress. And that's how programmers

159
00:05:54,020 --> 00:05:56,870
will do something. So we give them

160
00:05:56,870 --> 00:05:58,940
deadlines, we give them stress, and we get

161
00:05:58,940 --> 00:06:01,010
overtime. So they work more, they spend

162
00:06:01,010 --> 00:06:02,570
more, and we believe, and - managers in

163
00:06:02,570 --> 00:06:04,310
general assume that if people stay

164
00:06:04,310 --> 00:06:05,960
overtime, if they stay in the office,

165
00:06:05,960 --> 00:06:08,030
after the office is closed, these are the

166
00:06:08,030 --> 00:06:09,230
"good programmers". These are the good

167
00:06:09,230 --> 00:06:11,150
developers who stay over time to spend

168
00:06:11,150 --> 00:06:13,910
more time in the office. In our case, it's

169
00:06:13,910 --> 00:06:16,580
quite the opposite. So, we don't want

170
00:06:16,580 --> 00:06:18,560
programmers to work more. We want to work

171
00:06:18,560 --> 00:06:21,710
them less. We expect them to develop and

172
00:06:21,710 --> 00:06:23,540
to commit their code, and to close their

173
00:06:23,540 --> 00:06:25,490
tasks, and to close their tickets, and to

174
00:06:25,490 --> 00:06:27,710
do anything voluntarily - only when they

175
00:06:27,710 --> 00:06:29,990
want to do that. So, we allow our

176
00:06:29,990 --> 00:06:31,970
programmers to reject tasks if they

177
00:06:31,970 --> 00:06:33,770
don't want to do them. If they don't feel

178
00:06:33,770 --> 00:06:36,200
like it. So, we just give them tasks and

179
00:06:36,200 --> 00:06:37,730
allow them to choose whether it's a good

180
00:06:37,730 --> 00:06:39,770
time for them to work on them or not.

181
00:06:39,770 --> 00:06:41,600
Whether it's a good task for them or not. And,

182
00:06:41,600 --> 00:06:44,030
whether it's a good time for them to do

183
00:06:44,030 --> 00:06:46,310
it now. And, in most cases, most projects,

184
00:06:46,310 --> 00:06:48,020
our programmers, when the project is full

185
00:06:48,020 --> 00:06:50,090
of work - and, we have a lot of tasks - they

186
00:06:50,090 --> 00:06:53,660
spend three, four hours a day for the

187
00:06:53,660 --> 00:06:55,310
development, not more. So, we never work

188
00:06:55,310 --> 00:06:57,140
eight hours a day. Forget the

189
00:06:57,140 --> 00:06:58,730
overtime. We don't have overtime at all.

190
00:06:58,730 --> 00:07:01,130
So three, four hours a day. It's - in my

191
00:07:01,130 --> 00:07:03,200
experience - it's the maximum time a good

192
00:07:03,200 --> 00:07:05,750
programmer can spend a day for actually

193
00:07:05,750 --> 00:07:07,520
productive work. And they do it

194
00:07:07,520 --> 00:07:09,950
voluntarily because, I'll say about it

195
00:07:09,950 --> 00:07:11,510
later but, because we pay for results,

196
00:07:11,510 --> 00:07:13,730
that's why they're really interested to

197
00:07:13,730 --> 00:07:15,410
do the work, because, they get - you know -

198
00:07:15,410 --> 00:07:17,750
they get money for that. So, they do

199
00:07:17,750 --> 00:07:19,940
it when they feel like it. Sometimes they

200
00:07:19,940 --> 00:07:22,010
commit a lot of time one week and the

201
00:07:22,010 --> 00:07:24,200
next week they commit less time. So, the

202
00:07:24,200 --> 00:07:26,420
graph jumps up and down. And, it's not

203
00:07:26,420 --> 00:07:27,800
under control of the manager. The manager

204
00:07:27,800 --> 00:07:30,500
doesn't push the person - any developer - to

205
00:07:30,500 --> 00:07:32,960
any amount of work. If we don't

206
00:07:32,960 --> 00:07:34,940
have enough - you know - resources, if we

207
00:07:34,940 --> 00:07:37,250
don't have enough people to close all

208
00:07:37,250 --> 00:07:40,010
our tasks, then our managers have to find

209
00:07:40,010 --> 00:07:42,200
more people. Then we have to find more

210
00:07:42,200 --> 00:07:43,670
resources - I mean developers, programmers,

211
00:07:43,670 --> 00:07:46,010
resources, whatever. So, we find more of

212
00:07:46,010 --> 00:07:47,780
these resources, and it's a job of a

213
00:07:47,780 --> 00:07:49,940
manager how to - you know - to find more

214
00:07:49,940 --> 00:07:51,590
more resources and give that

215
00:07:51,590 --> 00:07:54,080
tasks to that people. But, we never stress

216
00:07:54,080 --> 00:07:55,820
programmers. We never expect them to work

217
00:07:55,820 --> 00:08:00,020
more than they want to. So, it's the job

218
00:08:00,020 --> 00:08:01,610
of a manager. That's what I'm saying. So,

219
00:08:01,610 --> 00:08:03,710
our managers are way more busy than

220
00:08:03,710 --> 00:08:05,570
traditional managers, because they really

221
00:08:05,570 --> 00:08:07,160
have to do the management work. They

222
00:08:07,160 --> 00:08:09,080
don't need to stress anybody. They need

223
00:08:09,080 --> 00:08:12,560
to schedule tasks on a timeline. So, we

224
00:08:12,560 --> 00:08:15,050
don't have ever overtimes at all. Number

225
00:08:15,050 --> 00:08:18,860
three is "Learnings." I'm going from this

226
00:08:18,860 --> 00:08:20,990
from the [least] critical to the

227
00:08:20,990 --> 00:08:22,430
more and more critical. So, "Learning".

228
00:08:22,430 --> 00:08:25,070
Traditionally, in any project which I've

229
00:08:25,070 --> 00:08:30,370
seen before, people, programmers, the core,

230
00:08:30,370 --> 00:08:34,010
are the main holder of information and

231
00:08:34,010 --> 00:08:35,330
knowledge. So, this is the most

232
00:08:35,330 --> 00:08:36,980
knowledgeable part of any project. This

233
00:08:36,980 --> 00:08:38,480
is the most important part of any

234
00:08:38,480 --> 00:08:41,450
project. If programmers know enough about

235
00:08:41,450 --> 00:08:43,910
the stuff they develop, then the project

236
00:08:43,910 --> 00:08:45,800
is more or less successful. The more they

237
00:08:45,800 --> 00:08:47,870
know, the more knowledgeable they are, the

238
00:08:47,870 --> 00:08:51,050
higher the chances of success. And, if

239
00:08:51,050 --> 00:08:53,090
somebody - someone new - [a]

240
00:08:53,090 --> 00:08:55,460
new programmer joins the project, then it

241
00:08:55,460 --> 00:08:57,620
takes some time to ramp up that person,

242
00:08:57,620 --> 00:09:01,250
to - you know - to teach that person all

243
00:09:01,250 --> 00:09:02,540
the new technologies and everything

244
00:09:02,540 --> 00:09:04,100
which is in the project. And, it takes a

245
00:09:04,100 --> 00:09:05,690
month. It takes two months. And, it takes

246
00:09:05,690 --> 00:09:07,880
sometimes three months. So, it takes long

247
00:09:07,880 --> 00:09:10,040
enough to bring somebody new to the

248
00:09:10,040 --> 00:09:11,420
project. This is the traditional way. That's

249
00:09:11,420 --> 00:09:14,840
what we have in most projects. And, in

250
00:09:14,840 --> 00:09:16,880
most cases, we have so-called experts -

251
00:09:16,880 --> 00:09:18,590
subject-matter experts - so, people who

252
00:09:18,590 --> 00:09:22,340
know most of the - they possess the

253
00:09:22,340 --> 00:09:24,050
majority of the knowledge in the project.

254
00:09:24,050 --> 00:09:26,210
So, they know a lot. And, these people are

255
00:09:26,210 --> 00:09:28,880
the key decision-makers in the project.

256
00:09:28,880 --> 00:09:31,130
If we lose that person in a traditional

257
00:09:31,130 --> 00:09:33,710
environment, if we lose that expert, we

258
00:09:33,710 --> 00:09:35,810
lose a lot. And that's why [these] people

259
00:09:35,810 --> 00:09:38,690
kind of start to control the owners of

260
00:09:38,690 --> 00:09:40,730
the project. It's not that the manager

261
00:09:40,730 --> 00:09:43,010
who controls the the development

262
00:09:43,010 --> 00:09:44,390
environment. It's the people, the experts,

263
00:09:44,390 --> 00:09:46,940
who control people with the money. So,

264
00:09:46,940 --> 00:09:48,530
they tell the customers what's going to

265
00:09:48,530 --> 00:09:50,330
happen, not the customer says what's

266
00:09:50,330 --> 00:09:51,530
gonna happen, but the people who are

267
00:09:51,530 --> 00:09:54,380
experts. They start to manipulate people

268
00:09:54,380 --> 00:09:56,810
with money. This is what I've

269
00:09:56,810 --> 00:09:58,640
seen in a lot of places. We change that

270
00:09:58,640 --> 00:10:02,660
completely. In our case, we don't really

271
00:10:02,660 --> 00:10:05,990
like the idea of having people who know

272
00:10:05,990 --> 00:10:10,310
too much in the project. We want

273
00:10:10,310 --> 00:10:14,420
our artifacts - our documents - to know

274
00:10:14,420 --> 00:10:18,350
"too much". We want our product to be self-

275
00:10:18,350 --> 00:10:21,320
explainable so that anybody tomorrow

276
00:10:21,320 --> 00:10:23,240
joins the project and can

277
00:10:23,240 --> 00:10:25,430
understand everything in just a few

278
00:10:25,430 --> 00:10:29,090
hours. It may sound crazy, but it really

279
00:10:29,090 --> 00:10:30,770
happens in our projects. We bring

280
00:10:30,770 --> 00:10:32,870
sometimes new programmers to the project

281
00:10:32,870 --> 00:10:34,940
which is one year old. And, that new

282
00:10:34,940 --> 00:10:37,550
developer starts writing code and making

283
00:10:37,550 --> 00:10:39,320
new classes, writing new unit tests,

284
00:10:39,320 --> 00:10:42,380
committing code, in just one hour. So, it

285
00:10:42,380 --> 00:10:44,090
doesn't take a month or two months to

286
00:10:44,090 --> 00:10:46,730
learn because we don't have experts. We

287
00:10:46,730 --> 00:10:48,410
don't have people who know "too much". We

288
00:10:48,410 --> 00:10:51,500
only have documents, and source code, and

289
00:10:51,500 --> 00:10:54,440
diagrams which explain everything. How we

290
00:10:54,440 --> 00:10:58,370
achieve that: We - first of all - we rotate

291
00:10:58,370 --> 00:11:00,980
people. We change people from project to

292
00:11:00,980 --> 00:11:03,170
project. We move them because we

293
00:11:03,170 --> 00:11:05,540
understand that the more experts we have

294
00:11:05,540 --> 00:11:06,690
the bigger is the trouble

295
00:11:06,690 --> 00:11:08,880
for us. We don't want experts. We want

296
00:11:08,880 --> 00:11:12,480
people [to] be - How [do] they say it? - like commodity

297
00:11:12,480 --> 00:11:14,460
programmers. So, they're like - they can

298
00:11:14,460 --> 00:11:15,840
work here, and then, tomorrow they can

299
00:11:15,840 --> 00:11:17,910
work there. So they can go from project

300
00:11:17,910 --> 00:11:19,920
to project and can contribute there as

301
00:11:19,920 --> 00:11:22,020
effectively as they were contributing

302
00:11:22,020 --> 00:11:23,730
yesterday in a different project. So,

303
00:11:23,730 --> 00:11:25,710
first of all, we did - i'll say a

304
00:11:25,710 --> 00:11:28,980
few words about it later - But, here we

305
00:11:28,980 --> 00:11:33,110
rotate them. We change them. And, second,

306
00:11:33,110 --> 00:11:38,640
we promote the idea of "always blame the

307
00:11:38,640 --> 00:11:40,050
project, don't blame yourselves". If you

308
00:11:40,050 --> 00:11:41,640
start working in a project, you don't

309
00:11:41,640 --> 00:11:43,110
understand anything. You don't need to

310
00:11:43,110 --> 00:11:45,240
learn. You need to say that the project

311
00:11:45,240 --> 00:11:47,460
is not is not sufficient enough for me

312
00:11:47,460 --> 00:11:49,740
to understand how it works. If you open

313
00:11:49,740 --> 00:11:51,540
the piece of code, if you look at it, you

314
00:11:51,540 --> 00:11:52,800
don't understand how it works.

315
00:11:52,800 --> 00:11:54,630
You don't learn it. You don't ask anybody.

316
00:11:54,630 --> 00:11:56,490
You don't find an expert. You just say,

317
00:11:56,490 --> 00:11:58,620
"Hey, your class or your piece of code is

318
00:11:58,620 --> 00:12:01,350
not clear enough. So, fix it." And then, we

319
00:12:01,350 --> 00:12:03,420
fix it for you. And then, it will be easier

320
00:12:03,420 --> 00:12:05,430
for you to understand it later. So, we

321
00:12:05,430 --> 00:12:07,950
always ask programmers to blame us - the

322
00:12:07,950 --> 00:12:11,790
project - for not being clear enough. And,

323
00:12:11,790 --> 00:12:14,520
we fix that. We improve the quality of

324
00:12:14,520 --> 00:12:16,020
the documentation. And, we bring it back

325
00:12:16,020 --> 00:12:18,420
to you and ask, "Can you fix it now?" If you

326
00:12:18,420 --> 00:12:20,520
can't fix it now, then again you blame

327
00:12:20,520 --> 00:12:22,680
us. And then, in a number of [those] cycles,

328
00:12:22,680 --> 00:12:25,110
eventually the project becomes as clear

329
00:12:25,110 --> 00:12:28,320
for you so you can understand it in half

330
00:12:28,320 --> 00:12:31,020
an hour. You probably have questions.

331
00:12:31,020 --> 00:12:34,650
We'll get back to it. [audience laughs] Number four: We have

332
00:12:34,650 --> 00:12:39,080
five in total. Number four: "Meetings"

333
00:12:39,080 --> 00:12:42,570
Traditionally, people believe that the

334
00:12:42,570 --> 00:12:44,880
more programmers communicate, the more

335
00:12:44,880 --> 00:12:47,250
meetings they have, the higher the

336
00:12:47,250 --> 00:12:49,440
chances of success in the project. The

337
00:12:49,440 --> 00:12:51,510
more - the closer - they stay together, the

338
00:12:51,510 --> 00:12:53,670
more of these face-to-face communications

339
00:12:53,670 --> 00:12:55,890
they have, the higher the chances that

340
00:12:55,890 --> 00:12:57,240
they will actually understand what's

341
00:12:57,240 --> 00:12:58,650
going on. And, they will produce something.

342
00:12:58,650 --> 00:13:01,290
So, that's why we have these Slack chats. We

343
00:13:01,290 --> 00:13:02,700
have these Skype calls. We have

344
00:13:02,700 --> 00:13:05,460
conference calls. Sometimes, the senior

345
00:13:05,460 --> 00:13:07,140
developers spend half of their day

346
00:13:07,140 --> 00:13:09,240
or maybe sometimes the full day in just

347
00:13:09,240 --> 00:13:11,010
meetings and just talking and talking

348
00:13:11,010 --> 00:13:13,140
and exchanging information. And, this is

349
00:13:13,140 --> 00:13:15,180
what - you know - managers believe is a good

350
00:13:15,180 --> 00:13:17,520
way to do. They think that senior

351
00:13:17,520 --> 00:13:19,110
developers have to be in [those] meetings.

352
00:13:19,110 --> 00:13:19,380
They [the senior developers]

353
00:13:19,380 --> 00:13:21,360
have to spend time on talking instead of

354
00:13:21,360 --> 00:13:23,850
writing code. And, even [the] Agile manifesto

355
00:13:23,850 --> 00:13:26,040
says so, like [that] face-to-face communications

356
00:13:26,040 --> 00:13:28,380
is always better than documentation. We

357
00:13:28,380 --> 00:13:29,960
completely disagree with that.

358
00:13:29,960 --> 00:13:32,400
We believe that the more meetings you

359
00:13:32,400 --> 00:13:34,440
have, the more informal communications,

360
00:13:34,440 --> 00:13:36,270
the more these chats you have, and the

361
00:13:36,270 --> 00:13:38,100
more face-to-face calls or whatever you

362
00:13:38,100 --> 00:13:40,590
call it by - "remote calls" - doesn't matter.

363
00:13:40,590 --> 00:13:43,140
The more you have of that, the bigger the

364
00:13:43,140 --> 00:13:44,780
trouble you [have] caused to the project.

365
00:13:44,780 --> 00:13:46,920
Because, you exchange the information

366
00:13:46,920 --> 00:13:49,530
which stays nowhere. We talk to each

367
00:13:49,530 --> 00:13:51,510
other. I explain [to] you how it works. You

368
00:13:51,510 --> 00:13:53,940
understand how it works. But tomorrow,

369
00:13:53,940 --> 00:13:55,710
somebody else will join the project. And,

370
00:13:55,710 --> 00:13:57,330
that person will not know how it works.

371
00:13:57,330 --> 00:13:58,920
So, we will have to explain again. We'll

372
00:13:58,920 --> 00:14:00,870
have to explain again. So, every minute

373
00:14:00,870 --> 00:14:02,940
spent in the meeting is the wasted money

374
00:14:02,940 --> 00:14:05,880
of the customer. So, the more we talk, the

375
00:14:05,880 --> 00:14:07,950
more we exchange information informally,

376
00:14:07,950 --> 00:14:09,870
the more money we waste, which the

377
00:14:09,870 --> 00:14:12,810
customer is paying us. So, good developers,

378
00:14:12,810 --> 00:14:15,180
I believe, and - we seriously preach

379
00:14:15,180 --> 00:14:17,250
this idea - the good developers - good

380
00:14:17,250 --> 00:14:19,020
engineers - they exchange informations

381
00:14:19,020 --> 00:14:20,910
through documents, through

382
00:14:20,910 --> 00:14:24,210
diagrams, through the source code, through

383
00:14:24,210 --> 00:14:26,130
formal written communication. [Good engineers use written communication] which is

384
00:14:26,130 --> 00:14:28,440
maintainable, which we can get back to in

385
00:14:28,440 --> 00:14:30,720
a year, in a month, in a week, in an hour,

386
00:14:30,720 --> 00:14:33,420
and understand why that decision was

387
00:14:33,420 --> 00:14:35,460
made, who made that decision, [and] what was the

388
00:14:35,460 --> 00:14:38,640
reason behind that decision. To achieve

389
00:14:38,640 --> 00:14:41,400
that - you know - not to enforce,

390
00:14:41,400 --> 00:14:43,740
but encourage people to actually use

391
00:14:43,740 --> 00:14:47,100
documents instead of calls, we don't

392
00:14:47,100 --> 00:14:52,050
allow them to call. [laughs] We just - that's so

393
00:14:52,050 --> 00:14:54,170
easy. [clapping in the audience]

394
00:14:55,400 --> 00:14:58,710
We just - we just don't. [laughs] They just don't

395
00:14:58,710 --> 00:15:01,620
have phone numbers of each other. So, we

396
00:15:01,620 --> 00:15:03,510
don't have that Slack chats. We don't

397
00:15:03,510 --> 00:15:05,010
have that Skype calls. They don't know, I

398
00:15:05,010 --> 00:15:06,450
mean, they know each other, but only

399
00:15:06,450 --> 00:15:08,280
through... We work on GitHub. So, they only

400
00:15:08,280 --> 00:15:10,350
know their GitHub accounts. All they

401
00:15:10,350 --> 00:15:12,030
can do in a project... If I don't know

402
00:15:12,030 --> 00:15:13,860
anything, for example: I joined the

403
00:15:13,860 --> 00:15:15,540
project. There are like 20 developers

404
00:15:15,540 --> 00:15:16,860
in there. I don't have some

405
00:15:16,860 --> 00:15:19,230
information. I can't call anybody. I can't

406
00:15:19,230 --> 00:15:21,570
ask anybody. All I can do is I can submit

407
00:15:21,570 --> 00:15:23,430
the ticket and say, "Hey, this piece of

408
00:15:23,430 --> 00:15:24,630
information is missing from that

409
00:15:24,630 --> 00:15:27,480
document. Can you please - you know - fix

410
00:15:27,480 --> 00:15:29,010
that and put that piece of information

411
00:15:29,010 --> 00:15:31,500
into that document?" And, somebody will fix

412
00:15:31,500 --> 00:15:33,090
that for me. I don't care who it's gonna

413
00:15:33,090 --> 00:15:35,430
be. I don't care who's gonna fix that. All

414
00:15:35,430 --> 00:15:37,170
I care about is that I look at the

415
00:15:37,170 --> 00:15:38,850
document and I find that missing piece

416
00:15:38,850 --> 00:15:42,060
of information. That's the idea. So, if you

417
00:15:42,060 --> 00:15:44,370
allow programmers - you know - to talk to

418
00:15:44,370 --> 00:15:46,620
each other, they will. If you allow them

419
00:15:46,620 --> 00:15:48,240
to call, they will, because it's naturally

420
00:15:48,240 --> 00:15:49,740
better for us - you know - to communicate -

421
00:15:49,740 --> 00:15:52,230
you know - face to face. Just - here's the - my

422
00:15:52,230 --> 00:15:53,670
friend sitting next to me. I'm gonna ask

423
00:15:53,670 --> 00:15:55,440
him. I'm not gonna - you know -

424
00:15:55,440 --> 00:15:57,360
get the document. I'm gonna just ask and

425
00:15:57,360 --> 00:15:58,080
get the information.

426
00:15:58,080 --> 00:16:00,120
We completely disallow that. For us, it's

427
00:16:00,120 --> 00:16:02,670
easy because we work remotely. So, all our

428
00:16:02,670 --> 00:16:04,830
developers are somewhere. They're like -

429
00:16:04,830 --> 00:16:06,300
you know - in different countries, in

430
00:16:06,300 --> 00:16:08,010
different time zones. So, we don't have a

431
00:16:08,010 --> 00:16:10,110
central place. So, we just keep them like

432
00:16:10,110 --> 00:16:12,000
that. And, we don't allow them to - you know -

433
00:16:12,000 --> 00:16:14,400
to get into some Slack chat and talk. All

434
00:16:14,400 --> 00:16:15,960
they can do is communicate through

435
00:16:15,960 --> 00:16:17,460
ticketing systems, through the source

436
00:16:17,460 --> 00:16:19,320
code. And that's how we achieve that

437
00:16:19,320 --> 00:16:21,300
results. So, we have only documents and

438
00:16:21,300 --> 00:16:22,980
only diagrams. And, I'm saying it again.

439
00:16:22,980 --> 00:16:25,020
Sometimes, it's very often we have

440
00:16:25,020 --> 00:16:27,060
projects which are one year old and the

441
00:16:27,060 --> 00:16:29,130
new developer jumps in and there's like

442
00:16:29,130 --> 00:16:31,410
50 thousand lines of code of Java code -

443
00:16:31,410 --> 00:16:33,570
like serious Java code. And, the new

444
00:16:33,570 --> 00:16:35,580
developer jumps in and starts writing

445
00:16:35,580 --> 00:16:37,200
code in just one hour.

446
00:16:37,200 --> 00:16:39,900
It's unachievable in a

447
00:16:39,900 --> 00:16:43,890
traditional project configuration. The

448
00:16:43,890 --> 00:16:46,640
last one. And [then] we [will] have questions. "Salaries":

449
00:16:46,640 --> 00:16:50,580
The biggest problem. So,

450
00:16:50,580 --> 00:16:53,550
traditionally, we believe

451
00:16:53,550 --> 00:16:57,000
that - I mean, everybody believes that - it's

452
00:16:57,000 --> 00:16:59,280
so important to keep the developer happy

453
00:16:59,280 --> 00:17:02,010
and to keep the developer safe. And, it's -

454
00:17:02,010 --> 00:17:04,260
you know - to get and keep the loyalty from

455
00:17:04,260 --> 00:17:06,190
the person. So, we need long time, long

456
00:17:06,190 --> 00:17:10,360
term contracts, good salary, paid every

457
00:17:10,360 --> 00:17:13,570
month, and that's how we're gonna

458
00:17:13,570 --> 00:17:15,640
achieve - you know - good results from that

459
00:17:15,640 --> 00:17:18,580
particular person. So, we usually pay for

460
00:17:18,580 --> 00:17:22,540
when - not we, but you - you usually get paid for

461
00:17:22,540 --> 00:17:25,630
per month, per time, sometimes per hour,

462
00:17:25,630 --> 00:17:28,720
sometimes per month. But, it's always for

463
00:17:28,720 --> 00:17:30,760
time. You just sit in front of the

464
00:17:30,760 --> 00:17:32,650
computer, and you do something, and they

465
00:17:32,650 --> 00:17:34,810
pay you for this time. You achieve the

466
00:17:34,810 --> 00:17:36,220
result, they pay you. You don't achieve

467
00:17:36,220 --> 00:17:39,940
the result, they still pay. You work,

468
00:17:39,940 --> 00:17:41,320
they pay you.

469
00:17:41,320 --> 00:17:44,050
You don't work, they still pay you. That's

470
00:17:44,050 --> 00:17:46,480
a huge problem, I believe, especially for

471
00:17:46,480 --> 00:17:49,630
A players. Remember that A, B, C players. We

472
00:17:49,630 --> 00:17:51,610
have "A players", people who really want to

473
00:17:51,610 --> 00:17:53,260
achieve something, and we have "C players"

474
00:17:53,260 --> 00:17:55,300
who just enjoy being in front of the

475
00:17:55,300 --> 00:17:58,750
computer. So, the A players,

476
00:17:58,750 --> 00:18:01,540
they don't like that. They will

477
00:18:01,540 --> 00:18:03,400
leave your project. And, they will not be

478
00:18:03,400 --> 00:18:05,290
productive at all because this

479
00:18:05,290 --> 00:18:07,990
format - you know - puts everybody on the

480
00:18:07,990 --> 00:18:10,240
same level. [That is] because, if I'm the A player

481
00:18:10,240 --> 00:18:12,670
and I produce results, I'm still

482
00:18:12,670 --> 00:18:14,290
paid more or less at the same level as

483
00:18:14,290 --> 00:18:16,270
the C player who produces zero results.

484
00:18:16,270 --> 00:18:20,380
And then another [issue] is "soft skills". So then, in

485
00:18:20,380 --> 00:18:22,210
the end, to make a decision [about] who's gonna

486
00:18:22,210 --> 00:18:23,920
get more salary, who's gonna get the raise,

487
00:18:23,920 --> 00:18:25,720
who's gonna get the - you know - who's gonna

488
00:18:25,720 --> 00:18:28,510
get fired - we use "soft skills". Who is the

489
00:18:28,510 --> 00:18:30,760
nicer guy in the team, who talks more,

490
00:18:30,760 --> 00:18:32,410
who is more communicative, who is the

491
00:18:32,410 --> 00:18:34,060
good match for the team - that's how we

492
00:18:34,060 --> 00:18:36,130
decide who gets the raise. [This is] instead of

493
00:18:36,130 --> 00:18:37,810
looking at the real results. Instead of

494
00:18:37,810 --> 00:18:39,280
looking at the numbers. I don't know [if it's]

495
00:18:39,280 --> 00:18:41,350
numbers but, [I mean] the actual results that

496
00:18:41,350 --> 00:18:43,720
people provide. So, we use the "soft skills".

497
00:18:43,720 --> 00:18:45,400
It completely demotivates A

498
00:18:45,400 --> 00:18:48,460
players, in my experience. So, we remove

499
00:18:48,460 --> 00:18:50,410
that. We don't have salaries at all. Well,

500
00:18:50,410 --> 00:18:52,770
we pay money, but we don't pay salaries. [laughs]

501
00:18:52,770 --> 00:18:55,560
We pay for results. Every time

502
00:18:55,560 --> 00:18:58,360
developers complete something - close the

503
00:18:58,360 --> 00:19:01,840
ticket, file a bug. Basically two things:

504
00:19:01,840 --> 00:19:03,970
Close the ticket, and file the bug. So, fix

505
00:19:03,970 --> 00:19:06,760
the bug or file a bug. We pay for that.

506
00:19:06,760 --> 00:19:09,250
And, we do micro payments. So, we pay our

507
00:19:09,250 --> 00:19:11,200
developers in like really small,

508
00:19:11,200 --> 00:19:12,910
micro payments. We can do that because we

509
00:19:12,910 --> 00:19:14,530
work with remote programmers and

510
00:19:14,530 --> 00:19:17,320
freelancers mostly. And, we pay them

511
00:19:17,320 --> 00:19:19,430
through like PayPal, BitCoin,

512
00:19:19,430 --> 00:19:21,050
whatever. So, we just - you know - we have

513
00:19:21,050 --> 00:19:22,670
these transactions - many transactions -

514
00:19:22,670 --> 00:19:24,950
every day. And, the more they commit, the

515
00:19:24,950 --> 00:19:26,330
more they, not commit but, the more

516
00:19:26,330 --> 00:19:28,130
results they produce... It's not about the

517
00:19:28,130 --> 00:19:29,630
number of lines of code,

518
00:19:29,630 --> 00:19:33,050
It's not about some productivity metrics.

519
00:19:33,050 --> 00:19:35,120
It's about, close the task, or not close

520
00:19:35,120 --> 00:19:36,740
the task. Here's the problem: You either

521
00:19:36,740 --> 00:19:38,420
solve the problem or you don't solve the

522
00:19:38,420 --> 00:19:40,010
problem. When the problem is solved... And,

523
00:19:40,010 --> 00:19:41,750
we have small problems. We always have

524
00:19:41,750 --> 00:19:43,640
small tasks, half an hour, or one hour,

525
00:19:43,640 --> 00:19:46,730
really micro tasks... The task is closed. We

526
00:19:46,730 --> 00:19:48,380
pay. The task is not closed - we don't pay.

527
00:19:48,380 --> 00:19:50,600
So, there's no salaries. This is how we

528
00:19:50,600 --> 00:19:52,940
make it so easy to measure the

529
00:19:52,940 --> 00:19:55,400
performance, and productivity, and

530
00:19:55,400 --> 00:19:57,590
motivation of everybody in the project.

531
00:19:57,590 --> 00:19:59,510
Because, we - I mean - we don't pay in the

532
00:19:59,510 --> 00:20:00,860
end of the month. We pay when the project -

533
00:20:00,860 --> 00:20:03,790
when the task - is closed.

534
00:20:03,790 --> 00:20:06,290
So, I'm basically done. I think you will

535
00:20:06,290 --> 00:20:07,880
have questions but let me [first] summarize. So,

536
00:20:07,880 --> 00:20:09,890
in this model of work which I just

537
00:20:09,890 --> 00:20:11,960
explained, we've been working for about

538
00:20:11,960 --> 00:20:16,160
three years already. 250 developers went

539
00:20:16,160 --> 00:20:18,080
through this model. So, 250 people

540
00:20:18,080 --> 00:20:19,910
actually tried that and got some money

541
00:20:19,910 --> 00:20:22,250
from us in different tasks

542
00:20:22,250 --> 00:20:24,470
and different projects. [These projects are] mostly Java, 90%

543
00:20:24,470 --> 00:20:29,780
is Java. 25 projects completed. Commercial

544
00:20:29,780 --> 00:20:31,580
and open source. We have open source

545
00:20:31,580 --> 00:20:33,350
projects as well, so you can look at them.

546
00:20:33,350 --> 00:20:34,850
Come to me later, [and] I'll give you my

547
00:20:34,850 --> 00:20:36,500
coordinates. You can look at the GitHub

548
00:20:36,500 --> 00:20:37,940
account. And, you will see a number of

549
00:20:37,940 --> 00:20:40,970
projects which were completed using this

550
00:20:40,970 --> 00:20:42,590
technology - this approach. And, you

551
00:20:42,590 --> 00:20:44,180
will see how we communicate to each

552
00:20:44,180 --> 00:20:46,760
other in GitHub tickets. So, you will see

553
00:20:46,760 --> 00:20:48,320
how we get rid of these meetings [completely].

554
00:20:48,320 --> 00:20:51,650
And, all our communications are there. You

555
00:20:51,650 --> 00:20:53,420
will see even how we pay programmers

556
00:20:53,420 --> 00:20:55,190
because all these like payment - you know -

557
00:20:55,190 --> 00:20:57,560
successful payment notifications are

558
00:20:57,560 --> 00:20:59,150
there in GitHub. It will be quite

559
00:20:59,150 --> 00:21:01,720
interesting. And, we produced about

560
00:21:01,720 --> 00:21:04,100
300,000 lines of Java code. It's pretty

561
00:21:04,100 --> 00:21:05,960
big amount of code. And, all of these

562
00:21:05,960 --> 00:21:06,860
projects were really successful

563
00:21:06,860 --> 00:21:09,020
comparing to what I had before for many

564
00:21:09,020 --> 00:21:11,390
years. So, I'm asking you to follow me.

565
00:21:11,390 --> 00:21:13,520
Subscribe to the blog and you will

566
00:21:13,520 --> 00:21:15,350
learn more. So, now questions from you

567
00:21:15,350 --> 00:21:17,470
guys.  [audience applause]

568
00:21:21,670 --> 00:21:24,860
Moderator: So, we have a plenty of questions in

569
00:21:24,860 --> 00:21:27,440
slider and I will start with the most

570
00:21:27,440 --> 00:21:31,760
voted ones. Okay? And, so you talk about

571
00:21:31,760 --> 00:21:34,070
information but, what about designing

572
00:21:34,070 --> 00:21:37,280
together? Because the design is - it's - a

573
00:21:37,280 --> 00:21:39,070
product of a conversation.

574
00:21:39,070 --> 00:21:41,750
Yegor: Well, yeah. It's... yeah. I've heard the

575
00:21:41,750 --> 00:21:43,700
question many times. So, people again

576
00:21:43,700 --> 00:21:45,530
believe that the design is the

577
00:21:45,530 --> 00:21:47,390
result of the conversation of the

578
00:21:47,390 --> 00:21:52,040
work of the group. But, in my experience

579
00:21:52,040 --> 00:21:54,110
it's not that true. In my experience, the

580
00:21:54,110 --> 00:21:56,090
design is always the result of one

581
00:21:56,090 --> 00:21:58,880
person thinking and asking people around.

582
00:21:58,880 --> 00:22:01,670
So, when we start a project, we always

583
00:22:01,670 --> 00:22:04,070
have some person as an architect - in

584
00:22:04,070 --> 00:22:05,150
the role of the architect - and that

585
00:22:05,150 --> 00:22:06,890
person is responsible for creating the

586
00:22:06,890 --> 00:22:08,750
skeleton, the prototype of the product.

587
00:22:08,750 --> 00:22:11,240
That person may collect information from

588
00:22:11,240 --> 00:22:12,650
other people, asking, "What do you think

589
00:22:12,650 --> 00:22:14,420
about this? I'm not sure about choosing

590
00:22:14,420 --> 00:22:16,280
the database. So, tell me which database

591
00:22:16,280 --> 00:22:18,530
would you think I should use? Tell me

592
00:22:18,530 --> 00:22:20,180
about the technology I should use." So,

593
00:22:20,180 --> 00:22:22,370
questions are welcome. That the architect

594
00:22:22,370 --> 00:22:24,680
has to have - has to ask - questions, this

595
00:22:24,680 --> 00:22:26,240
is perfect. But, the final decision -

596
00:22:26,240 --> 00:22:28,580
the final design - has to be made by one

597
00:22:28,580 --> 00:22:31,020
person. That's what I believe.

598
00:22:31,020 --> 00:22:34,910
Moderator: Buf if they get paid for answers, if they don't get paid, they don't get to ask questions.

599
00:22:34,910 --> 00:22:37,610
Yegor: We pay for answering questions, of course. But,

600
00:22:37,610 --> 00:22:39,470
all questions are like tickets. So, let's

601
00:22:39,470 --> 00:22:40,880
say, for example, you're a good database

602
00:22:40,880 --> 00:22:42,980
engineer, yeah. And, I am like an architect.

603
00:22:42,980 --> 00:22:44,210
And, I don't really know what kind of

604
00:22:44,210 --> 00:22:45,770
database to choose, MySQL or

605
00:22:45,770 --> 00:22:47,660
PostgreSQL. So, I'm sorry, I'm submitting a

606
00:22:47,660 --> 00:22:50,150
ticket and saying, "Hey, I['m] miss[ing] a

607
00:22:50,150 --> 00:22:53,210
comparison of MySQL and PostgreSQL. I

608
00:22:53,210 --> 00:22:55,130
need a comparison provided for me. This

609
00:22:55,130 --> 00:22:57,260
is a task for you for an hour. So, you

610
00:22:57,260 --> 00:22:59,030
work for an hour, give me the table, a

611
00:22:59,030 --> 00:23:01,400
spreadsheet, comparing why MySQL is

612
00:23:01,400 --> 00:23:03,020
better for us and why PostgreSQL is better

613
00:23:03,020 --> 00:23:04,820
for us. And, I say, "Thank you very much." I

614
00:23:04,820 --> 00:23:07,880
know that. Now, let me make a decision. So,

615
00:23:07,880 --> 00:23:09,500
that's how it should be - iteratively

616
00:23:09,500 --> 00:23:11,180
goes through questions, and answers, and

617
00:23:11,180 --> 00:23:13,010
then finally, I will be able to make my

618
00:23:13,010 --> 00:23:16,190
decision. Again, it's not easy. It's way

619
00:23:16,190 --> 00:23:17,810
easier to sit together in the same room

620
00:23:17,810 --> 00:23:18,950
and say, "Hey, what do you think about

621
00:23:18,950 --> 00:23:20,930
PostgreSQL?" and you say, "Well, MySQL

622
00:23:20,930 --> 00:23:22,610
is better." "Oh, that's right, thank you very

623
00:23:22,610 --> 00:23:24,290
much!" and then I make a decision. But then,

624
00:23:24,290 --> 00:23:26,390
in two months, somebody will come and

625
00:23:26,390 --> 00:23:28,220
say, "Hey, why the decision for MySQL

626
00:23:28,220 --> 00:23:30,590
was made?" and, I'm gone from the project

627
00:23:30,590 --> 00:23:32,300
because - you know - they fired me already.

628
00:23:32,300 --> 00:23:33,950
You're gone as well. [audience laughs] They fired

629
00:23:33,950 --> 00:23:36,320
you, too. And then, the question is, "Who

630
00:23:36,320 --> 00:23:38,240
made the decision?" Well, there were two

631
00:23:38,240 --> 00:23:39,650
guys - you know - yeah. I still have the

632
00:23:39,650 --> 00:23:41,300
phone number of the one guy. Let me call

633
00:23:41,300 --> 00:23:44,390
him. But instead, we're gonna have the

634
00:23:44,390 --> 00:23:46,490
detailed conversation. And, the ticket for

635
00:23:46,490 --> 00:23:47,930
that. And, everybody will be able to jump

636
00:23:47,930 --> 00:23:50,210
in and see, "Yeah, that guy who is

637
00:23:50,210 --> 00:23:52,520
fired asked the question, "Why MySQL?"

638
00:23:52,520 --> 00:23:54,470
And, another guy submitted the detailed

639
00:23:54,470 --> 00:23:57,730
explanation of them. That's a perfect meeting traceability.

640
00:23:57,920 --> 00:23:58,940
Moderator: My friends, there

641
00:23:58,940 --> 00:24:00,710
are so many questions in slider. Go

642
00:24:00,710 --> 00:24:02,420
ahead and vote for questions, because,

643
00:24:02,420 --> 00:24:05,360
oh, let's make it moderated, OK? I'll

644
00:24:05,360 --> 00:24:08,060
ask the top most questions. And, there are

645
00:24:08,060 --> 00:24:09,740
many. So, there is the - it's not a

646
00:24:09,740 --> 00:24:11,480
question. It's like a complaint: Three

647
00:24:11,480 --> 00:24:13,760
hours of work, are you kidding? My manager

648
00:24:13,760 --> 00:24:15,590
will kill me. The stock will decrease. And,

649
00:24:15,590 --> 00:24:19,010
the company will close. Yegor: For three hours

650
00:24:19,010 --> 00:24:20,960
of work? Moderator: Three hours of work, are you kidding me? Yegor: Yeah

651
00:24:20,960 --> 00:24:23,330
per day. Yeah. I believe honestly that a

652
00:24:23,330 --> 00:24:25,670
productive - like a good engineer - Well, I'm

653
00:24:25,670 --> 00:24:28,160
talking about myself, first of all. So, I

654
00:24:28,160 --> 00:24:30,410
can't imagine working for eight hours a

655
00:24:30,410 --> 00:24:32,510
day on something which is not my

656
00:24:32,510 --> 00:24:34,250
personal project - you know - which I really

657
00:24:34,250 --> 00:24:36,920
love. So, working for somebody, for some

658
00:24:36,920 --> 00:24:39,890
project, for money, 3-4 hours a day, is the

659
00:24:39,890 --> 00:24:42,140
total maximum and normal - you know - good

660
00:24:42,140 --> 00:24:44,300
engineer can really handle, I think.

661
00:24:44,300 --> 00:24:44,660
So,

662
00:24:44,660 --> 00:24:47,090
otherwise, it's eight hours and three

663
00:24:47,090 --> 00:24:49,190
five hours of Facebook watching - you know -

664
00:24:49,190 --> 00:24:49,990
out of this.

665
00:24:49,990 --> 00:24:52,850
So, it's gonna be three hours, in any case.

666
00:24:52,850 --> 00:24:54,890
So, you're not going to be productive for

667
00:24:54,890 --> 00:24:56,930
eight hours, it's not possible. You can

668
00:24:56,930 --> 00:24:59,060
really write code for three [hours]. Sometimes,

669
00:24:59,060 --> 00:25:01,670
even two hours is enough. But, I think

670
00:25:01,670 --> 00:25:03,290
that, in general, programmers have to be

671
00:25:03,290 --> 00:25:05,990
paid way better. So, we need to earn way

672
00:25:05,990 --> 00:25:07,910
more. We need to make more money per hour

673
00:25:07,910 --> 00:25:10,370
and spend less time in front of the

674
00:25:10,370 --> 00:25:12,770
computer. You know [what I mean]? So, I don't know how

675
00:25:12,770 --> 00:25:14,510
much you're making, but an average like

676
00:25:14,510 --> 00:25:16,220
say, for example, now programmers are

677
00:25:16,220 --> 00:25:18,470
making like - personaly, I dunno - 25 euros an

678
00:25:18,470 --> 00:25:20,180
hour, thirty euros an hour.

679
00:25:20,180 --> 00:25:22,460
They have to make a hundred [Euros], but work

680
00:25:22,460 --> 00:25:26,060
three times less. In this case - but - they

681
00:25:26,060 --> 00:25:27,770
work three times less. But - they have to be

682
00:25:27,770 --> 00:25:30,740
paid for results, not for the time. So now,

683
00:25:30,740 --> 00:25:32,630
we are all paid for eight hours - you know -

684
00:25:32,630 --> 00:25:33,890
from the morning till the evening.

685
00:25:33,890 --> 00:25:36,320
Whatever I do, you pay me. But, it should

686
00:25:36,320 --> 00:25:37,460
be the other way. It should be the

687
00:25:37,460 --> 00:25:39,920
opposite: You pay me way more, but only

688
00:25:39,920 --> 00:25:42,020
for the results I produce. In the end, it

689
00:25:42,020 --> 00:25:43,670
will be the same amount of

690
00:25:43,670 --> 00:25:45,230
money in cash. Maybe a little bit more,

691
00:25:45,230 --> 00:25:47,450
but I will be focused on results.

692
00:25:47,450 --> 00:25:48,920
I'll be really motivated and

693
00:25:48,920 --> 00:25:51,140
interested to actually work, instead of

694
00:25:51,140 --> 00:25:53,000
just pretending I'm working, just sitting

695
00:25:53,000 --> 00:25:55,820
there. And, just going to, okay, so the

696
00:25:55,820 --> 00:25:57,290
three hours is that. Three hours is the

697
00:25:57,290 --> 00:26:00,170
maximum. Well, four. Moderator: The next question: Who

698
00:26:00,170 --> 00:26:02,780
decides how much time the problem

699
00:26:02,780 --> 00:26:05,810
takes to solve? If it's the

700
00:26:05,810 --> 00:26:08,000
business, how can [they] estimate? They always

701
00:26:08,000 --> 00:26:10,000
think the task is easy.

702
00:26:10,000 --> 00:26:14,720
Yegor: Well, yeah. We just, in our

703
00:26:14,720 --> 00:26:18,260
approach, we always give the same

704
00:26:18,260 --> 00:26:21,260
budget for all tasks. So, we always give

705
00:26:21,260 --> 00:26:23,780
it a really small budget. And then, we use

706
00:26:23,780 --> 00:26:25,340
this technology where we allow

707
00:26:25,340 --> 00:26:28,220
developers to return back the code in

708
00:26:28,220 --> 00:26:31,460
incomplete state. So, I guess you're

709
00:26:31,460 --> 00:26:32,990
developers, right? Who are the developers?

710
00:26:32,990 --> 00:26:36,890
Who actually writes code here? Like 75%,

711
00:26:36,890 --> 00:26:39,620
maybe more. So, you should understand that.

712
00:26:39,620 --> 00:26:42,530
Let's say you get a task [where] you have to

713
00:26:42,530 --> 00:26:44,390
implement that feature. And, the feature

714
00:26:44,390 --> 00:26:47,150
is way bigger than one hour. So, we don't

715
00:26:47,150 --> 00:26:49,460
ask you to - you know - to... We ask you to

716
00:26:49,460 --> 00:26:52,040
cheat, basically. We ask you to implement

717
00:26:52,040 --> 00:26:53,960
the - provide the high level

718
00:26:53,960 --> 00:26:56,540
implementation with certain parts not

719
00:26:56,540 --> 00:26:59,420
implemented yet. So, we say, this feature,

720
00:26:59,420 --> 00:27:00,950
for example, I have to be able to click

721
00:27:00,950 --> 00:27:04,100
the button and log in. And, this is one

722
00:27:04,100 --> 00:27:05,300
hour for you. And, you understand that's

723
00:27:05,300 --> 00:27:06,530
one hour. It's not enough. It should be

724
00:27:06,530 --> 00:27:08,720
like five hours. So, you just create a

725
00:27:08,720 --> 00:27:10,970
button. It says "log in". It's clickable.

726
00:27:10,970 --> 00:27:13,910
But, it doesn't log in. So, you just return

727
00:27:13,910 --> 00:27:15,860
back the code to me - to the master

728
00:27:15,860 --> 00:27:18,110
branch - and say, the button is there but,

729
00:27:18,110 --> 00:27:20,510
there is a marker [which] says "to-do later:

730
00:27:20,510 --> 00:27:22,730
Implement the log-in this and that way."

731
00:27:22,730 --> 00:27:25,130
So, you kind of help us to improve - to

732
00:27:25,130 --> 00:27:27,350
design that. And, by introducing the

733
00:27:27,350 --> 00:27:29,720
high level solution, and then keeping the

734
00:27:29,720 --> 00:27:32,000
markers of where other programmers have

735
00:27:32,000 --> 00:27:34,010
to contribute more. So, you return back

736
00:27:34,010 --> 00:27:36,140
the code with the markers. And then, these

737
00:27:36,140 --> 00:27:39,230
markers we convert back to new tasks. So,

738
00:27:39,230 --> 00:27:41,720
we ask developers to help us break down

739
00:27:41,720 --> 00:27:44,120
the problem into smaller pieces. That's

740
00:27:44,120 --> 00:27:46,490
how we do it. Moderator: Next question. Yegor: We call it

741
00:27:46,490 --> 00:27:48,110
"puzzle driven development". So, we call

742
00:27:48,110 --> 00:27:49,970
these pieces "puzzles". So, we give you

743
00:27:49,970 --> 00:27:51,980
the task. You return it back with puzzles

744
00:27:51,980 --> 00:27:53,840
inside. And then, somebody else will solve

745
00:27:53,840 --> 00:27:55,610
[those] puzzles. And while people

746
00:27:55,610 --> 00:27:56,870
solve the puzzles, they will introduce

747
00:27:56,870 --> 00:27:59,150
new puzzles. So, when we start developing,

748
00:27:59,150 --> 00:28:00,860
it's just a few puzzles. And then, it goes

749
00:28:00,860 --> 00:28:02,270
like that. [Yegor gestures an expanding mountain with his hands] So, there are like

750
00:28:02,270 --> 00:28:04,130
hundreds of puzzles. And then, when all of

751
00:28:04,130 --> 00:28:06,050
them get fixed, the software gets back [together],

752
00:28:06,050 --> 00:28:08,300
and it starts to work well. It becomes

753
00:28:08,300 --> 00:28:10,610
completely functional. If you're interested,

754
00:28:10,610 --> 00:28:12,170
come back to me later. I'll explain.

755
00:28:12,170 --> 00:28:13,100
Moderator: Next question.

756
00:28:13,100 --> 00:28:15,620
Isn't there a lack of quality? Developers

757
00:28:15,620 --> 00:28:17,660
deliver the code and go. There is no

758
00:28:17,660 --> 00:28:19,580
motivation in delivering good code. Who

759
00:28:19,580 --> 00:28:21,950
decides whether to pay or not?

760
00:28:22,100 --> 00:28:24,110
Yegor: Yeah, that's a good question. So, definitely. And,

761
00:28:24,110 --> 00:28:26,540
we really ask our developers

762
00:28:26,540 --> 00:28:29,960
to be lazy. So, be lazy. Don't try to do

763
00:28:29,960 --> 00:28:31,670
more than you can do. But, be prepared

764
00:28:31,670 --> 00:28:34,550
that the quality control is really

765
00:28:34,550 --> 00:28:36,770
strict in front of you. So, when you get

766
00:28:36,770 --> 00:28:38,120
the task and you have to implement

767
00:28:38,120 --> 00:28:40,790
something, do as little as you can. You

768
00:28:40,790 --> 00:28:42,620
don't need to - you know - to make it hard.

769
00:28:42,620 --> 00:28:44,690
You don't make the quality higher than

770
00:28:44,690 --> 00:28:46,460
the quality bar we have in front of you.

771
00:28:46,460 --> 00:28:48,890
And, the quality bar is quite high. So, for

772
00:28:48,890 --> 00:28:50,960
you to return back the code to us, to the

773
00:28:50,960 --> 00:28:53,120
master branch, it has to pass two code

774
00:28:53,120 --> 00:28:55,850
reviews plus the automated merge

775
00:28:55,850 --> 00:28:57,890
instruments. So, we're going to merge your

776
00:28:57,890 --> 00:29:00,500
branch back to master branch only if all

777
00:29:00,500 --> 00:29:02,510
the tests and all the static analysis

778
00:29:02,510 --> 00:29:04,310
and all the coverage control, everything

779
00:29:04,310 --> 00:29:07,250
passes. So, for example, they'll give you

780
00:29:07,250 --> 00:29:09,020
the example of how high is the quality

781
00:29:09,020 --> 00:29:12,170
bar for us. Static analysis is the step

782
00:29:12,170 --> 00:29:15,200
in the build for us. So, we

783
00:29:15,200 --> 00:29:16,910
don't with - The build is not going to

784
00:29:16,910 --> 00:29:17,480
pass

785
00:29:17,480 --> 00:29:20,420
if the static analyzer says there are

786
00:29:20,420 --> 00:29:23,060
any bugs, any warnings, [or] any complaints. You

787
00:29:23,060 --> 00:29:25,940
know what static analysis is, right? Do

788
00:29:25,940 --> 00:29:29,180
you know what static analysis is? Okay,

789
00:29:29,180 --> 00:29:29,570
good.

790
00:29:29,570 --> 00:29:31,220
So you understand the static analysis is

791
00:29:31,220 --> 00:29:33,560
usually is used as a tool which is - you

792
00:29:33,560 --> 00:29:35,510
know - which we run on the code and which

793
00:29:35,510 --> 00:29:37,580
generates some - you know - spreadsheets,

794
00:29:37,580 --> 00:29:39,470
some interesting web pages, which we'll

795
00:29:39,470 --> 00:29:42,050
look at and say, "Hey, the rate of our

796
00:29:42,050 --> 00:29:44,540
static analysis warnings is a bit lower

797
00:29:44,540 --> 00:29:47,180
than the yesterday. That's great, but,

798
00:29:47,180 --> 00:29:48,800
there's still a lot of them - Still a lot

799
00:29:48,800 --> 00:29:50,570
of complaints." Look at your project. Run

800
00:29:50,570 --> 00:29:52,400
the static analyzer for your project. No

801
00:29:52,400 --> 00:29:53,840
matter what language you're using, you

802
00:29:53,840 --> 00:29:55,730
will see there many of them. In our case,

803
00:29:55,730 --> 00:29:58,490
there is zero. So, we run the static

804
00:29:58,490 --> 00:30:00,320
analysis on the entire code base. And,

805
00:30:00,320 --> 00:30:02,630
there should be zero warnings, which is

806
00:30:02,630 --> 00:30:04,040
really difficult to achieve. If you're

807
00:30:04,040 --> 00:30:05,420
serious, if your static analysis is

808
00:30:05,420 --> 00:30:07,100
configured properly, it's really

809
00:30:07,100 --> 00:30:08,360
difficult to achieve. Especially, there

810
00:30:08,360 --> 00:30:09,590
are many people working there. And, the

811
00:30:09,590 --> 00:30:11,630
code base is quite big in our case. No

812
00:30:11,630 --> 00:30:13,070
matter how big is the code base, no

813
00:30:13,070 --> 00:30:13,760
matter how

814
00:30:13,760 --> 00:30:15,470
many developers are there, we're

815
00:30:15,470 --> 00:30:17,360
not gonna merge your stuff unless

816
00:30:17,360 --> 00:30:19,640
there's no bugs. There's no static

817
00:30:19,640 --> 00:30:21,590
analysis complaints. So, that's an

818
00:30:21,590 --> 00:30:23,570
example of how high is the quality bar.

819
00:30:23,570 --> 00:30:25,970
So, what I'm saying is that, we're kind of

820
00:30:25,970 --> 00:30:27,650
building a conflict. Here's your task.

821
00:30:27,650 --> 00:30:29,750
You're gonna get the money when it's

822
00:30:29,750 --> 00:30:31,460
merged. But, you're not gonna get [your code] merged

823
00:30:31,460 --> 00:30:34,580
if the quality is not high enough. And, we

824
00:30:34,580 --> 00:30:37,490
do it for every single task. So, for each -

825
00:30:37,490 --> 00:30:39,740
for every single increment. For every

826
00:30:39,740 --> 00:30:41,540
single developer who is trying to change

827
00:30:41,540 --> 00:30:43,760
the code and commit the new stuff into

828
00:30:43,760 --> 00:30:45,980
our code base into master. We always

829
00:30:45,980 --> 00:30:47,450
check the quality. We'll always check the

830
00:30:47,450 --> 00:30:49,340
quality. So, when the project is one year

831
00:30:49,340 --> 00:30:52,100
old, we merged already a thousand pull

832
00:30:52,100 --> 00:30:54,470
requests. And, for all these thousand pull

833
00:30:54,470 --> 00:30:56,660
requests, for thousand[s of] steps, micro

834
00:30:56,660 --> 00:30:58,460
increments, we check the quality again,

835
00:30:58,460 --> 00:31:00,470
again, and again. So, in the end, the

836
00:31:00,470 --> 00:31:02,150
quality is really high. Again, you can

837
00:31:02,150 --> 00:31:03,770
check our GitHub projects in Java. You

838
00:31:03,770 --> 00:31:05,540
will see how high is

839
00:31:05,540 --> 00:31:08,450
the quality of our code. Moderator: Okay, great.

840
00:31:08,450 --> 00:31:10,220
There was a comment. Maybe you have

841
00:31:10,220 --> 00:31:12,740
something to say about it. Even prisoners

842
00:31:12,740 --> 00:31:17,240
are allowed one call. Yegor: [laughs] Well, yeah. For the

843
00:31:17,240 --> 00:31:18,770
call. Yeah. I've heard that complain a lot.

844
00:31:18,770 --> 00:31:20,570
So, people saying like, "Hey, we're people.

845
00:31:20,570 --> 00:31:21,890
We love to call each other. We love

846
00:31:21,890 --> 00:31:23,630
to say good morning sometimes. What kind

847
00:31:23,630 --> 00:31:25,370
of work is that when there is like, yeah,

848
00:31:25,370 --> 00:31:27,860
people don't talk to each other?" I also

849
00:31:27,860 --> 00:31:30,920
thought so. I had that fear years ago

850
00:31:30,920 --> 00:31:32,540
when I introduced that [no call policy]. I was like hey,

851
00:31:32,540 --> 00:31:34,310
programmers not gonna like it, because

852
00:31:34,310 --> 00:31:36,290
they're kind of used to talking to each other.

853
00:31:36,290 --> 00:31:38,780
They love these, all these Slack chats, and

854
00:31:38,780 --> 00:31:41,660
hashtags, and everything. But, when we're

855
00:31:41,660 --> 00:31:43,430
talking about A players, when we're

856
00:31:43,430 --> 00:31:45,110
talking about senior developers, they

857
00:31:45,110 --> 00:31:47,930
really don't like that. They like to stay

858
00:31:47,930 --> 00:31:49,430
focused on their stuff, which is really

859
00:31:49,430 --> 00:31:51,350
interesting for them. They have other

860
00:31:51,350 --> 00:31:53,060
places where we can - where they can - talk

861
00:31:53,060 --> 00:31:54,710
to somebody. They can - you know - have beer.

862
00:31:54,710 --> 00:31:57,410
They can smile. They can joke. That the

863
00:31:57,410 --> 00:31:59,360
work and the project is the place where

864
00:31:59,360 --> 00:32:00,530
people are interested in professional

865
00:32:00,530 --> 00:32:04,190
growth, not - you know - interpersonal

866
00:32:04,190 --> 00:32:06,500
communications, in my experience. Because,

867
00:32:06,500 --> 00:32:08,510
let's face the reality: The project is

868
00:32:08,510 --> 00:32:11,330
just a temporary group of

869
00:32:11,330 --> 00:32:13,520
people who have nothing in common

870
00:32:13,520 --> 00:32:15,770
basically. In most cases, they're

871
00:32:15,770 --> 00:32:17,030
from different countries, they're from

872
00:32:17,030 --> 00:32:18,170
different cultures, they're from

873
00:32:18,170 --> 00:32:20,420
different - they're different ages, they're

874
00:32:20,420 --> 00:32:22,340
just different people. They come together

875
00:32:22,340 --> 00:32:24,500
because they are Java experts, for

876
00:32:24,500 --> 00:32:26,950
example. And, this is what they have in common.

877
00:32:26,950 --> 00:32:29,470
That's it. They're not friends. They never

878
00:32:29,470 --> 00:32:31,210
been to school together. They don't live

879
00:32:31,210 --> 00:32:32,680
together. They don't have - you know - their

880
00:32:32,680 --> 00:32:34,360
families are not friends. So,

881
00:32:34,360 --> 00:32:36,430
they want to focus on stuff which they

882
00:32:36,430 --> 00:32:37,810
have in common. And, when traditional

883
00:32:37,810 --> 00:32:40,630
projects - you know - trying to make a team

884
00:32:40,630 --> 00:32:42,310
of them, trying to use these

885
00:32:42,310 --> 00:32:44,260
team-building things, and trying to - you

886
00:32:44,260 --> 00:32:45,580
know - say like, "Hey, we're one family",

887
00:32:45,580 --> 00:32:48,160
They're not one family. They don't have

888
00:32:48,160 --> 00:32:50,530
anything in common. Like - you know - I'm

889
00:32:50,530 --> 00:32:55,080
40. He's 16. What kind of family is this?

890
00:32:58,680 --> 00:33:01,330
And, this is really annoying for A

891
00:33:01,330 --> 00:33:02,740
players. This is really annoying for

892
00:33:02,740 --> 00:33:04,630
senior developers to be in the team like

893
00:33:04,630 --> 00:33:06,220
that when people are trying to put them

894
00:33:06,220 --> 00:33:07,780
together and say it's one family, one

895
00:33:07,780 --> 00:33:09,640
team, "we're all together". They['re] just [like,] "Leave

896
00:33:09,640 --> 00:33:10,090
me alone.

897
00:33:10,090 --> 00:33:11,980
Let me code. Let me - [you] know - write the code.

898
00:33:11,980 --> 00:33:13,720
Let me focus on that [which] we are doing."

899
00:33:13,720 --> 00:33:15,160
Exactly that. We're just saying remove

900
00:33:15,160 --> 00:33:18,880
that interpersonal - you know - stop and

901
00:33:18,880 --> 00:33:21,220
focus on this, on the technical, on the

902
00:33:21,220 --> 00:33:22,840
technical problems, and people really

903
00:33:22,840 --> 00:33:25,330
appreciate it - A players. C players, yeah,

904
00:33:25,330 --> 00:33:27,010
they complain. They complain they say

905
00:33:27,010 --> 00:33:28,870
like, "Hey, where's your Slack chat? How can

906
00:33:28,870 --> 00:33:30,460
I call you? I can explain. You let me call

907
00:33:30,460 --> 00:33:31,960
you and I explain everything." We're just

908
00:33:31,960 --> 00:33:33,550
saying, "No, here's the ticket. Submit the

909
00:33:33,550 --> 00:33:35,560
ticket. You don't like the objective code?

910
00:33:35,560 --> 00:33:37,060
You don't like the documentation? Submit

911
00:33:37,060 --> 00:33:38,950
the ticket." And he's like, "Yeah, it's not

912
00:33:38,950 --> 00:33:40,750
really how I'm used to them to do that.

913
00:33:40,750 --> 00:33:42,970
Let me call, it'd be easier for me. It's a

914
00:33:42,970 --> 00:33:44,770
clear indication that he's a C player.

915
00:33:44,770 --> 00:33:47,470
Or she. Moderator: This is always a man. So, let's

916
00:33:47,470 --> 00:33:49,810
move on. If people are only paid for

917
00:33:49,810 --> 00:33:52,300
results, how do you handle periods where

918
00:33:52,300 --> 00:33:54,730
people have less productivity? They have

919
00:33:54,730 --> 00:33:56,650
personal issues, for example. But, they are

920
00:33:56,650 --> 00:34:01,060
temporary. Yegor: It's their own problems. I mean,

921
00:34:01,060 --> 00:34:03,340
yeah. But, we, there... We're adults, I mean.

922
00:34:03,340 --> 00:34:06,580
It's business. So, we, this is... We pay

923
00:34:06,580 --> 00:34:08,890
people. I mean, our rates are high.

924
00:34:08,890 --> 00:34:11,800
That's first. So, we pay people. We work

925
00:34:11,800 --> 00:34:13,330
remotely. And, there are people from

926
00:34:13,330 --> 00:34:14,710
mostly from not really expensive

927
00:34:14,710 --> 00:34:17,470
countries. And, we pay like 40 bucks an

928
00:34:17,470 --> 00:34:19,600
hour, 45 dollars an hour. It's pretty high

929
00:34:19,600 --> 00:34:21,970
rate. So, we allow them to make money

930
00:34:21,970 --> 00:34:24,700
while they are productive,

931
00:34:24,700 --> 00:34:27,340
while they have time. So, they are - you

932
00:34:27,340 --> 00:34:28,960
know - they should be smart enough. They

933
00:34:28,960 --> 00:34:31,540
should be grown-up people to manage

934
00:34:31,540 --> 00:34:33,370
their own finance, to manage their own

935
00:34:33,370 --> 00:34:35,530
money. So, save something for the time

936
00:34:35,530 --> 00:34:37,750
which for your vacation or whatever. So,

937
00:34:37,750 --> 00:34:38,980
we're not trying to - you know - we're

938
00:34:38,980 --> 00:34:40,090
trying to respect our

939
00:34:40,090 --> 00:34:41,890
people with - our programmers - we're trying

940
00:34:41,890 --> 00:34:44,080
to treat them as - you know - as equal.

941
00:34:44,080 --> 00:34:46,150
They're not kids. They're not

942
00:34:46,150 --> 00:34:47,470
children. So, they don't need to be

943
00:34:47,470 --> 00:34:49,180
patronized like in traditional teams.

944
00:34:49,180 --> 00:34:50,350
Like, yeah, we're gonna cover your

945
00:34:50,350 --> 00:34:53,110
vacation, just sit in the office. It, again,

946
00:34:53,110 --> 00:34:54,850
it doesn't motivate A players. It

947
00:34:54,850 --> 00:34:56,650
motivates C players. With those, like, hey,

948
00:34:56,650 --> 00:34:58,570
I'm productive now because it's luck.

949
00:34:58,570 --> 00:35:00,310
Maybe tomorrow I will stop being

950
00:35:00,310 --> 00:35:01,750
productive because, who knows. What's

951
00:35:01,750 --> 00:35:03,130
gonna happen? They're afraid of that. And,

952
00:35:03,130 --> 00:35:04,590
that's why they want to be patronized,

953
00:35:04,590 --> 00:35:06,760
protected. They want somebody to take

954
00:35:06,760 --> 00:35:08,980
care of them. We just let them take

955
00:35:08,980 --> 00:35:10,630
care of themselves. So, we treat them

956
00:35:10,630 --> 00:35:11,950
equally. This is the project. We have

957
00:35:11,950 --> 00:35:13,450
money here. You produce something. You

958
00:35:13,450 --> 00:35:14,770
give the results. We transfer the money

959
00:35:14,770 --> 00:35:17,050
from here to here. And then, we're - you

960
00:35:17,050 --> 00:35:20,950
know - just go our own ways. So, we don't

961
00:35:20,950 --> 00:35:22,990
have any vacations, paid vacations,

962
00:35:22,990 --> 00:35:25,180
anything. Moderator: Who is responsible for the

963
00:35:25,180 --> 00:35:27,700
final production result? Yegor: Well, for the

964
00:35:27,700 --> 00:35:28,930
final productions result, is the

965
00:35:28,930 --> 00:35:30,820
architect. So, technically, we have an

966
00:35:30,820 --> 00:35:32,410
architect, like I said, in any project. And,

967
00:35:32,410 --> 00:35:34,240
that person is responsible for final

968
00:35:34,240 --> 00:35:37,330
production result and responsible. We

969
00:35:37,330 --> 00:35:38,820
don't like the word responsible, because,

970
00:35:38,820 --> 00:35:41,950
responsibility means some - you know - means

971
00:35:41,950 --> 00:35:43,570
that you're gonna lose something if

972
00:35:43,570 --> 00:35:45,550
something happens. In our case, it doesn't

973
00:35:45,550 --> 00:35:47,860
happen. So, in our case, if the project

974
00:35:47,860 --> 00:35:50,500
doesn't work in the end, then, you know,

975
00:35:50,500 --> 00:35:52,570
that's it. So, there is no - you know -

976
00:35:52,570 --> 00:35:54,340
financial responsibility of that person.

977
00:35:54,340 --> 00:35:55,600
We're not gonna say like, "Hey, now you're

978
00:35:55,600 --> 00:35:56,890
gonna pay us something because you were

979
00:35:56,890 --> 00:35:58,930
the architect for a month and you didn't

980
00:35:58,930 --> 00:36:01,660
produce anything." So, the word "responsible",

981
00:36:01,660 --> 00:36:03,640
in this case, we don't like to use, even

982
00:36:03,640 --> 00:36:05,350
with the customer. So, when customers come

983
00:36:05,350 --> 00:36:07,750
to us and say like, "Hey, are you gonna be

984
00:36:07,750 --> 00:36:10,150
responsible? Is [it] gonna work?" We say, "No, we

985
00:36:10,150 --> 00:36:11,620
can't guarantee that the software will

986
00:36:11,620 --> 00:36:13,570
work. All we can guarantee [is] that the way

987
00:36:13,570 --> 00:36:16,150
we manage the development is effective."

988
00:36:16,150 --> 00:36:18,250
It's like it's, I'm always giving this

989
00:36:18,250 --> 00:36:20,440
example with the taxi driver. You go into

990
00:36:20,440 --> 00:36:22,270
the taxi and you ask, "Can you guarantee

991
00:36:22,270 --> 00:36:24,640
that you will take me to the airport?" If

992
00:36:24,640 --> 00:36:26,350
the taxi driver is professional enough,

993
00:36:26,350 --> 00:36:28,330
he will say like, "No, I can't guarantee

994
00:36:28,330 --> 00:36:30,310
you that, because, anything may happen. The

995
00:36:30,310 --> 00:36:31,750
car may break. There's gonna be a

996
00:36:31,750 --> 00:36:33,520
roadblock, whatever. Anything can happen.

997
00:36:33,520 --> 00:36:35,140
All I can guarantee you [is] that I have a

998
00:36:35,140 --> 00:36:36,850
GPS, I have a map, I have a driving

999
00:36:36,850 --> 00:36:38,740
license, I have the gas in my car.

1000
00:36:38,740 --> 00:36:41,290
Let's go. Let's go and see what happens.

1001
00:36:41,290 --> 00:36:43,300
So, we are together in this

1002
00:36:43,300 --> 00:36:44,560
journey, you know? The same should happen

1003
00:36:44,560 --> 00:36:46,990
in software development. Any attempt to

1004
00:36:46,990 --> 00:36:49,180
put responsibility on

1005
00:36:49,180 --> 00:36:52,480
on the team like say, "Hey guys. Whatever

1006
00:36:52,480 --> 00:36:53,530
happens, you have to

1007
00:36:53,530 --> 00:36:55,870
make it work. It's just against the idea

1008
00:36:55,870 --> 00:36:57,970
of software development. It's wrong. You

1009
00:36:57,970 --> 00:36:59,680
shouldn't. Well, if you [are] a

1010
00:36:59,680 --> 00:37:01,000
customer, you shouldn't do it like that.

1011
00:37:01,000 --> 00:37:03,340
You should check how your

1012
00:37:03,340 --> 00:37:05,290
team works. How effective are the

1013
00:37:05,290 --> 00:37:07,510
mechanisms and methods they use to

1014
00:37:07,510 --> 00:37:09,250
organize their work? And then, see what

1015
00:37:09,250 --> 00:37:09,760
happens.

1016
00:37:09,760 --> 00:37:12,010
And also, it's also related to the

1017
00:37:12,010 --> 00:37:13,600
estimate. So, sometimes people say like

1018
00:37:13,600 --> 00:37:15,130
"Can you estimate the project? How much

1019
00:37:15,130 --> 00:37:17,980
will it cost?" We can't estimate, but, we

1020
00:37:17,980 --> 00:37:20,080
never put that in the contract [that] we can't

1021
00:37:20,080 --> 00:37:21,490
give you some number and say like, "We

1022
00:37:21,490 --> 00:37:22,900
estimate it's gonna cost a hundred

1023
00:37:22,900 --> 00:37:25,090
thousand dollars." Yes, that's an estimate.

1024
00:37:25,090 --> 00:37:28,930
But, it may be a million easy. So, we just

1025
00:37:28,930 --> 00:37:31,390
gonna see what happens. We will

1026
00:37:31,390 --> 00:37:33,610
drive. We're in the taxi. We will drive.

1027
00:37:33,610 --> 00:37:36,310
But, if on that way straight there, if

1028
00:37:36,310 --> 00:37:40,150
there will be some - you know - detours and

1029
00:37:40,150 --> 00:37:41,980
some roadblocks, then we will make

1030
00:37:41,980 --> 00:37:43,870
detours. And, it will cost you more and

1031
00:37:43,870 --> 00:37:46,780
more and more. The only thing we can

1032
00:37:46,780 --> 00:37:48,340
guarantee is that we're not gonna cheat.

1033
00:37:48,340 --> 00:37:49,810
We're not gonna - you know - make extra

1034
00:37:49,810 --> 00:37:51,790
detours. That's what you can blame

1035
00:37:51,790 --> 00:37:53,500
us for. But, don't blame us for - you know -

1036
00:37:53,500 --> 00:37:55,750
for the road being longer than you were

1037
00:37:55,750 --> 00:37:58,480
expecting. That's the philosophy. I mean,

1038
00:37:58,480 --> 00:37:59,980
not every customer will like that, but...

1039
00:37:59,980 --> 00:38:02,380
Moderator: Okay, let's move on. How do you know that

1040
00:38:02,380 --> 00:38:04,540
your developers didn't form up hidden

1041
00:38:04,540 --> 00:38:08,680
secret chat groups? [Yegor and audience laugh] How do you know that

1042
00:38:08,680 --> 00:38:11,560
your developers does not form up a

1043
00:38:11,560 --> 00:38:13,780
secret chat groups? Yegor: Well, yeah. We thought

1044
00:38:13,780 --> 00:38:15,670
about that. And - you know - when they start

1045
00:38:15,670 --> 00:38:16,810
working, they - sometimes they try.

1046
00:38:16,810 --> 00:38:23,260
Sometimes they try. [audience laughs] I had - it's a real

1047
00:38:23,260 --> 00:38:25,570
story. I had a story with the customer. So,

1048
00:38:25,570 --> 00:38:27,310
the customer came to us. And, I explained

1049
00:38:27,310 --> 00:38:29,350
the idea, like how we work, and he

1050
00:38:29,350 --> 00:38:30,880
said like, "I love it. Let's do it your way."

1051
00:38:30,880 --> 00:38:33,370
And, then we started to work. And then, in

1052
00:38:33,370 --> 00:38:35,980
about a month, I realized that I

1053
00:38:35,980 --> 00:38:37,960
don't have enough - you know - tickets. So,

1054
00:38:37,960 --> 00:38:39,520
he's coming to me and giving me some

1055
00:38:39,520 --> 00:38:41,170
information which I can't find hidden in

1056
00:38:41,170 --> 00:38:43,510
tickets. So, he's saying like, "Hey, we're

1057
00:38:43,510 --> 00:38:45,070
using that technology. It's really great."

1058
00:38:45,070 --> 00:38:46,540
and I'm like, "Yeah, but, who made that

1059
00:38:46,540 --> 00:38:47,980
decision?" and he's like, "Yeah, that

1060
00:38:47,980 --> 00:38:49,720
developer made the decision." and I'm

1061
00:38:49,720 --> 00:38:51,610
like, "Where? Where's the ticket?" and he's

1062
00:38:51,610 --> 00:38:53,380
like, "Forget it. No problem. You already

1063
00:38:53,380 --> 00:38:55,060
made it." And then, I realized that he

1064
00:38:55,060 --> 00:38:57,570
created the chat - the customer. And, [he]

1065
00:38:57,570 --> 00:39:00,340
invited developers there. And, they

1066
00:39:00,340 --> 00:39:02,650
started to talk there. And, I'm like - and

1067
00:39:02,650 --> 00:39:04,600
me and my managers were like - watching in

1068
00:39:04,600 --> 00:39:06,100
the tickets. And, the tickets - the amount of

1069
00:39:06,100 --> 00:39:07,120
tickets - go down.

1070
00:39:07,120 --> 00:39:09,340
The conversations stop there. But, the

1071
00:39:09,340 --> 00:39:10,750
information - but the code - is being

1072
00:39:10,750 --> 00:39:12,820
written. So, they write code. They make

1073
00:39:12,820 --> 00:39:14,320
decisions. But, we don't see what's going

1074
00:39:14,320 --> 00:39:16,630
on. And then, I talk to the customer and

1075
00:39:16,630 --> 00:39:18,640
said like, "Do you have some - you know - do

1076
00:39:18,640 --> 00:39:20,410
they talk to them?" And he's like, "Yeah.

1077
00:39:20,410 --> 00:39:22,420
What's wrong with that?" And I'm like, "Dude.

1078
00:39:22,420 --> 00:39:25,270
I told you; that's completely against our -

1079
00:39:25,270 --> 00:39:26,890
you know - our principles, our

1080
00:39:26,890 --> 00:39:28,480
philosophy." And he's like, "All right. All

1081
00:39:28,480 --> 00:39:30,700
right. I promise you. No more talking." And

1082
00:39:30,700 --> 00:39:33,550
then, again, the same. So, he just

1083
00:39:33,550 --> 00:39:35,890
opened the box. And, he ruined the whole

1084
00:39:35,890 --> 00:39:37,600
thing. So, in the end, the project failed.

1085
00:39:37,600 --> 00:39:39,880
In the end, it was nowhere, because we

1086
00:39:39,880 --> 00:39:41,260
completely lost control of the situation.

1087
00:39:41,260 --> 00:39:42,970
You know - you understand why, because

1088
00:39:42,970 --> 00:39:44,800
these people started it all. They started

1089
00:39:44,800 --> 00:39:46,780
to make decisions. And then, this -

1090
00:39:46,780 --> 00:39:49,210
we completely lost control. That's it. So,

1091
00:39:49,210 --> 00:39:50,770
the project failed, and quite -

1092
00:39:50,770 --> 00:39:52,570
Yeah. They told him like - this customer - I

1093
00:39:52,570 --> 00:39:53,920
told him like, "Look, I can't be

1094
00:39:53,920 --> 00:39:55,510
responsible anymore for anything because

1095
00:39:55,510 --> 00:39:57,610
you['re] just doing things which

1096
00:39:57,610 --> 00:39:59,530
are completely wrong. You shouldn't do

1097
00:39:59,530 --> 00:40:01,240
that. You shouldn't allow them to have

1098
00:40:01,240 --> 00:40:03,700
that conversation." So, it happened. And,

1099
00:40:03,700 --> 00:40:05,680
second question. Yes. Moderator: So, one more

1100
00:40:05,680 --> 00:40:07,570
question. What stops me from generating

1101
00:40:07,570 --> 00:40:11,020
the bugs [that] I will fix for your money?

1102
00:40:11,020 --> 00:40:13,540
What stops me from generating

1103
00:40:13,540 --> 00:40:16,990
the bugs that will get fixed for your

1104
00:40:16,990 --> 00:40:17,860
money? So, basically -

1105
00:40:17,860 --> 00:40:19,330
Yegor: Okay, yeah. That's a good question. So, what

1106
00:40:19,330 --> 00:40:20,800
stops the developer from generating -

1107
00:40:20,800 --> 00:40:22,810
while writing - the code and introducing

1108
00:40:22,810 --> 00:40:24,970
some bugs which will be fixed

1109
00:40:24,970 --> 00:40:27,190
later. And then, the developer will make

1110
00:40:27,190 --> 00:40:28,510
more money, right? That's the question.

1111
00:40:28,510 --> 00:40:31,270
Well, first of all we assign. But, we

1112
00:40:31,270 --> 00:40:34,120
assign bugs to fix not to the same

1113
00:40:34,120 --> 00:40:35,560
people who introduced them. So, we

1114
00:40:35,560 --> 00:40:37,810
randomly assign tickets. So, let's say

1115
00:40:37,810 --> 00:40:39,490
you're the developer. So, you create a

1116
00:40:39,490 --> 00:40:41,050
piece of code. And, in that piece of code

1117
00:40:41,050 --> 00:40:42,880
there is some bugs intentionally made.

1118
00:40:42,880 --> 00:40:44,860
Okay, you made them intentionally. But,

1119
00:40:44,860 --> 00:40:46,510
then these bugs will be fixed by that

1120
00:40:46,510 --> 00:40:48,370
person, that person, and that person - randomly.

1121
00:40:48,370 --> 00:40:50,680
So, there is no real reason for the

1122
00:40:50,680 --> 00:40:52,510
developer to introduce bugs. Like, there's

1123
00:40:52,510 --> 00:40:55,860
no, I haven't seen that any, like never. So -

1124
00:40:55,860 --> 00:40:59,770
Say again? [listens to audience member speaking inaudibly, then audience laughs] Well, unless they have a secret

1125
00:40:59,770 --> 00:41:02,830
chat. But, *we* assign tickets. So, they

1126
00:41:02,830 --> 00:41:04,570
don't make this decision. We always give

1127
00:41:04,570 --> 00:41:06,400
them tickets. We say like, "Hey, Jeff. This

1128
00:41:06,400 --> 00:41:08,110
is your ticket." You can't choose it. I

1129
00:41:08,110 --> 00:41:09,760
mean, you can say, "No, give me that

1130
00:41:09,760 --> 00:41:12,430
one." It's always randomly only always

1131
00:41:12,430 --> 00:41:13,900
to random people. That's how we actually

1132
00:41:13,900 --> 00:41:15,700
solve the problem with no experts at all.

1133
00:41:15,700 --> 00:41:18,040
So, because every time we give

1134
00:41:18,040 --> 00:41:20,220
bugs and troubles to differ

1135
00:41:20,220 --> 00:41:21,480
people. So, they always have to

1136
00:41:21,480 --> 00:41:23,610
understand the new piece of code. So,

1137
00:41:23,610 --> 00:41:24,900
there [is] always something in the code

1138
00:41:24,900 --> 00:41:26,790
base - something new arrived into the

1139
00:41:26,790 --> 00:41:29,790
person, and it helps. So, I haven't

1140
00:41:29,790 --> 00:41:31,500
seen that. Moderator: Okay, great.

1141
00:41:31,500 --> 00:41:34,700
Another question: So, how do you decide

1142
00:41:34,700 --> 00:41:37,290
how much do you pay for a particular

1143
00:41:37,290 --> 00:41:39,870
task? Yegor: Well, like I said, we always give the

1144
00:41:39,870 --> 00:41:42,240
same. We say in the same budget. Let's say

1145
00:41:42,240 --> 00:41:45,810
one hour, and one hour means multiplied

1146
00:41:45,810 --> 00:41:48,360
by the hourly rate of the developer. So,

1147
00:41:48,360 --> 00:41:50,100
we have like a group of developers. Let's

1148
00:41:50,100 --> 00:41:52,410
say 20 people in the project. And, that

1149
00:41:52,410 --> 00:41:54,180
person is 50 bucks an hour, or that

1150
00:41:54,180 --> 00:41:56,250
person is 30 bucks an hour. So, and then,

1151
00:41:56,250 --> 00:41:58,740
we have a bug for one hour. So, we

1152
00:41:58,740 --> 00:42:00,510
randomly select the person and give it

1153
00:42:00,510 --> 00:42:03,200
to that person. Okay, it's gonna cost $50.

1154
00:42:03,200 --> 00:42:04,880
[audience member speaking inaudibly]

1155
00:42:04,880 --> 00:42:07,410
Yeah, like, I just explained before. So, we just

1156
00:42:07,410 --> 00:42:10,080
always say one hour. We don't know - we

1157
00:42:10,080 --> 00:42:12,150
don't estimate. We don't care. So, we just

1158
00:42:12,150 --> 00:42:13,950
say one hour. Well, in our case, it's half

1159
00:42:13,950 --> 00:42:16,080
an hour. We try it - one hour - but, now we

1160
00:42:16,080 --> 00:42:17,760
use 30 minutes. So, it's always half an

1161
00:42:17,760 --> 00:42:21,540
hour. And, we allow developers to cheat and

1162
00:42:21,540 --> 00:42:24,360
return back the code - return back to fix

1163
00:42:24,360 --> 00:42:27,180
not completely fixed. So, sometimes we

1164
00:42:27,180 --> 00:42:31,350
have bugs which takes an oh, five, seven,

1165
00:42:31,350 --> 00:42:33,660
ten steps. So, let's say, I create - there's

1166
00:42:33,660 --> 00:42:36,420
a bug. I assign it to you. You fix it not

1167
00:42:36,420 --> 00:42:38,160
completely. You just say, "Hey, I just - for

1168
00:42:38,160 --> 00:42:40,500
example - I introduced the unit test which

1169
00:42:40,500 --> 00:42:42,870
which proves that this is not the

1170
00:42:42,870 --> 00:42:43,410
problem."

1171
00:42:43,410 --> 00:42:45,060
But, I don't know what the problem is. So,

1172
00:42:45,060 --> 00:42:47,520
I introduce new unit test. And, I put a

1173
00:42:47,520 --> 00:42:49,200
new marker saying like, "Hey, please

1174
00:42:49,200 --> 00:42:51,240
investigate further, because it looks

1175
00:42:51,240 --> 00:42:52,950
like - here's the new unit test but - it

1176
00:42:52,950 --> 00:42:54,540
doesn't prove that the bug is there." I've

1177
00:42:54,540 --> 00:42:56,400
returned that code back. The new

1178
00:42:56,400 --> 00:42:58,470
developer picks up the marker and tries

1179
00:42:58,470 --> 00:43:00,930
again. Again, it can [be] that the developer

1180
00:43:00,930 --> 00:43:03,060
can't do anything. Again, the new code comes

1181
00:43:03,060 --> 00:43:04,830
into master branch. The new developer is

1182
00:43:04,830 --> 00:43:06,540
assigned. A new developer is assigned. So,

1183
00:43:06,540 --> 00:43:07,890
we can jump back and forth, back and

1184
00:43:07,890 --> 00:43:09,960
forth, on these iterations, sometimes for

1185
00:43:09,960 --> 00:43:11,880
ten times, and eventually that will be

1186
00:43:11,880 --> 00:43:13,920
fixed. So, the total cost of the bug will

1187
00:43:13,920 --> 00:43:15,780
be 10 multiplied by 30 minutes. There's

1188
00:43:15,780 --> 00:43:18,510
gonna be 300 minutes - the total budget.

1189
00:43:18,510 --> 00:43:20,910
But, it's going to be 10 increments. 10

1190
00:43:20,910 --> 00:43:23,070
Increments with 10 people, which is a way

1191
00:43:23,070 --> 00:43:24,570
more manageable situation

1192
00:43:24,570 --> 00:43:27,570
comparing to one person [spending] 300

1193
00:43:27,570 --> 00:43:29,880
minutes. "Hey, here's your task work for

1194
00:43:29,880 --> 00:43:32,060
300 minutes" which means, what, five hours.

1195
00:43:32,060 --> 00:43:33,380
Maybe,

1196
00:43:33,380 --> 00:43:35,120
you will fix it. In our case, it's going to be

1197
00:43:35,120 --> 00:43:37,190
ten people touching that problem. Ten

1198
00:43:37,190 --> 00:43:39,920
people. Eventually, the person number ten

1199
00:43:39,920 --> 00:43:40,910
will fix it.

1200
00:43:40,910 --> 00:43:43,190
Which one is manageable - more manageable?

1201
00:43:43,190 --> 00:43:46,310
Obviously, the - our approach. So, we don't

1202
00:43:46,310 --> 00:43:47,990
estimate. We just always say thirty

1203
00:43:47,990 --> 00:43:49,430
minutes, half an hour, whatever. You do

1204
00:43:49,430 --> 00:43:50,930
half an hour, do something, do something,

1205
00:43:50,930 --> 00:43:53,150
do something. And then, we iterate, iterate,

1206
00:43:53,150 --> 00:43:56,750
iterate. Moderator: Okay, great. And what about "Ops"

1207
00:43:56,750 --> 00:43:58,280
now what about operations and

1208
00:43:58,280 --> 00:44:00,040
infrastructure? How do you pay for that?

1209
00:44:00,040 --> 00:44:04,010
Yegor: Well for DevOps, it's kind of something

1210
00:44:04,010 --> 00:44:06,770
we haven't touched like seriously. So, I'm

1211
00:44:06,770 --> 00:44:08,780
not sure how it's gonna work for DevOps.

1212
00:44:08,780 --> 00:44:10,430
For the current situation with DevOps

1213
00:44:10,430 --> 00:44:12,050
where people actually - you know - configure

1214
00:44:12,050 --> 00:44:14,270
servers and install something on servers.

1215
00:44:14,270 --> 00:44:16,400
So, my answer is, "I don't know."

1216
00:44:16,400 --> 00:44:18,800
We'll try. We haven't touched that area

1217
00:44:18,800 --> 00:44:20,840
yet. Maybe with the immutable servers -

1218
00:44:20,840 --> 00:44:22,250
which i think is gonna be the future -

1219
00:44:22,250 --> 00:44:24,350
when we'll have mutable servers, and then

1220
00:44:24,350 --> 00:44:26,120
the entire infrastructure will be kept

1221
00:44:26,120 --> 00:44:28,130
in the in the repository as a code, and

1222
00:44:28,130 --> 00:44:30,320
everything will happen in the code, and

1223
00:44:30,320 --> 00:44:31,340
then the infrastructure will be

1224
00:44:31,340 --> 00:44:32,870
automatically deployed to production

1225
00:44:32,870 --> 00:44:35,930
like in one-click, then probably the

1226
00:44:35,930 --> 00:44:38,030
DevOps will be equal to developer. And

1227
00:44:38,030 --> 00:44:39,710
then, the problem will be solved. But,

1228
00:44:39,710 --> 00:44:41,330
right now it's more like people with the

1229
00:44:41,330 --> 00:44:45,620
SSH consoles. So, it's difficult to do.

1230
00:44:45,620 --> 00:44:48,170
Moderator: If the server is down, the spacer duty also

1231
00:44:48,170 --> 00:44:50,960
just open a ticket? Yegor: Well, in most cases. Yeah, in

1232
00:44:50,960 --> 00:44:52,820
most cases of the server is down, we also

1233
00:44:52,820 --> 00:44:54,440
open the ticket. We also open the ticket

1234
00:44:54,440 --> 00:44:56,090
and say, "Hey, somebody fixd that." And,

1235
00:44:56,090 --> 00:44:57,800
sometimes - it's a good question -

1236
00:44:57,800 --> 00:44:59,270
sometimes, we need that ticket to be

1237
00:44:59,270 --> 00:45:01,400
fixed now instead of in five hours. And,

1238
00:45:01,400 --> 00:45:04,070
[in] that case, we increase the budget. We just

1239
00:45:04,070 --> 00:45:05,600
say, "It's not 30 minutes. It's two hours."

1240
00:45:05,600 --> 00:45:07,280
Even though we understand it's a work

1241
00:45:07,280 --> 00:45:09,230
for five minutes or 15 minutes. We

1242
00:45:09,230 --> 00:45:11,810
just multiply - we boost - the budget. We say

1243
00:45:11,810 --> 00:45:14,120
like "multiple"

1244
00:45:14,120 --> 00:45:15,920
and it means like the developer will

1245
00:45:15,920 --> 00:45:17,420
pick it up immediately and will try to

1246
00:45:17,420 --> 00:45:19,670
do everything to close because the

1247
00:45:19,670 --> 00:45:21,320
budget is way higher. So, we can - we

1248
00:45:21,320 --> 00:45:23,570
actually motivate by money which is [a]

1249
00:45:23,570 --> 00:45:24,380
really good thing.

1250
00:45:24,380 --> 00:45:26,210
So, most people say that money is not a

1251
00:45:26,210 --> 00:45:27,800
motivator. We completely disagree with

1252
00:45:27,800 --> 00:45:30,110
that. Money is the greatest motivator.

1253
00:45:30,110 --> 00:45:32,240
That is the greatest motivator ever

1254
00:45:32,240 --> 00:45:34,940
invented in human history. We don't

1255
00:45:34,940 --> 00:45:36,260
need anything else. We don't need [these] -

1256
00:45:36,260 --> 00:45:37,940
you know - tennis tables in the office. We

1257
00:45:37,940 --> 00:45:39,290
don't need these coffee machines. We need

1258
00:45:39,290 --> 00:45:41,480
money. Just give money to people in a

1259
00:45:41,480 --> 00:45:43,010
clear way so the developers will

1260
00:45:43,010 --> 00:45:44,750
understand why the money coming there,

1261
00:45:44,750 --> 00:45:46,850
how to increase the amount of money. I

1262
00:45:46,850 --> 00:45:48,830
don't need this soft skills. I don't - I

1263
00:45:48,830 --> 00:45:50,750
don't need to please my manager. I just

1264
00:45:50,750 --> 00:45:53,270
need to close more tickets. This is, I

1265
00:45:53,270 --> 00:45:54,950
think, the way more productive - you know -

1266
00:45:54,950 --> 00:45:56,960
approach to motivating people. There's

1267
00:45:56,960 --> 00:46:00,920
the question here. Maybe, yeah. [audience member speaks inaudibly] Taxes? Well,

1268
00:46:00,920 --> 00:46:03,140
we pay taxes. Programmers pay taxes. So,

1269
00:46:03,140 --> 00:46:04,580
programmers want to send money to -

1270
00:46:04,580 --> 00:46:06,050
programmers, they pay all their own

1271
00:46:06,050 --> 00:46:09,800
taxes, of course. Yeah. Moderator: How much do you pay

1272
00:46:09,800 --> 00:46:12,470
managers and how well for managers? Yegor: We

1273
00:46:12,470 --> 00:46:14,630
pay in the same, well, we also pay by task.

1274
00:46:14,630 --> 00:46:18,380
And so, the more tasks a manager is

1275
00:46:18,380 --> 00:46:20,660
managing, is assigning, closing, assigning,

1276
00:46:20,660 --> 00:46:22,550
closing, the more money the manager gets.

1277
00:46:22,550 --> 00:46:24,260
So, we're always trying to attach

1278
00:46:24,260 --> 00:46:26,720
everybody in the project for results. So,

1279
00:46:26,720 --> 00:46:29,090
the more the project produces - the more

1280
00:46:29,090 --> 00:46:31,640
we deliver to the customer more code - the

1281
00:46:31,640 --> 00:46:33,650
more everybody - you know - gets. And, the

1282
00:46:33,650 --> 00:46:35,120
customers appreciate it. And, we build

1283
00:46:35,120 --> 00:46:37,040
them like that. So, in

1284
00:46:37,040 --> 00:46:40,250
the end of the week, end of

1285
00:46:40,250 --> 00:46:41,720
the month, we just show the customer the

1286
00:46:41,720 --> 00:46:44,300
total amount of tasks we closed, bugs we

1287
00:46:44,300 --> 00:46:46,550
solved, and this is the amount of money the

1288
00:46:46,550 --> 00:46:50,360
customer pays. Yeah. Moderator: Okay. Who is breaking

1289
00:46:50,360 --> 00:46:53,510
up work in chunks? What if the chunk is

1290
00:46:53,510 --> 00:46:55,640
too big? Yegor: Well, I just explained that. I

1291
00:46:55,640 --> 00:46:57,230
explained that. We just - you know - we just

1292
00:46:57,230 --> 00:46:58,760
give the full chunk to [the] developer and

1293
00:46:58,760 --> 00:47:00,410
then we allow [the] developer to cheat. So,

1294
00:47:00,410 --> 00:47:02,750
that's how we break it. Moderator: Do you even have

1295
00:47:02,750 --> 00:47:04,160
permanent employees in the company,

1296
00:47:04,160 --> 00:47:08,090
except yourself? Yegor: Well, we - I'm, in general,

1297
00:47:08,090 --> 00:47:10,160
against permanent employment. I think

1298
00:47:10,160 --> 00:47:11,840
it's a modern form of

1299
00:47:11,840 --> 00:47:12,320
slavery.

1300
00:47:12,320 --> 00:47:15,380
So, this long-term employment is,

1301
00:47:15,380 --> 00:47:17,960
well, it's really good for lazy

1302
00:47:17,960 --> 00:47:19,700
developers. It's really, it's

1303
00:47:19,700 --> 00:47:21,680
perfect. Because, like I said, no matter

1304
00:47:21,680 --> 00:47:23,660
what you do, you get the money. But, it's

1305
00:47:23,660 --> 00:47:25,160
not really good for people who are - you

1306
00:47:25,160 --> 00:47:27,790
know - who are

1307
00:47:27,800 --> 00:47:30,260
self-motivated, who are interested to

1308
00:47:30,260 --> 00:47:32,480
grow, who wants to get pleasure from the

1309
00:47:32,480 --> 00:47:34,190
work. So, this full time employment -

1310
00:47:34,190 --> 00:47:36,230
long-term employment - is really not

1311
00:47:36,230 --> 00:47:37,820
productive. It's everywhere right now. All

1312
00:47:37,820 --> 00:47:40,220
companies pay that. And - you know - it's

1313
00:47:40,220 --> 00:47:42,140
really difficult to fight against that.

1314
00:47:42,140 --> 00:47:43,910
But, I think the future is different. I

1315
00:47:43,910 --> 00:47:45,980
think, in the future, we'll get more money

1316
00:47:45,980 --> 00:47:47,810
per hour, but, we'll get

1317
00:47:47,810 --> 00:47:49,880
paid for results. I think that's the

1318
00:47:49,880 --> 00:47:51,860
future for professional development - not

1319
00:47:51,860 --> 00:47:53,030
what we have right now. What we have

1320
00:47:53,030 --> 00:47:54,770
right now is, like I'm saying, it's modern

1321
00:47:54,770 --> 00:47:58,160
slavery of the 21st century. It's gonna

1322
00:47:58,160 --> 00:47:59,930
go. Moderator: Okay, and the last question. From

1323
00:47:59,930 --> 00:48:02,450
me actually to the audience. Yegor: Yeah. Moderator: How many

1324
00:48:02,450 --> 00:48:07,450
people agree with this approach? Yegor: See? 50%.

1325
00:48:07,450 --> 00:48:10,250
Moderator: Thank you, Yegor. That was awesome. Yegor: Thank

1326
00:48:10,250 --> 00:48:11,760
you very much.

1327
00:48:11,760 --> 00:48:19,730
[Applause]