1
00:00:04,542 --> 00:00:07,546
MIRABAI: Howdy.

2
00:00:07,546 --> 00:00:09,413
So, first off, who am I?

3
00:00:09,413 --> 00:00:11,587
My name is Mirabai Knight, I'm based in New York City,

4
00:00:11,587 --> 00:00:17,373
and I'm a professional stenographer for Deaf and hard of hearing college students and professionals.

5
00:00:17,373 --> 00:00:22,593
About six years ago, I got really frustrated with my proprietary steno software.

6
00:00:22,593 --> 00:00:27,127
It was, like, $4,000, you know, $700 bucks a year for upgrades, really buggy,

7
00:00:27,127 --> 00:00:29,037
bloated, terrible DRM.

8
00:00:29,037 --> 00:00:30,566
It was just driving me nuts, right?

9
00:00:30,566 --> 00:00:36,841
So I decided there needed to be an open source alternative to steno -- professional, realtime,

10
00:00:36,841 --> 00:00:39,877
computer-connected stenography.

11
00:00:39,877 --> 00:00:41,794
So I figured no one else was going to write it,

12
00:00:41,794 --> 00:00:42,948
if I didn't do it myself.

13
00:00:42,948 --> 00:00:48,619
I tried to find a Python tutor, and I was extremely lucky to happen upon Joshua Lifton, PhD,

14
00:00:48,619 --> 00:00:50,553
recently of the MIT Media Lab,

15
00:00:50,553 --> 00:00:55,725
who started out tutoring me in Python, and then got so wrapped up in the project

16
00:00:55,725 --> 00:00:57,207
that he basically came to me and said --

17
00:00:57,207 --> 00:00:58,991
listen, I'll develop this for you.

18
00:00:58,991 --> 00:01:00,331
You can pay me whatever you can afford,

19
00:01:00,331 --> 00:01:02,133
but I want to make this happen.

20
00:01:02,133 --> 00:01:03,710
So he took over the project.

21
00:01:03,710 --> 00:01:05,165
Has been coding it ever since,

22
00:01:05,165 --> 00:01:08,612
and then, when he had to sort of tone down the amount of time he could give to it,

23
00:01:08,612 --> 00:01:10,005
because of his full-time job,

24
00:01:10,005 --> 00:01:12,294
Hesky Fisher, whose day job is at Google,

25
00:01:12,294 --> 00:01:13,743
took over the project.

26
00:01:13,743 --> 00:01:18,546
So I've been just ridiculously fortunate to get these incredible top coders to help me with the project.

27
00:01:18,546 --> 00:01:23,352
Also, we've got about 200 users who weigh in, 
bug testing, giving input.

28
00:01:23,352 --> 00:01:25,085
It's been a really cool project.

29
00:01:25,085 --> 00:01:26,626
It's been around for about three years now,

30
00:01:26,626 --> 00:01:29,026
and it's going strong.

31
00:01:29,026 --> 00:01:30,872
What is stenography?

32
00:01:30,872 --> 00:01:34,798
It's a phonetic-mnemonic text input system.

33
00:01:34,798 --> 00:01:36,897
It's been around since about 1911,

34
00:01:36,897 --> 00:01:40,503
and to be a professional stenographer, like me,

35
00:01:40,503 --> 00:01:43,502
you basically need a speed of 240 words per minute,

36
00:01:43,502 --> 00:01:45,939
with 99.9% accuracy.

37
00:01:45,939 --> 00:01:50,585
That means an error or omission about every thousand words or four pages.

38
00:01:50,585 --> 00:01:54,796
You can see this old-fashioned steno machine obviously doesn't hook up to a computer.

39
00:01:54,796 --> 00:01:56,320
That's from around 1930.

40
00:01:56,320 --> 00:01:58,020
Cost about $50.

41
00:01:58,020 --> 00:02:02,868
This much bigger sort of split, ergonomic steno machine, which is the one I use professionally,

42
00:02:02,868 --> 00:02:05,247
costs $4,000.

43
00:02:05,247 --> 00:02:07,828
But this one down here, which just looks like a regular qwerty keyboard,

44
00:02:07,828 --> 00:02:13,200
except for some mysterious rectangular keys pasted onto it,

45
00:02:13,200 --> 00:02:18,375
costs about $50 for the keyboard and $20 for our laser cut steno keys, which we sell

46
00:02:18,375 --> 00:02:19,543
through the Plover store.

47
00:02:19,543 --> 00:02:26,222
That is the main innovation of Plover, that we're basically cutting the hardware cost from $4,000 down to $50.

48
00:02:26,222 --> 00:02:28,985
But why should you care?

49
00:02:28,985 --> 00:02:29,919
You're all developers.

50
00:02:29,919 --> 00:02:30,453
I'm not.

51
00:02:30,453 --> 00:02:32,690
I'm still trying to learn Python, but it's a hard slog.

52
00:02:32,690 --> 00:02:36,045
And when you think of stenographers, 
you probably think of, you know,

53
00:02:36,045 --> 00:02:40,428
old-fashioned ladies with buns in their hair, 
sitting in a courtroom, going, "La la la la la!"

54
00:02:40,428 --> 00:02:41,497
Why should you care?

55
00:02:41,497 --> 00:02:44,216
It's not interesting, it's not cool, it's not useful.

56
00:02:44,216 --> 00:02:46,834
But you're wrong.

57
00:02:46,834 --> 00:02:49,038
Steno is good for a ridiculous amount of things.

58
00:02:49,038 --> 00:02:51,272
It's incredibly elegant, useful, efficient.

59
00:02:51,272 --> 00:02:55,777
I won't really go into all the details of these six things here.

60
00:02:55,777 --> 00:03:00,424
I wrote a series of essays detailing each one of them.

61
00:03:00,424 --> 00:03:01,749
Speed, wearability, ergonomics --

62
00:03:01,749 --> 00:03:05,053
if you're interested in them, check out the link from my PyCon speakers page.

63
00:03:05,053 --> 00:03:09,296
But I'll go briefly into paying work and accessibility for people with disabilities.

64
00:03:09,296 --> 00:03:13,275
Mostly, I'll focus on the last one, which is fluency and efficiency of text entry,

65
00:03:13,275 --> 00:03:15,230
which is why you guys might care.

66
00:03:15,230 --> 00:03:17,498
So this speech will have sort of a two-pronged approach.

67
00:03:17,498 --> 00:03:21,210
First I'm going to give you the hard sell to try to convince you that you, as a coder,

68
00:03:21,210 --> 00:03:26,132
should learn steno to write code, blog posts, 
emails, novels -- whatever you want to do with it.

69
00:03:26,132 --> 00:03:29,202
That it's incredibly useful, 
and that you personally should learn it.

70
00:03:29,202 --> 00:03:30,573
That might work for some of you.

71
00:03:30,573 --> 00:03:32,073
Probably won't work for all of you.

72
00:03:32,073 --> 00:03:33,678
So at the end of that section,

73
00:03:33,678 --> 00:03:38,581
I'll go on to sort of talking about why you might care about steno being out there in the world in general.

74
00:03:38,581 --> 00:03:40,014
Even if you don't personally learn it,

75
00:03:40,014 --> 00:03:44,687
why it's really important that other people have the opportunity to use it.

76
00:03:44,687 --> 00:03:46,524
This is just the steno alphabet.

77
00:03:46,524 --> 00:03:51,793
I'm not going to go into details, but my slides are up on the page, so you can check this out at your leisure later.

78
00:03:51,793 --> 00:03:55,365
The steno keyboard has 22 keys.

79
00:03:55,365 --> 00:04:00,070
Basically it's a two-dimensional syllabic input system,

80
00:04:00,070 --> 00:04:01,724
as opposed to the qwerty system,

81
00:04:01,724 --> 00:04:05,967
where obviously you take any word, break it down into its component letters,

82
00:04:05,967 --> 00:04:08,440
and then type out each letter one by one.

83
00:04:08,440 --> 00:04:12,219
With steno, you input each syllable in a single stroke,

84
00:04:12,219 --> 00:04:16,453
though sometimes you can actually get more syllables per stroke, depending on the word.

85
00:04:16,453 --> 00:04:19,920
So the left hand handles the initial consonants of the word,

86
00:04:19,920 --> 00:04:21,791
the right hand handles the final consonants,

87
00:04:21,791 --> 00:04:24,058
and the thumbs handle the vowels in the middle.

88
00:04:24,058 --> 00:04:26,764
There's no spacebar.

89
00:04:26,764 --> 00:04:29,999
Each stroke is submitted after all of the keys are released.

90
00:04:29,999 --> 00:04:32,567
And most words can be written phonetically.

91
00:04:32,567 --> 00:04:34,604
One example here is "straps".

92
00:04:34,604 --> 00:04:36,273
You can look on the steno keyboard.

93
00:04:36,273 --> 00:04:38,239
S-T-R-A-P-S.

94
00:04:38,239 --> 00:04:41,517
You hit all of those together in one stroke, 
and it comes out "straps".

95
00:04:41,517 --> 00:04:42,911
Very intuitive.

96
00:04:42,911 --> 00:04:49,049
Now, Plover is unique -- as opposed to all the other proprietary steno software,

97
00:04:49,049 --> 00:04:51,448
in that it is essentially a keyboard emulator.

98
00:04:51,448 --> 00:04:55,090
Most proprietary software is a word processor,

99
00:04:55,090 --> 00:04:59,149
and if you want to interact with the rest of the OS, it's in this very sort of awkward sandboxed system,

100
00:04:59,149 --> 00:05:01,231
with lots of delays and convoluted things.

101
00:05:01,231 --> 00:05:02,695
It's very inconvenient.

102
00:05:02,695 --> 00:05:06,733
But Plover is stripped down and designed just to let you interact with your system,

103
00:05:06,733 --> 00:05:10,811
as if you were using a qwerty keyboard.

104
00:05:10,811 --> 00:05:13,506
So, for an example, if you wanted to do a two-syllable word,

105
00:05:13,506 --> 00:05:15,972
you start out with PAOEU.

106
00:05:15,972 --> 00:05:17,579
P, long I; PAOEU.

107
00:05:17,579 --> 00:05:18,786
And you get the word "pie".

108
00:05:18,786 --> 00:05:20,013
P-I-E.

109
00:05:20,013 --> 00:05:21,415
Then you add on THON,

110
00:05:21,415 --> 00:05:24,719
and suddenly Plover will delete the I and the E of "pie",

111
00:05:24,719 --> 00:05:28,656
and send out Y-T-H-O-N, to get python.

112
00:05:28,656 --> 00:05:31,024
Very simple.

113
00:05:31,024 --> 00:05:32,366
As I mentioned,

114
00:05:32,366 --> 00:05:34,321
there are some words which you can write in a single stroke,

115
00:05:34,321 --> 00:05:38,866
even if they're ten letters and four syllables long,

116
00:05:38,866 --> 00:05:40,032
such as deplorable.

117
00:05:40,032 --> 00:05:42,868
D-P-L-O-R-B-L, dplorbl.

118
00:05:42,868 --> 00:05:47,881
Peripheral, P-R-I-F-R-L, prifrl.

119
00:05:47,881 --> 00:05:49,950
I'm going to -- oh, yes, question?

120
00:05:49,950 --> 00:05:51,780
AUDIENCE MEMBER: Could you go back 
to deplorable for a second?

121
00:05:51,780 --> 00:05:53,150
MIRABAI: Go back to deplorable?

122
00:05:53,150 --> 00:05:53,914
AUDIENCE MEMBER: Yes.

123
00:05:53,914 --> 00:05:57,085
MIRABAI: Are you confused about why there's no D on the left side of the keyboard?

124
00:05:57,085 --> 00:05:59,686
Yes, I guess I should have mentioned that,

125
00:05:59,686 --> 00:06:02,321
since the steno keyboard only has 22 keys,

126
00:06:02,321 --> 00:06:07,333
the letters that don't appear
are produced by specific chords.

127
00:06:07,333 --> 00:06:13,877
So in this case, TK is D, HR is L, and down here, EU is I.

128
00:06:13,877 --> 00:06:20,111
You can see that in more detail if you look at the steno alphabet that I posted earlier.

129
00:06:20,111 --> 00:06:25,152
So I'm going to show you an example of me inputting a very small snippet from Codecademy,

130
00:06:25,152 --> 00:06:28,285
which I'm trying to use to learn Python, with steno.

131
00:06:28,285 --> 00:06:30,148
But just to keep in mind, while you're watching this,

132
00:06:30,148 --> 00:06:34,126
it took 143 keystrokes 
to do the same amount of code in qwerty,

133
00:06:34,126 --> 00:06:39,362
whereas it took 54 keystrokes to do it in steno.

134
00:06:39,362 --> 00:06:42,753
All right, let me rewind here.

135
00:06:42,753 --> 00:06:48,840
Down below, you can see the phonetic equivalent of the steno that I'm writing.

136
00:07:03,084 --> 00:07:05,153
Pretty simple.

137
00:07:05,153 --> 00:07:10,460
(applause)

138
00:07:10,460 --> 00:07:13,459
All right, let me go back here.

139
00:07:13,459 --> 00:07:14,890
Right.

140
00:07:14,890 --> 00:07:20,068
Another big advantage of this two-dimensional 
versus one-dimensional text input system

141
00:07:20,068 --> 00:07:25,006
is that you can have mnemonic hooks for all of your words and commands.

142
00:07:25,006 --> 00:07:26,306
So I'm a Vim user.

143
00:07:26,306 --> 00:07:27,739
AUDIENCE MEMBER: Woo!

144
00:07:27,739 --> 00:07:28,888
MIRABAI: Yeah, Vim!

145
00:07:28,888 --> 00:07:29,551
Woo!

146
00:07:29,551 --> 00:07:35,121
And most of these sort of one-letter commands in Vim are sort of intuitive.

147
00:07:35,121 --> 00:07:36,914
You know, back is go back one word.

148
00:07:36,914 --> 00:07:37,452
That makes sense, right?

149
00:07:37,452 --> 00:07:41,655
But r is used -- say if you want to record a macro, right?

150
00:07:41,655 --> 00:07:43,254
R is used for replace a letter.

151
00:07:43,254 --> 00:07:45,225
M is used for mark my spot.

152
00:07:45,225 --> 00:07:49,290
So you have to have these weird, convoluted things like reQord maQro, with a q.

153
00:07:49,290 --> 00:07:50,496
It doesn't really work.

154
00:07:50,496 --> 00:07:53,734
Because you only have one letter to sort of hang your mnemonic hook on.

155
00:07:53,734 --> 00:07:56,542
Whereas, with steno, 
since you have an entire syllable per stroke,

156
00:07:56,542 --> 00:08:00,089
you can do really complex commands like GLERCH,

157
00:08:00,089 --> 00:08:02,083
for "launch browser and perform a Google search",

158
00:08:02,083 --> 00:08:08,030
or QUARN, short for "quarantine", which -- when I make an error in Python -- er, in Plover --

159
00:08:08,030 --> 00:08:11,659
I can cut that highlighted text and append it to a document that I review later

160
00:08:11,659 --> 00:08:16,496
to help fix bugs in Plover's orthographic system.

161
00:08:16,496 --> 00:08:22,858
Now, I might have convinced some of you that steno is cool and useful and efficient and awesome

162
00:08:22,858 --> 00:08:23,810
and you want to learn it.

163
00:08:23,810 --> 00:08:24,746
I might not have.

164
00:08:24,746 --> 00:08:28,406
But I want to make the larger case that steno is really important to have in the world.

165
00:08:28,406 --> 00:08:30,586
The altruistic argument.

166
00:08:30,586 --> 00:08:33,082
In this country, there are 35 million people with hearing loss,

167
00:08:33,082 --> 00:08:35,648
and about 2% of those know Sign Language,

168
00:08:35,648 --> 00:08:38,987
so 98% can only really be helped by realtime captioning.

169
00:08:38,987 --> 00:08:43,262
There's only 400 certified captioners like me in this country, and most of them are near retirement age.

170
00:08:43,262 --> 00:08:46,902
Plus there are 3 to 8 million people who can't use their voices to speak,

171
00:08:46,902 --> 00:08:51,297
and steno is virtually the only technology that would allow them to converse with people they know

172
00:08:51,297 --> 00:08:55,241
in a truly realtime way, at true conversational speeds,

173
00:08:55,241 --> 00:09:01,781
as opposed to tapping things out in a pictographic system that's very slow and awkward.

174
00:09:01,781 --> 00:09:06,312
Unfortunately, the reason why there are so few stenographers and why so many of them are so old

175
00:09:06,312 --> 00:09:10,951
is because the only way to learn steno these days is through for-profit trade schools

176
00:09:10,951 --> 00:09:14,162
that are really kind of scams, you know?

177
00:09:14,162 --> 00:09:15,755
They'll let anybody in.

178
00:09:15,755 --> 00:09:21,095
They'll cost a huge amount of money in tuition, plus buying the machine, buying the software,

179
00:09:21,095 --> 00:09:24,899
and the national dropout rate is 85%.

180
00:09:24,899 --> 00:09:28,734
Because, honestly, most of the people going to steno school don't have other prospects

181
00:09:28,734 --> 00:09:31,741
and aren't really well suited to learn the system.

182
00:09:31,741 --> 00:09:38,076
Most people can reach about 140 to 160 words per minute in steno in, I would say, three to six months.

183
00:09:38,076 --> 00:09:43,316
But getting that extra additional speed from 160 to 225, which is graduation speed,

184
00:09:43,316 --> 00:09:45,364
can take two to six years.

185
00:09:45,364 --> 00:09:51,024
The problem is that we currently don't have any amateurs or hobbyists or tinkerers,

186
00:09:51,024 --> 00:09:56,773
or anyone who wants to pick up steno for fun, for free, or at least low cost, considering this $50 system that we've got.

187
00:09:56,773 --> 00:10:02,902
It's like if the only way you could learn guitar is if you had to go to music school and buy a jewel encrusted,

188
00:10:02,902 --> 00:10:08,270
professional-level guitar, that you had to go through a four-year program, before you could even use it.

189
00:10:08,270 --> 00:10:11,243
If there's no amateurs, there's no future for the program.

190
00:10:11,243 --> 00:10:12,545
And that's what's happening with steno.

191
00:10:12,545 --> 00:10:13,352
It's disappearing.

192
00:10:13,352 --> 00:10:18,255
People aren't using this technology, because they're locked into this proprietary system.

193
00:10:18,255 --> 00:10:22,527
That's why Plover now is totally free software,

194
00:10:22,527 --> 00:10:26,430
the hardware is really much more affordable than the $4,000 proprietary software (sic),

195
00:10:26,430 --> 00:10:31,999
and our next sort of endeavor is going to be trying to raise money for a video game system

196
00:10:31,999 --> 00:10:37,178
that will teach Plover in an addictive way, kind of retro-style arcade games.

197
00:10:37,178 --> 00:10:43,442
Because gamification is a big buzzword in learning these days, but for steno, it's really, really perfect.

198
00:10:43,442 --> 00:10:45,209
You start slow, you get faster,

199
00:10:45,209 --> 00:10:49,940
you start with simple words, you sort of build up your wordhoard, as you go along.

200
00:10:49,940 --> 00:10:53,415
It's almost like an RPG plus a platformer, combined.

201
00:10:53,415 --> 00:10:55,521
So we're hoping that, as the steno community grows,

202
00:10:55,521 --> 00:10:58,989
we'll be able to get more people in, to help develop this system of video games,

203
00:10:58,989 --> 00:11:03,329
and then the steno community hopefully will just explode.

204
00:11:03,329 --> 00:11:06,563
There's a bunch of resources online.

205
00:11:06,563 --> 00:11:08,976
If you go to ploversteno.org.

206
00:11:08,976 --> 00:11:14,273
User groups, forums, mailing lists, all sorts of stuff.

207
00:11:14,273 --> 00:11:16,474
And you can teach yourself steno on your own, try it out.

208
00:11:16,474 --> 00:11:22,254
And I would also be happy 
to answer any questions at this point.

209
00:11:22,254 --> 00:11:23,480
(applause)

210
00:11:23,480 --> 00:11:28,553
Oh, thank you.

211
00:11:28,553 --> 00:11:29,784
AUDIENCE MEMBER: More demos.

212
00:11:29,784 --> 00:11:30,789
MIRABAI: More demos?

213
00:11:30,789 --> 00:11:32,793
Okay.

214
00:11:32,793 --> 00:11:34,191
AUDIENCE MEMBER: Or questions.

215
00:11:34,191 --> 00:11:36,510
MIRABAI: First I'll ask for questions.

216
00:11:36,510 --> 00:11:38,304
If there are no more questions, I'll give you more demos.

217
00:11:38,304 --> 00:11:41,602
No questions?

218
00:11:41,602 --> 00:11:43,369
AUDIENCE MEMBER: Hi.

219
00:11:43,369 --> 00:11:46,702
I switched to Colemak this year,

220
00:11:46,702 --> 00:11:48,439
which is just a different keyboard layout.

221
00:11:48,439 --> 00:11:50,675
It's more... I don't know.

222
00:11:50,675 --> 00:11:52,582
It's better.

223
00:11:52,582 --> 00:11:54,046
Whatever you want to call it.

224
00:11:54,046 --> 00:11:59,384
And I found that I accidentally locked myself out of everybody else's computers.

225
00:11:59,384 --> 00:12:04,026
Because I can't type on other people's computers anymore.

226
00:12:04,026 --> 00:12:06,057
Which kind of sucks.

227
00:12:06,057 --> 00:12:07,259
I don't know.

228
00:12:07,259 --> 00:12:09,093
The jury's still out.

229
00:12:09,093 --> 00:12:13,863
I wonder if you experience this at all, with steno.

230
00:12:13,863 --> 00:12:17,801
You sit down at a library computer.

231
00:12:17,801 --> 00:12:19,169
MIRABAI: I mean, it is a disadvantage.

232
00:12:19,169 --> 00:12:20,671
It's annoying having to switch back to qwerty.

233
00:12:20,671 --> 00:12:25,207
We're currently looking at possible commercial solutions,

234
00:12:25,207 --> 00:12:26,878
for a low cost, very portable steno machine,

235
00:12:26,878 --> 00:12:30,581
that you can just plug into any computer, and it'll work as a qwerty keyboard.

236
00:12:30,581 --> 00:12:32,984
Sort of emulate a qwerty keyboard, without having to install any software.

237
00:12:32,984 --> 00:12:34,661
Which I think will help.

238
00:12:34,661 --> 00:12:36,253
I mean, it's annoying having to carry a keyboard around,

239
00:12:36,253 --> 00:12:38,497
but that might be a sort of compromise solution.

240
00:12:38,497 --> 00:12:46,733
AUDIENCE MEMBER: Since Colemak is still the same shape as qwerty, it overrode qwerty in my brain.

241
00:12:46,733 --> 00:12:48,309
It replaced it.

242
00:12:48,309 --> 00:12:50,704
Does steno do that to you?

243
00:12:50,704 --> 00:12:53,505
MIRABAI: No, I'm still pretty good with qwerty, yeah.

244
00:12:53,505 --> 00:12:59,714
It's just annoying.

245
00:12:59,714 --> 00:13:00,315
AUDIENCE MEMBER: Hi.

246
00:13:00,315 --> 00:13:06,054
I'm an iPhone 5 user, and I always have a problem with the qwerty keyboard on it.

247
00:13:06,054 --> 00:13:10,230
Really hard to use, to send emails and so forth.

248
00:13:10,230 --> 00:13:17,672
Do you think steno will be a good application for the iPhone 5, for regular users, for the faster typing?

249
00:13:17,672 --> 00:13:21,477
MIRABAI: Well, because it's a two-handed system, it's not too easy to transfer.

250
00:13:21,477 --> 00:13:23,637
I hope eventually there will be a wearable solution,

251
00:13:23,637 --> 00:13:28,344
where basically you have the keyboard, you know, on your pants, and you can just walk around.

252
00:13:28,344 --> 00:13:31,076
I think this really sort of rises and falls

253
00:13:31,076 --> 00:13:34,423
depending on whether Google Glass
takes off or sinks like a stone.

254
00:13:34,423 --> 00:13:36,139
If people start having wearable options,

255
00:13:36,139 --> 00:13:38,279
and then get frustrated with the input methods available,

256
00:13:38,279 --> 00:13:40,591
I think steno could be a really good solution to that,

257
00:13:40,591 --> 00:13:45,048
but if wearable computing is really just relegated back to the '90s, who knows?

258
00:13:45,048 --> 00:13:50,090
I don't think that it's going to be usable for the next few years, unfortunately.

259
00:13:50,090 --> 00:13:51,195
AUDIENCE MEMBER: Thank you very much.

260
00:13:51,195 --> 00:13:53,355
MIRABAI: Sure.

261
00:13:53,355 --> 00:13:57,125
AUDIENCE MEMBER: So as a Vim addict, I'm glad to hear that you are one too,

262
00:13:57,125 --> 00:14:02,096
because Dvorak was always a no-go for me, because the colon's in the wrong place,

263
00:14:02,096 --> 00:14:05,269
and the hj keys are a no-go.

264
00:14:05,269 --> 00:14:07,036
So how do you handle that?

265
00:14:07,036 --> 00:14:10,707
And also maybe you could address how you handle things like parens and square brackets.

266
00:14:10,707 --> 00:14:12,845
Things of that nature.

267
00:14:12,845 --> 00:14:20,883
MIRABAI: Well, let me put this back up again, because I use both parentheses and square brackets in this demo.

268
00:14:20,883 --> 00:14:26,004
Basically, for me, the sort of iconic Python double parentheses with nothing in them

269
00:14:26,004 --> 00:14:27,922
I just defined as PRENS.

270
00:14:27,922 --> 00:14:29,157
I write it out at PRENS.

271
00:14:29,157 --> 00:14:31,322
If I want an open parenthesis, I do PREN,

272
00:14:31,322 --> 00:14:33,262
and if I want a closed one, I do PRENT.

273
00:14:33,262 --> 00:14:35,506
Brackets are BRKT and BR*KT.

274
00:14:35,506 --> 00:14:37,702
It's actually fairly intuitive, once you get used to them.

275
00:14:37,702 --> 00:14:41,705
And the nice thing about steno is that your fingers never stray from the home row,

276
00:14:41,705 --> 00:14:46,339
so you're not doing that sort of tippy-tappy thing with the qwerty or even the Dvorak keyboard.

277
00:14:46,339 --> 00:14:51,345
You just stay there, and you make all the possible combinations that you can with the 22 keys

278
00:14:51,345 --> 00:14:53,179
underneath your fingers.

279
00:14:53,179 --> 00:15:00,123
AUDIENCE MEMBER: So then moving around is also chording now, instead of using -- like, hjkl keys for moving around?

280
00:15:00,123 --> 00:15:03,625
MIRABAI: Yeah, you do them sort of like this.

281
00:15:03,625 --> 00:15:05,358
It's very simple.

282
00:15:05,358 --> 00:15:06,727
AUDIENCE MEMBER: Where does that configuration live?

283
00:15:06,727 --> 00:15:08,331
MIRABAI: Say again?

284
00:15:08,331 --> 00:15:10,238
AUDIENCE MEMBER: Where does that configuration live?

285
00:15:10,238 --> 00:15:13,135
Is that your vimrc, or is that the steno?

286
00:15:13,135 --> 00:15:15,172
MIRABAI: Oh, no, no, I still just use regular Vim mappings.

287
00:15:15,172 --> 00:15:17,273
Like, if I want to go back a word, I still write b.

288
00:15:17,273 --> 00:15:20,921
I just write b on the steno keyboard, which is PW plus asterisk.

289
00:15:20,921 --> 00:15:25,114
(laughter)

290
00:15:25,114 --> 00:15:28,149
AUDIENCE MEMBER: You said it takes two to six years to become a fully trained steno.

291
00:15:28,149 --> 00:15:29,152
MIRABAI: To get really fast, yeah.

292
00:15:29,152 --> 00:15:33,594
AUDIENCE MEMBER: How much time would it take to get faster than you are as a qwerty --

293
00:15:33,594 --> 00:15:38,533
MIRABAI: My estimate is it's probably around two to three months, depending on how much time you put into it,

294
00:15:38,533 --> 00:15:41,564
and how quickly you pick it up, but not that much.

295
00:15:41,564 --> 00:15:42,436
AUDIENCE MEMBER: Thanks.

296
00:15:42,436 --> 00:15:43,503
MIRABAI: Yeah, sure.

297
00:15:43,503 --> 00:15:47,674
AUDIENCE MEMBER: Probably about the same time as learning Dvorak, honestly.

298
00:15:47,674 --> 00:15:50,704
AUDIENCE MEMBER: Do you know anything about the state of the art for languages other than English?

299
00:15:50,704 --> 00:15:55,175
So, for example, I am from Russia, so I would have to type Cyrillics if I were to use this thing.

300
00:15:55,175 --> 00:15:57,576
MIRABAI: Yes, this is about foreign languages in steno.

301
00:15:57,576 --> 00:16:00,589
We actually do have a number of people on the Plover user group from other countries.

302
00:16:00,589 --> 00:16:02,882
There's one from Poland.

303
00:16:02,882 --> 00:16:07,654
I think there's a Russian-American who is interested in using steno in Russian.

304
00:16:07,654 --> 00:16:11,188
AUDIENCE MEMBER: Polish uses the Latin alphabet.

305
00:16:11,188 --> 00:16:13,092
MIRABAI: Yeah, no, I know.

306
00:16:13,092 --> 00:16:17,227
But basically, Plover can output anything in the Unicode character set.

307
00:16:17,227 --> 00:16:21,168
And I think there are stenographic theories for other languages.

308
00:16:21,168 --> 00:16:23,009
Some of them you might have to start from scratch,

309
00:16:23,009 --> 00:16:26,064
but I think for most European languages, there already is an established stenographic theory

310
00:16:26,064 --> 00:16:28,172
that you would just have to import into the Plover system.

311
00:16:28,172 --> 00:16:29,741
So there's a lot of potential there.

312
00:16:29,741 --> 00:16:32,314
Not many people are using it at this point.

313
00:16:32,314 --> 00:16:36,984
AUDIENCE MEMBER: I'm just curious whether you would have enough keys, because Russian has 33 letters rather than 26.

314
00:16:36,984 --> 00:16:38,564
MIRABAI: Quite possibly.

315
00:16:38,564 --> 00:16:40,435
Since it's phonetic.

316
00:16:40,435 --> 00:16:46,068
I believe there is a pre-existing Russian stenographic system, but I'm not completely positive.

317
00:16:46,068 --> 00:16:46,637
AUDIENCE MEMBER: Okay, thanks.

318
00:16:46,637 --> 00:16:48,385
MIRABAI: Sure.

319
00:16:48,385 --> 00:16:49,003
AUDIENCE MEMBER: Hi.

320
00:16:49,003 --> 00:16:53,270
I was just wondering -- so it looked in the picture like the $50 keyboard that you were talking about --

321
00:16:53,270 --> 00:16:54,892
it looks like a normal keyboard.

322
00:16:54,892 --> 00:16:56,412
MIRABAI: Not quite.

323
00:16:56,412 --> 00:16:57,714
AUDIENCE MEMBER: So are there hardware differences?

324
00:16:57,714 --> 00:16:58,447
MIRABAI: There are, yeah.

325
00:16:58,447 --> 00:17:00,522
The main thing is that it needs to have n-key rollover.

326
00:17:00,522 --> 00:17:05,346
Because most keyboards won't accept chords of more than, like, two keys at a time.

327
00:17:05,346 --> 00:17:10,391
So the one we use -- the cheapest n-key rollover keyboard is the Microsoft Sidewinder X4,

328
00:17:10,391 --> 00:17:11,455
which is like a gaming keyboard.

329
00:17:14,055 --> 00:17:15,013
AUDIENCE MEMBER: Sorry, follow up.

330
00:17:15,013 --> 00:17:22,757
So is it possible to get, like, other keyboards other than that $50 one, or are they all more expensive than that?

331
00:17:22,757 --> 00:17:24,059
MIRABAI: They're all more expensive than that.

332
00:17:24,059 --> 00:17:26,979
Yeah, like the Majestouch costs about $120.

333
00:17:26,979 --> 00:17:27,556
Works great.

334
00:17:27,556 --> 00:17:31,000
But nothing cheaper than $50, unfortunately.

335
00:17:31,000 --> 00:17:32,664
AUDIENCE MEMBER: How did you come up with the name?

336
00:17:32,664 --> 00:17:33,722
MIRABAI: Oh, yes.

337
00:17:33,722 --> 00:17:36,506
Actually, let me show you the logo.

338
00:17:36,506 --> 00:17:37,574
If you can see this little guy --

339
00:17:37,574 --> 00:17:39,406
a plover is a sort of a wading bird.

340
00:17:39,406 --> 00:17:42,243
I came up with it for two reasons.

341
00:17:42,243 --> 00:17:47,416
First of all, it's a two syllable, six letter word that you can play in one stroke on the steno keyboard,

342
00:17:47,416 --> 00:17:50,759
and secondly, I like that it's a phonetic system,

343
00:17:50,759 --> 00:17:52,753
but Plover is a non-phonetic word.

344
00:17:52,753 --> 00:17:54,087
Sort of perverse that way.

345
00:17:54,087 --> 00:17:55,625
But you can see on his little wing there --

346
00:17:55,625 --> 00:17:59,970
the dark brown patches spell out Plover.

347
00:17:59,970 --> 00:18:01,630
Because his little wing is a steno keyboard.

348
00:18:01,630 --> 00:18:04,223
So P-L-O-F-R spells out Plover.

349
00:18:04,223 --> 00:18:08,207
(applause)

350
00:18:08,207 --> 00:18:10,505
AUDIENCE MEMBER: You mentioned just a few moments ago

351
00:18:10,505 --> 00:18:13,309
that you type bracket and paren by typing combinations of...

352
00:18:13,309 --> 00:18:16,745
Like, that sound like bracket and paren.

353
00:18:16,745 --> 00:18:19,890
What do you do when you actually want to type the word bracket?

354
00:18:19,890 --> 00:18:21,482
MIRABAI: Oh, well, that's yet another definition.

355
00:18:21,482 --> 00:18:26,062
Basically, for the actual character bracket, I take out the vowel.

356
00:18:26,062 --> 00:18:31,500
So it's sort of BRKT and BR*KT, but for the word "bracket", I just write BRAKT, which is one stroke, again.

357
00:18:31,500 --> 00:18:32,869
AUDIENCE MEMBER: Like everything.

358
00:18:32,869 --> 00:18:34,870
MIRABAI: Yeah!

359
00:18:34,870 --> 00:18:36,814
AUDIENCE MEMBER: So this is really cool.

360
00:18:36,814 --> 00:18:41,446
My question is: You mentioned that you have, like, a personal dictionary.

361
00:18:41,446 --> 00:18:42,479
MIRABAI: Yes.

362
00:18:42,479 --> 00:18:44,212
AUDIENCE MEMBER: How is that formed?

363
00:18:44,212 --> 00:18:45,881
And sort of what's the interface for that?

364
00:18:45,881 --> 00:18:47,149
MIRABAI: Sure.

365
00:18:47,149 --> 00:18:49,013
AUDIENCE MEMBER: And I guess the other question is...

366
00:18:49,013 --> 00:18:53,922
Let's assume that this comes to dominate, like, the hacker world.

367
00:18:53,922 --> 00:18:54,821
MIRABAI: I hope so!

368
00:18:54,821 --> 00:18:57,226
That's what I'm going for!

369
00:18:57,226 --> 00:19:03,037
AUDIENCE MEMBER: How would we do interoperability if we all have personal dictionaries, basically, to use with this?

370
00:19:03,037 --> 00:19:05,948
MIRABAI: Yeah, well, I mean, that goes back to the question of what you have to do

371
00:19:05,948 --> 00:19:07,738
when you're using someone else's computer.

372
00:19:07,738 --> 00:19:10,115
But the dictionary is inherent.

373
00:19:10,115 --> 00:19:11,874
Each stenographer has their own dictionary.

374
00:19:11,874 --> 00:19:16,476
And I was going to record a demo of using the steno keyboard to define a new word,

375
00:19:16,476 --> 00:19:18,047
but it was a little bit clumsy.

376
00:19:18,047 --> 00:19:19,453
It's actually working pretty well now.

377
00:19:19,453 --> 00:19:20,847
It's one of our new features.

378
00:19:20,847 --> 00:19:22,083
It's really useful.

379
00:19:22,083 --> 00:19:24,362
But yeah, you would have to have some way of transporting your dictionary.

380
00:19:24,362 --> 00:19:26,153
But it's, like, a 4-meg text file.

381
00:19:26,153 --> 00:19:31,793
It's not that hard to -- you can use a thumb drive, or if we do get this $200 solution

382
00:19:31,793 --> 00:19:36,451
of just a plug-in keyboard that works with any computer, the dictionary will be on there.

383
00:19:36,451 --> 00:19:40,167
AUDIENCE MEMBER: And to define those words, you, I assume, do another chord

384
00:19:40,167 --> 00:19:43,072
That says, "I'm going to define a new word and here it is?"

385
00:19:43,072 --> 00:19:44,379
MIRABAI: Correct, yep.

386
00:19:44,379 --> 00:19:45,575
Yeah, it's actually pretty seamless.

387
00:19:45,575 --> 00:19:48,210
AUDIENCE MEMBER: Cool!

388
00:19:48,210 --> 00:19:51,311
AUDIENCE MEMBER: Hi, I used to work at a court reporting firm, actually,

389
00:19:51,311 --> 00:19:53,118
so this all sounds too familiar.

390
00:19:53,118 --> 00:19:59,454
And reading the description, this will be great for a realtime court reporting application.

391
00:19:59,454 --> 00:20:03,224
Do you have plans for expanding this in the future, in a commercial way?

392
00:20:03,224 --> 00:20:04,492
MIRABAI: No.

393
00:20:04,492 --> 00:20:06,062
I want Plover to remain open source.

394
00:20:06,062 --> 00:20:09,364
Honestly, I'm not really interested in breaking into the court reporting market,

395
00:20:09,364 --> 00:20:10,705
because I don't do court reporting,

396
00:20:10,705 --> 00:20:12,700
and I find that stuff incredibly tedious.

397
00:20:12,700 --> 00:20:17,673
So the $4,000 proprietary software already handles the court reporting.

398
00:20:17,673 --> 00:20:20,910
But I do, you know, realtime stenography for the Deaf and hard of hearing.

399
00:20:20,910 --> 00:20:26,145
AUDIENCE MEMBER: Cool, thank you.

400
00:20:26,145 --> 00:20:30,865
AUDIENCE MEMBER: At 240 words per minute, you'd think you would be a great candidate for the MUD gaming community.

401
00:20:30,865 --> 00:20:32,720
Have you talked to these people yet?

402
00:20:32,720 --> 00:20:34,324
MIRABAI: No, actually.

403
00:20:34,324 --> 00:20:35,991
That's a really good idea.

404
00:20:35,991 --> 00:20:40,069
Yeah, I'd love for there to be, like, a text-based RPG that teaches you steno.

405
00:20:40,069 --> 00:20:41,611
I think that would be a really good application,

406
00:20:41,611 --> 00:20:45,843
and a lot cheaper to develop than a big graphics extravaganza.

407
00:20:45,843 --> 00:20:49,972
Yeah, right now the typing game community are pretty excited about Plover.

408
00:20:49,972 --> 00:20:57,012
But that's a pretty small niche in the total online gaming world. So we'll see what happens.

409
00:20:57,012 --> 00:20:58,481
Demos?

410
00:20:58,481 --> 00:21:03,017
All right. I got a whole playlist here.

411
00:21:03,017 --> 00:21:04,228
AUDIENCE MEMBER: You went too fast!

412
00:21:04,228 --> 00:21:05,518
MIRABAI: Sorry about that.

413
00:21:05,518 --> 00:21:10,730
(humming)

414
00:21:10,730 --> 00:21:17,565
MIRABAI: See, I gave the exact same presentation in New York City, and I didn't have enough time.

415
00:21:17,565 --> 00:21:22,777
I think I must have just gone on adrenaline this time.

416
00:21:22,777 --> 00:21:24,107
AUDIENCE MEMBER: Sorry, may I ask another quick question?

417
00:21:24,107 --> 00:21:24,738
MIRABAI: Sure!

418
00:21:24,738 --> 00:21:26,108
AUDIENCE MEMBER: So how does it work with the thumbs?

419
00:21:26,108 --> 00:21:30,892
So the vowels are on the thumbs, and you have to press either one or two keys

420
00:21:30,892 --> 00:21:32,615
at the same time with your thumb, right?

421
00:21:32,615 --> 00:21:33,782
MIRABAI: Exactly, yes.

422
00:21:33,782 --> 00:21:35,117
AUDIENCE MEMBER: Okay, cool.

423
00:21:35,117 --> 00:21:36,351
MIRABAI: Yep, that's exactly right.

424
00:21:36,351 --> 00:21:37,927
AUDIENCE MEMBER: It's convenient to press two keys at once?

425
00:21:37,927 --> 00:21:39,189
MIRABAI: It is, yeah.

426
00:21:39,189 --> 00:21:40,390
I mean, the steno keyboard is designed that way.

427
00:21:40,390 --> 00:21:43,229
Basically the footprint of the steno keyboard is very small.

428
00:21:43,229 --> 00:21:46,195
It's basically the size of a typical hand, you know.

429
00:21:46,195 --> 00:21:49,482
AUDIENCE MEMBER: Okay, that makes sense, yeah.

430
00:21:49,482 --> 00:21:53,836
MIRABAI: So here is a short demo of me using the Filco Majestouch that I mentioned earlier,

431
00:21:53,836 --> 00:22:00,743
which is about a $120 qwerty keyboard, to write in Plover, with the corresponding keys that I'm pressing.

432
00:22:00,743 --> 00:22:02,350
I'm not sure how visible that is.

433
00:22:02,350 --> 00:22:04,145
Eh, I should have made it bigger.

434
00:22:04,145 --> 00:22:08,615
Let me give you another one.

435
00:22:08,615 --> 00:22:11,987
Here is the online typing game that I mentioned.

436
00:22:11,987 --> 00:22:16,824
It's called Typeracer.

437
00:22:16,824 --> 00:22:17,975
AUDIENCE MEMBER: Are you the world record holder?

438
00:22:17,975 --> 00:22:19,328
MIRABAI: I'm not, actually.

439
00:22:19,328 --> 00:22:20,484
AUDIENCE MEMBER: Who is?

440
00:22:20,484 --> 00:22:21,567
MIRABAI: I get too nervous.

441
00:22:21,567 --> 00:22:26,033
Well, the guy who's the world record holder on Typeracer is a qwerty typist.

442
00:22:26,033 --> 00:22:26,986
AUDIENCE MEMBER: What?

443
00:22:26,986 --> 00:22:27,905
MIRABAI: I know.

444
00:22:27,905 --> 00:22:29,004
It's ridiculous, right?

445
00:22:29,004 --> 00:22:31,820
Honestly, I get too nervous, and I get sort of hung up on it.

446
00:22:31,820 --> 00:22:32,880
I'm pretty close.

447
00:22:32,880 --> 00:22:34,610
I think I beat him in several races.

448
00:22:34,610 --> 00:22:37,316
But he's just really focused.

449
00:22:37,316 --> 00:22:39,612
(laughter)

450
00:22:39,612 --> 00:22:41,319
But I certainly beat Jumbo here.

451
00:22:41,319 --> 00:22:42,184
Jimbo.

452
00:22:42,184 --> 00:22:44,086
Whoever it was.

453
00:22:44,086 --> 00:22:45,493
(laughter)

454
00:22:45,493 --> 00:22:52,592
And here's another application that I can't really give you the benefit of the voice, particularly.

455
00:22:52,592 --> 00:22:58,835
ESPEAK VOICE: This is a demonstration of how Plover can be used to speak at a normal conversational pace.

456
00:22:58,835 --> 00:23:02,169
You can see that I'm triggering the speech engine manually,

457
00:23:02,169 --> 00:23:05,675
but while the computer is speaking the previous sentence,

458
00:23:05,675 --> 00:23:07,809
I'm able to write out the next one,

459
00:23:07,809 --> 00:23:11,683
resulting in a smooth and even flow of speech.

460
00:23:11,683 --> 00:23:17,092
This demonstration uses eSpeak, which is a free program included with Linux,

461
00:23:17,092 --> 00:23:22,496
but Plover works with all realtime text-to-speech engines,

462
00:23:22,496 --> 00:23:27,968
and can be used to do virtually anything the qwerty keyboard can do.

463
00:23:27,968 --> 00:23:31,412
Why is it so much faster and more efficient than qwerty?

464
00:23:31,412 --> 00:23:36,651
For a steno machine, an entire syllable, word, or phrase...

465
00:23:36,651 --> 00:23:40,011
MIRABAI: Anyway, that's a small example of how people can use...

466
00:23:40,011 --> 00:23:43,717
(applause)

467
00:23:43,717 --> 00:23:47,464
How people can use Plover to speak if they can't speak with their voices,

468
00:23:47,464 --> 00:23:51,155
whether due to surgery, or whether they have congenital malformations, or whatever,

469
00:23:51,155 --> 00:23:53,227
which is a surprising number of people.

470
00:23:53,227 --> 00:23:55,963
And it's very difficult for them to converse at a conversational pace

471
00:23:55,963 --> 00:23:58,732
without a realtime system like this.

472
00:23:58,732 --> 00:23:59,999
Another question?

473
00:23:59,999 --> 00:24:01,969
AUDIENCE MEMBER: How big is the dictionary you use?

474
00:24:01,969 --> 00:24:04,799
MIRABAI: My dictionary is currently about 160,000 words,

475
00:24:04,799 --> 00:24:07,309
though there are dictionaries that are 200,000, you know.

476
00:24:07,309 --> 00:24:08,909
It's a good starter dictionary.

477
00:24:08,909 --> 00:24:11,352
I use my own dictionary as the sort of Plover default dictionary,

478
00:24:11,352 --> 00:24:16,375
which people can then download and alter and add to at their heart's delight.

479
00:24:16,375 --> 00:24:20,629
AUDIENCE MEMBER: I was wondering, if it's not too deep of a hole to fall into,

480
00:24:20,629 --> 00:24:21,986
how...

481
00:24:21,986 --> 00:24:25,298
How do I ask this?

482
00:24:25,298 --> 00:24:29,464
Is it emulating an input source, essentially?

483
00:24:29,464 --> 00:24:32,705
So as far as your computer is concerned, these are the letters that you typed?

484
00:24:32,705 --> 00:24:34,069
MIRABAI: Yes, absolutely.

485
00:24:34,069 --> 00:24:40,627
AUDIENCE MEMBER: Okay, and then it's designed to have different types of output for different systems

486
00:24:40,627 --> 00:24:42,445
that expect things?

487
00:24:42,445 --> 00:24:45,046
MIRABAI: No, it's just a qwerty keyboard emulator.

488
00:24:45,046 --> 00:24:46,548
It sends out --

489
00:24:46,548 --> 00:24:48,550
you write the letter in one stroke,

490
00:24:48,550 --> 00:24:51,883
but then it's the one that does the work of converting those into individual letters,

491
00:24:51,883 --> 00:24:53,921
and then in the example that I showed,

492
00:24:53,921 --> 00:24:55,428
where I wrote PAOEU, and it said pie,

493
00:24:55,428 --> 00:24:57,089
and then I wrote THON,

494
00:24:57,089 --> 00:24:58,297
it deleted "ie",

495
00:24:58,297 --> 00:25:00,194
and then sent out "thon".

496
00:25:00,194 --> 00:25:02,396
So it is basically using delete tokens and send tokens.

497
00:25:02,396 --> 00:25:05,332
AUDIENCE MEMBER: Okay, so it is actually piping out a couple of backspaces, and then...

498
00:25:05,332 --> 00:25:08,378
MIRABAI: Exactly, yeah, so it's really transferable to just about every program.

499
00:25:08,378 --> 00:25:09,603
AUDIENCE MEMBER: Cool.

500
00:25:09,603 --> 00:25:12,103
MIRABAI: Yeah.

501
00:25:12,103 --> 00:25:14,794
AUDIENCE MEMBER: How would you compare qwerty to steno for hand strain?

502
00:25:14,794 --> 00:25:17,176
Because it looks so elegant and just, like, gentle.

503
00:25:17,176 --> 00:25:17,910
MIRABAI: Yes, absolutely.

504
00:25:17,910 --> 00:25:19,446
It's something I didn't go into too much,

505
00:25:19,446 --> 00:25:21,317
but the ergonomic argument is a really powerful one.

506
00:25:21,317 --> 00:25:23,217
I started out as a qwerty transcriptionist,

507
00:25:23,217 --> 00:25:24,984
and I could type around 110 words per minute,

508
00:25:24,984 --> 00:25:27,297
but after doing that for eight hours a day,

509
00:25:27,297 --> 00:25:29,256
my wrists and my hands were just killing me.

510
00:25:29,256 --> 00:25:29,757
And I was worried --

511
00:25:29,757 --> 00:25:30,991
at the time, I was in steno school,

512
00:25:30,991 --> 00:25:34,496
and I was worried that I would just jeopardize my entire career by, you know, burning out my hands.

513
00:25:34,496 --> 00:25:37,713
And as soon as I switched to steno, it was miraculous.

514
00:25:37,713 --> 00:25:40,214
I mean, I do use the split ergonomic steno keyboard,

515
00:25:40,214 --> 00:25:43,334
but it's much more like playing the piano than playing a typewriter.

516
00:25:43,334 --> 00:25:46,638
You're stroking, and then you can relax, and then you stroke, and then you can relax.

517
00:25:46,638 --> 00:25:49,841
So you're not constantly moving your fingers without any cessation,

518
00:25:49,841 --> 00:25:51,678
and you're not keeping that tension in your fingers.

519
00:25:51,678 --> 00:25:56,458
So I can type, you know, at around 200 words per minute for eight hours without a break,

520
00:25:56,458 --> 00:26:01,695
and I don't feel any pain at all.

521
00:26:01,695 --> 00:26:06,993
AUDIENCE MEMBER: I was just curious if there were any special cases where it breaks, like words like there, their, and they're,

522
00:26:06,993 --> 00:26:08,763
or anagrams, or, like --

523
00:26:08,763 --> 00:26:10,430
MIRABAI: Yeah, homonyms are handled really well.

524
00:26:10,430 --> 00:26:14,102
I mean, you have to sort of either make up or learn the specific rules of differentiating them.

525
00:26:14,102 --> 00:26:18,375
But yeah, I don't have any issues with homonyms, particularly.

526
00:26:18,375 --> 00:26:19,415
AUDIENCE MEMBER: Awesome, thank you.

527
00:26:19,415 --> 00:26:20,607
MIRABAI: Sure.

528
00:26:20,607 --> 00:26:23,046
AUDIENCE MEMBER: Hey, I just thought of another one.

529
00:26:23,046 --> 00:26:25,243
So you know, when you learn a totally new skill,

530
00:26:25,243 --> 00:26:28,289
there's sort of a point, a ramp-up period,

531
00:26:28,289 --> 00:26:30,757
where you have to devote yourself and be disciplined to learn the skill,

532
00:26:30,757 --> 00:26:32,579
and there's a point, like, where you're hooked.

533
00:26:32,579 --> 00:26:34,721
And from that point on, it's all coasting.

534
00:26:34,721 --> 00:26:36,213
It's like -- okay, you want to learn more of the language.

535
00:26:36,213 --> 00:26:37,966
You want to learn more of the vocabulary.

536
00:26:37,966 --> 00:26:43,048
So if you were speaking to a person who's, like, grown up on qwerty, like, what does it take to get to the --

537
00:26:43,048 --> 00:26:44,932
okay, now you're obsessed.

538
00:26:44,932 --> 00:26:47,167
Like, you've got to learn how to do this -- point?

539
00:26:47,167 --> 00:26:48,005
MIRABAI: Yeah, that's a really good question.

540
00:26:48,005 --> 00:26:49,230
And it is quite variable.

541
00:26:49,230 --> 00:26:51,266
There is a fairly steep initial learning curve.

542
00:26:51,266 --> 00:26:56,504
You can type really simple phonetic words like cat and straps once you learn the chording keyboard,

543
00:26:56,504 --> 00:26:59,707
but there's a certain point where the frustration sets in,

544
00:26:59,707 --> 00:27:01,442
and you're writing words that you think are phonetic,

545
00:27:01,442 --> 00:27:03,506
and they don't necessarily translate the way you want them to.

546
00:27:03,506 --> 00:27:06,179
We're definitely working to make that much easier.

547
00:27:06,179 --> 00:27:08,167
You know, having sort of a prompting system,

548
00:27:08,167 --> 00:27:12,854
that if you type out a word letter by letter, it'll show you the equivalent steno chord for it.

549
00:27:12,854 --> 00:27:18,459
Or, you know, the video game, I think, will really help people to sort of just put in concentrated

550
00:27:18,459 --> 00:27:23,341
sort of flow-state work into it, and then suddenly wake up and be like --

551
00:27:23,341 --> 00:27:27,102
I've been doing this for two hours, and now I can actually write this email without ever getting frustrated

552
00:27:27,102 --> 00:27:30,439
or having to stop or having to look up a word specifically.

553
00:27:30,439 --> 00:27:33,042
So right now, Plover is not particularly user friendly,

554
00:27:33,042 --> 00:27:37,482
but I think it will become more so.

555
00:27:37,482 --> 00:27:41,679
AUDIENCE MEMBER: How do keys such as control, alt, meta, and shift work?

556
00:27:41,679 --> 00:27:43,690
MIRABAI: Again, they're just defined in the dictionary.

557
00:27:43,690 --> 00:27:48,006
We have a special command syntax, so alt is... ALT.

558
00:27:48,006 --> 00:27:49,965
AUDIENCE MEMBER: Sweet.

559
00:27:49,965 --> 00:27:51,256
MIRABAI: Yeah.

560
00:27:51,256 --> 00:27:56,061
AUDIENCE MEMBER: How you do combine, say, control and shift?

561
00:27:56,061 --> 00:27:57,865
Is that, again... Alt-shift?

562
00:27:57,865 --> 00:28:00,100
MIRABAI: Well, I don't believe that right now

563
00:28:00,100 --> 00:28:07,921
we're able to have a sort of "hold down shift while I write the next word", but you can sort of nest the commands

564
00:28:07,921 --> 00:28:09,696
in a single dictionary definition.

565
00:28:09,696 --> 00:28:12,667
So for me, alt-tab is TABLT.

566
00:28:12,667 --> 00:28:14,606
So that'll just do alt-tab.

567
00:28:14,606 --> 00:28:18,097
But I think eventually, we do want to implement sort of a stickykey situation,

568
00:28:18,097 --> 00:28:22,109
where if you write one, you'll know that the next one that's written will be combined with the previous.

569
00:28:22,109 --> 00:28:22,800
AUDIENCE MEMBER: Thank you.

570
00:28:22,800 --> 00:28:25,111
MIRABAI: Sure.

571
00:28:25,111 --> 00:28:32,446
AUDIENCE MEMBER: Okay, I saw your demo, where there was stenographic input, and the output

572
00:28:32,446 --> 00:28:37,817
was valid Python chords or functions, and so on.

573
00:28:37,817 --> 00:28:40,217
I didn't understand how that works.

574
00:28:40,217 --> 00:28:49,726
Because -- are you using pre-canned special things that you trigger, or are you really freely programming

575
00:28:49,726 --> 00:28:55,034
with brackets and braces and curly things and all the punctuation?

576
00:28:55,034 --> 00:28:58,612
MIRABAI: Well, one thing that's really great about Python, as you guys know, is that it's very readable.

577
00:28:58,612 --> 00:29:04,876
It doesn't use that much specific syntax or punctuation that isn't common in English speech.

578
00:29:04,876 --> 00:29:08,190
So, you know, you can see a lot of these things are just simple English words.

579
00:29:08,190 --> 00:29:09,980
Original is ORJ.

580
00:29:09,980 --> 00:29:11,114
You know, lower is LOER.

581
00:29:11,114 --> 00:29:12,716
WORD, FIRST, PRINT, FIRST.

582
00:29:12,716 --> 00:29:19,626
All of these commands -- you know, I did define raw input as RIPT, which was raw_input("

583
00:29:19,626 --> 00:29:22,363
so that's a single stroke for that specific Python command,

584
00:29:22,363 --> 00:29:24,530
but that's something that you'll use all the time,

585
00:29:24,530 --> 00:29:27,032
so it's worth defining it once,

586
00:29:27,032 --> 00:29:28,565
and then being able to do it in a single stroke.

587
00:29:28,565 --> 00:29:31,940
Similarly for variables, if you come up with a specific variable

588
00:29:31,940 --> 00:29:34,480
that you know you're going to be writing in your program many times over,

589
00:29:34,480 --> 00:29:38,243
you can just take a second, define it as a one stroke, mnemonic syllable,

590
00:29:38,243 --> 00:29:42,684
and then every time you reference that variable in the future, you just have to do that one stroke to get it.

591
00:29:42,684 --> 00:29:43,816
AUDIENCE MEMBER: Okay.

592
00:29:43,816 --> 00:29:46,603
I was just wondering whether I can write a new program,

593
00:29:46,603 --> 00:29:48,721
which is not yet defined,

594
00:29:48,721 --> 00:29:54,094
and I have certain identifiers which are new, and all of that.

595
00:29:54,094 --> 00:29:56,462
MIRABAI: Oh, absolutely, yes.

596
00:29:56,462 --> 00:29:57,364
AUDIENCE MEMBER: Okay, I'll try it.

597
00:29:57,364 --> 00:29:58,766
Thank you.

598
00:29:58,766 --> 00:29:59,800
MIRABAI: Yes, it's pretty simple.

599
00:29:59,800 --> 00:30:03,269
Anything -- you know, again, basically you can output anything that the qwerty keyboard can.

600
00:30:03,269 --> 00:30:09,942
So you just have to know what strokes correspond to which letters or punctuation or whatever you're using.

601
00:30:09,942 --> 00:30:13,913
AUDIENCE MEMBER: Okay.

602
00:30:13,913 --> 00:30:16,149
AUDIENCE MEMBER: Do you have any issues with screen?

603
00:30:16,149 --> 00:30:17,618
MIRABAI: I'm sorry?

604
00:30:17,618 --> 00:30:18,434
AUDIENCE MEMBER: Screen?

605
00:30:18,434 --> 00:30:22,221
Control-A, control-left-right, screen?

606
00:30:22,221 --> 00:30:28,961
You know, for -- you SSH something, and then you start screen, and then you have multiple screen sessions on it.

607
00:30:28,961 --> 00:30:32,298
MIRABAI: That hasn't really come up.

608
00:30:32,298 --> 00:30:41,040
I imagine it would be handled similarly to the way one handles it using just a qwerty keyboard.

609
00:30:41,040 --> 00:30:44,810
AUDIENCE MEMBER: But there's a lot of metakey combinations that one uses with screen.

610
00:30:44,810 --> 00:30:50,450
Like, in parallel, where you're doing control-A and something else all at the same time on qwerty.

611
00:30:50,450 --> 00:30:52,991
MIRABAI: You can nest those in any dictionary definition.

612
00:30:52,991 --> 00:30:56,517
AUDIENCE MEMBER: Okay, so you would do, like, two different chords, then?

613
00:30:56,517 --> 00:31:01,294
MIRABAI: Well, I sort of like to see it at a higher level, where if there's a particular command,

614
00:31:01,294 --> 00:31:05,431
you would just assign sort of a mnemonic hook to that command,

615
00:31:05,431 --> 00:31:08,640
and then assign that to whatever metakey combinations are necessary to produce it.

616
00:31:08,640 --> 00:31:10,176
AUDIENCE MEMBER: Cool, thank you.

617
00:31:10,176 --> 00:31:12,939
MIRABAI: Sure.

618
00:31:12,939 --> 00:31:18,586
AUDIENCE MEMBER: I think I know the answer to this question, but there's no way to get this to work on a laptop right now?

619
00:31:18,586 --> 00:31:19,812
MIRABAI: Unfortunately, no.

620
00:31:19,812 --> 00:31:21,435
There are no n-key rollover laptops.

621
00:31:21,435 --> 00:31:25,048
It's really frustrating, because I would pay any amount to get one, but currently, no.

622
00:31:25,048 --> 00:31:26,483
Yeah.

623
00:31:30,107 --> 00:31:31,079
AUDIENCE MEMBER: Hi.

624
00:31:31,079 --> 00:31:34,010
Are you familiar with transcriptionist work?

625
00:31:34,010 --> 00:31:36,477
MIRABAI: Yes, I was a transcriptionist for many years.

626
00:31:36,477 --> 00:31:39,349
AUDIENCE MEMBER: Okay, so how does this compare to, like, their crazy macro systems?

627
00:31:39,349 --> 00:31:40,666
MIRABAI: It's much better.

628
00:31:40,666 --> 00:31:42,015
Much more efficient.

629
00:31:42,015 --> 00:31:44,669
Yeah, with transcriptionists, mostly they use, like, text expansion systems,

630
00:31:44,669 --> 00:31:48,291
where they have to write the first two or three letters of a word, and then it'll sort of expand out,

631
00:31:48,291 --> 00:31:52,762
but with steno, most of those you can do in one stroke, so it's still almost triple the efficiency.

632
00:31:52,762 --> 00:31:53,912
AUDIENCE MEMBER: Wow, thanks.

633
00:31:53,912 --> 00:31:56,299
MIRABAI: Sure.

634
00:31:56,299 --> 00:31:58,835
AUDIENCE MEMBER: So thank you for captioning the Ada Initiative's latest video.

635
00:31:58,835 --> 00:32:00,035
MIRABAI: Absolutely!

636
00:32:00,035 --> 00:32:03,904
AUDIENCE MEMBER: I didn't realize how lucky we were to get one of the 400.

637
00:32:03,904 --> 00:32:10,285
But I just imagine that when you have a very -- you can hear code in your head, when you're writing these things.

638
00:32:10,285 --> 00:32:15,251
I was wondering if you could record a video of you speaking your code, or maybe speak it right now.

639
00:32:15,251 --> 00:32:16,685
MIRABAI: Sure!

640
00:32:16,685 --> 00:32:25,100
So let me start with this one here.

641
00:32:25,100 --> 00:32:29,334
All right, this is a sdration of how Plover can be ufed to speak at a nol kfrbl pace.

642
00:32:29,334 --> 00:32:32,413
You can see tha I'm trirging the speech engine manwael,

643
00:32:32,413 --> 00:32:37,204
but while the kpurt is spaoeg the praoef snens, I'm able to write out the negt one,

644
00:32:37,204 --> 00:32:41,176
rulting in a smooth and aoen flow of speech.

645
00:32:41,176 --> 00:32:42,416
This sdraition uses espeak...

646
00:32:42,416 --> 00:32:44,640
AUDIENCE MEMBER: Can you do it for code?

647
00:32:44,640 --> 00:32:45,584
Your code one?

648
00:32:45,584 --> 00:32:52,224
MIRABAI: Oh my code one, sure, yeah, absolutely.

649
00:32:52,224 --> 00:32:53,358
Yeah, you can see it's captioned down at the bottom.

650
00:32:53,358 --> 00:33:02,002
Orj kwals ript kpa ent ai word kln prent rr f orj pp s afl prens kln rr tab word kwals orj

651
00:33:02,002 --> 00:33:09,117
pp loer prens rr tab firs kwals word brkt 0 br*kt rr tab print firs rr else kln

652
00:33:09,117 --> 00:33:14,424
rr tab print kwgs kpa not ai word tppl kwgs

653
00:33:14,424 --> 00:33:16,981
(applause)

654
00:33:16,981 --> 00:33:22,321
Yeah, that is literally what's going on in my head as I'm writing steno.

655
00:33:22,321 --> 00:33:24,158
So there you go.

656
00:33:24,158 --> 00:33:27,928
MIRABAI: No more questions?

657
00:33:27,928 --> 00:33:30,698
All right, then I guess I will -- oh, is there one?

658
00:33:30,698 --> 00:33:32,311
Yeah, okay.

659
00:33:32,311 --> 00:33:38,239
AUDIENCE MEMBER: So I was just pulling up your website, and seeing that these are sort of acrylic pieces

660
00:33:38,239 --> 00:33:39,807
that I glue on my keyboard, basically.

661
00:33:39,807 --> 00:33:40,841
MIRABAI: Yeah, laser cut acrylic.

662
00:33:40,841 --> 00:33:44,744
AUDIENCE MEMBER: Do you see any prospects for maybe a little more sophisticated hardware improving things?

663
00:33:44,744 --> 00:33:45,946
MIRABAI: Absolutely!

664
00:33:45,946 --> 00:33:48,480
AUDIENCE MEMBER: I mean, a lot of people are now getting hobbyist 3D printers.

665
00:33:48,480 --> 00:33:52,086
Do you imagine people printing out their own steno keyboards or something?

666
00:33:52,086 --> 00:33:55,662
MIRABAI: I was actually talking about a possibility of a Raspberry Pi-based sort of steno kit,

667
00:33:55,662 --> 00:33:58,190
that we could sell for people to assemble,

668
00:33:58,190 --> 00:34:02,396
and as I said, we are also thinking about a commercial venture of just selling sort of low cost,

669
00:34:02,396 --> 00:34:03,832
pre-made steno machines.

670
00:34:03,832 --> 00:34:05,763
So yeah, there's a lot of potential in that.

671
00:34:05,763 --> 00:34:06,836
AUDIENCE MEMBER: From Adafruit!

672
00:34:06,836 --> 00:34:08,003
MIRABAI: From Adafruit?

673
00:34:08,003 --> 00:34:09,807
Maybe!

674
00:34:09,807 --> 00:34:10,437
We'll see what happens.

675
00:34:10,437 --> 00:34:12,670
AUDIENCE MEMBER: So is it just 22 switches, basically, that you would need for one of these?

676
00:34:12,670 --> 00:34:13,605
MIRABAI: Exactly, yeah.

677
00:34:13,605 --> 00:34:14,167
Absolutely.

678
00:34:14,167 --> 00:34:19,079
It's very simple.

679
00:34:19,079 --> 00:34:19,948
All right, so I guess that's it.

680
00:34:19,948 --> 99:59:59,999
Thank you!