The Story of Python and how it took over the world | Python: The Documentary — Transcript

Full transcript

1. 0:00Did you think Python was going to get popular from the beginning or?
2. 0:05[Music]
3. 0:14Python has contributed to making critical thinking a  little more approachable by more people.
4. 0:21You can spend one to two hours,
5. 0:23learning programming fundamentals
6. 0:25and then you can write like 10 to 20 line of code
7. 0:28that do something interesting.
8. 0:29The language, the community, the ecosystem,
9. 0:32the market, the whole thing has had such an  impact on humanity.
10. 0:37Today, I think it's safe to say that almost anywhere there's a computer,
11. 0:40there's probably some Python.
12. 0:42It's literally on Mars.
13. 0:45The impact of AI, I think we're still  seeing what that's going to be.
14. 0:49Python is incredibly important for that.
15. 0:52To fully understand how Python got so big, we have to go back in time.
16. 1:00Maybe not that far back.
17. 1:02Everything started back in Amsterdam in the 80s
18. 1:05during my time at CWI, a Dutch research facility.
19. 1:11The CWI had been behind a lot of major programming language developments.
20. 1:17Algol 60 and Algol 68 for instance.
21. 1:20And Lambert Meertens had tried teaching programming to artists
22. 1:25and discovered things about
23. 1:28programming that are easy, if you're a scientist  or a geek. But for somebody like an artist,
24. 1:34those things weren't obvious because you had to  know something about the computer.
25. 1:39The cause of the frustration was the low level of the languages that were available.
26. 1:44The computers were immensely  expensive and compared with that
27. 1:48a programmer was really really cheap
28. 1:51and so programming languages
29. 1:53were designed with that economic relationship in mind.
30. 1:57It didn't matter if it took a long time to
31. 2:00program as long as it didn't demand too much of  the computer because that's where the cost was.
32. 2:06We looked at each other and said, can't we do a  better job? Can't we design a language that is
33. 2:11easy to learn, easy to teach, but also easy to use.
34. 2:15So that beginners could understand how to program
35. 2:18without having to go into all those messy hardware details.
36. 2:22And that's how the ABC project was born.
37. 2:27One of the members of the team  was Guido van Rossum.
38. 2:31[Music]
39. 2:39I was hired to work on ABC for which  there was a small prototype but every
40. 2:46part of the prototype had to be expanded  to fully working according to the spec.
41. 2:54In 1985, we made the release of ABC. And  then we hit the problem: how to make the
42. 3:01world know that there was ABC?
43. 3:05We did not have a good way of reaching our intended audience.
44. 3:11The web did not exist.
45. 3:12Most people had never heard of a computer or even seen one.
46. 3:17If anybody wanted a copy, they had to send us  a letter and we'd send them a floppy disc.
47. 3:24So I'm afraid ABC reached very few people and  at some point the directors for reason, that
48. 3:30I still don't understand, killed that project.
49. 3:33I had spent, well, 3 and 1/2 years of hard work
50. 3:39on this language and the implementation to make it as as good as we could.
51. 3:44So it was disappointing.
52. 3:46[Music]
53. 3:49Guido was transferred to another  project on a distributed operating system,
54. 3:55the amoeba system.
55. 3:58One of my jobs at that project as the programmer, not the researcher,
56. 4:03was to write, or help write, a large number of user applications.
57. 4:09And I didn't actually think
58. 4:12that for many of those utilities C was a  very good language.
59. 4:18And I thought, well,
60. 4:19oh man, if we just could program ABC instead of  C, each of those utilities would be like half a
61. 4:27page of code and I could write the entire suite  of utilities in a couple of weeks instead of the
62. 4:32years that it looked like it was going to take us.
63. 4:35Except that ABC was so high and abstract,
64. 4:40it wasn't a good language to talk to servers  and file systems and processes and it sort of
65. 4:49the whole operating system thing was abstracted  away from ABC. So I thought, well, there really
66. 4:56ought to be a language that sort of bridges the  gap between C and shell. And we looked at Perl and
67. 5:03we didn't think it was any good as a programming  language. It was about as bad as as Basic,
68. 5:11although in different ways. So, Python was also  in a part built as an alternative to Perl. That's
69. 5:18where I decided, oh well, I'm going to see if  I can make my own programming language.
70. 5:24[Music]
71. 5:28One Christmas holiday, I believe, he decided  to spend his time designing and building a
72. 5:34new programming language for the operating  system based on the principles that he'd
73. 5:39learned while on the ABC project.
74. 5:41It was logical that I would start with ABC as an example.
75. 5:46The most prominent feature that  Python borrowed from ABC is actually the use
76. 5:51of indentation for statement grouping. But it  was also logical that I sort of dropped the
77. 5:59things from ABC that I didn't particularly  like. That's how we ended up with Python.
78. 6:10Okay. I finally was confident enough of  the interpreter was complete enough that
79. 6:18I showed it to Lambert. He showed me a  language he had developed stealthily,
80. 6:24which he called Python, named after Monty  Python. I said look what I made. It was
81. 6:31clear to me that it had some resemblance  to ABC. The command line prompt is even
82. 6:36the same. So Lambert watched me type a few  things and.. Guido was excited about it and he
83. 6:43apparently thought I would be excited as well.  At some point he had seen enough and he said,
84. 6:48"Can I try something?" And he types one very  short line of code and it crashes the interpreter.
85. 6:56And he knew that that it would. The point  escaped me. I didn't know why he thought
86. 7:02this was better than ABC and why he had  done this whole project. I went back and
87. 7:08the next day I had a fix for it. But it was  a pretty crushing experience to be honest.
88. 7:28It must have been at least half a year  later, that I first got acquainted with
89. 7:32Python. So this is the office where Guido  and I were holed up when he created Python.
90. 7:40There were two important users. Sjoerd was  one of them and the other was Jack Jansen.
91. 7:47Guido was sitting over there and I was sitting  over there. If you want to prototype a program,
92. 7:54you sort of write the outlines of the program,  but it doesn't actually work. When you do this
93. 8:01in Python, it actually works. It's readable.  It's very easy to program and the indentation
94. 8:07is totally natural as far as I'm concerned  anyway. Sjoerd and Jack were the most active
95. 8:13in using Python. You find bugs, you find the  things that you want to extend and then telling
96. 8:18me about it because all they had to do was say,  "Hey, Guido." It was the first real interactive
97. 8:26systems programming language. So, you could just  sit and write and run it immediately without any
98. 8:30compilation. So, it was much more fun to use  than the old programming languages, but slower.
99. 8:39I think it was mostly Jack's idea in  the end, while we all worked on it,
100. 8:43to make an open-source release. CWI allowed  Guido to distribute Python to the world as
101. 8:50long as its copyright notice was there. They  would have held back if they had known it
102. 8:55would be such an incredible success, which  is good that they didn't know. Nobody knew.
103. 9:00I didn't either. Because if they had held back,  it would never have become an incredible success.
104. 9:08Then we had to actually physically do  the release which turned out to be an
105. 9:13incredible pain because Usenet was  the only thing we had to release it.
106. 9:22Usenet was a network of, you could say, bulletin  boards. People would write messages. They would
107. 9:30get distributed over the world and then there  were the groups for all the different computer
108. 9:35languages. Usenet had a very strict limit on  the size of a single post. First put your entire
109. 9:44source tree in a tarball. Then compress it and you  have to do another step where you turn the binary
110. 9:52data into ASCII encoding. So you have this huge  file with gobblygook, but it's at least printable
111. 10:01gobblygook. And then you apply another tool that  snips it into pieces. Then there is a script where
112. 10:09if you have downloaded all 21 parts, you can  undo all those operations in reverse. That was
113. 10:16what people put up with. So we did all the work on  our side and plenty of people did all the work on
114. 10:22their side to unpack this because I had apparently  written a very good teaser about what this was.
115. 10:34[Music] Very quickly I started getting feedback  via email or Usenet from people who had done
116. 10:42something exciting with it or who had  found a bug and that just kept coming.
117. 10:50We worked at Johnson Space Center. So our interest  was putting together flight simulations and tools
118. 10:57for the design engineers that would be used  for the shuttle program at the time. I remember
119. 11:05needing a scripting language to control a C++  library project we were working on at the time.
120. 11:14Not too long into that search, I found Guido's  Python distribution at CWI. I was able to FTP
121. 11:21that down, unpack it, compile it, and run it in  a matter of 10 minutes or so. That was my first
122. 11:33impressive experience with Python. 20 years ago, a  computer that cost a million dollars and filled an
123. 11:41entire room had less capability than an 11 pounds  desktop computer today. With costs going down and
124. 11:48capabilities going up, the computer has entered  our lives quickly and unobtrusively. There were
125. 11:53lots of things changing in a major way that made  computing particularly exciting to be in. At that
126. 11:59time, computers suddenly were coming out of the  machine room and landing up on your desk. And
127. 12:04then of course the web happened. Then it all  exploded. After the internet came to Europe,
128. 12:10it was much easier for software to be distributed.  Python very slowly took off. Then there was
129. 12:17a growing Python community which was very  supportive like a big family you could almost say.
130. 12:30Part of the benefit of this is that I don't..  I'm like, there's no strain on my body at all.
131. 12:37I was working on developing a kind of distributed  equipment database and a colleague at the lab,
132. 12:45Michael McLay, knew that I was looking for  what was called a scripting language that was
133. 12:51also sophisticated that would enable you to build  substantial things and not get bogged down in the
134. 12:57ragged edges. Scripting languages were designed  to glue things together, to connect things that
135. 13:04already worked, that already existed. Immediately  on trying it, I was just amazed and thought this
136. 13:10can't be as good as it is. And it was. Mike was  interested in bringing Guido over and Guido was
137. 13:18interested in coming over. 99% of the people that  I got in touch with were in the United States.
138. 13:26So Guido ended up coming and then Mike and Guido  and I wanted to arrange a workshop so people who
139. 13:33were interested in Python could gather and talk  about what they were working on and what they
140. 13:38felt was needed. This is the t-shirt for the first  conference. And so this yeah that that was just
141. 13:47sort of flowy writing and on the on the back it  gave the date of the conference. Spam, spam, spam,
142. 13:55spam. It's such a fantastic song. That workshop  really is for me the beginning of the Python
143. 14:02community. November 1994, Gaithersburg, Maryland  in this windowless government office building at
144. 14:09NIST. That was a great experience. It was about  two days. We had 20 people. I still have the
145. 14:15t-shirt from that workshop. There was just kind  of this feel of, hey, there's something happening.
146. 14:20This will be fun to hang out with the people  that I chat with online, to actually see Guido
147. 14:26in person and actually make decisions. And we knew  that if you impressed him with the feature and it
148. 14:31was an easy thing to install, he would add it to  the language right there so that we can at least
149. 14:35try it out in beta form. There was like a whole  room of people who knew what you were talking
150. 14:39about because they had all studied Python deeply  or almost all. One of the people was Barry Warsaw.
151. 14:51[Music] Around 1994, I joined a company, CNRI,  Corporation for National Research Initiatives,
152. 15:01and we were building these software agents,  little programs that can move around to
153. 15:07different computers on a network, and they pickle  themselves up and move around to another place and
154. 15:12reconstitute themselves and do some more work.  A friend of mine from NIST sent me an email and
155. 15:18said, "Hey, we've got this guy coming over from  the Netherlands and he's going to talk about his
156. 15:22language. Do you want to come?" And so we went  to the workshop and just absolutely fell in love
157. 15:28with of course Python and of course Guido. Even  then, it had the feeling that there was a there
158. 15:36there was something that was going to happen  that there was potential. I remember coming
159. 15:42back to CNRI saying, "Oh, this is great. we're  going to use Python for sure in our project. And
160. 15:48one of our colleagues at CNRI said, "Well, what  do you think? Why don't we just hire Guido?"
161. 16:04CNRI was set up to sort of promote information  infrastructure and foster research, help build
162. 16:11pilot projects, take the results and put  them out into the public so that others
163. 16:16could then leverage it. Part of CNRI's mission  was to cultivate internet based things. That's
164. 16:22why our name Corporation for National Research  Initiatives has the word "initiatives" in there.
165. 16:32We had a need for a language that would be easier  for people to deal with than any other traditional
166. 16:40languages that were around at the time. I ran  across a photograph. There was a party at CNRI,
167. 16:46years ago. It was probably I think 1996.  And it's it's a Guido. Yeah. I used to say
168. 16:56Guido. Many people say Guido, but I said Guido  and I'm sure I don't pronounce it right. Oh,
169. 17:02he was a very stylish person and he was a very  strong fellow. Got introduced to him and it just
170. 17:07seemed like it was a good fit and so I made the  job offer. For me, it was fantastic. There is no
171. 17:13doubt about that. One of the things he insists on  if you want to get him to work for you is that he
172. 17:18has some fraction of his time, preferably as much  as halftime if not more, to work on the language
173. 17:25itself. the fact that they could work full-time  on Python because it turned out that the Knowbots
174. 17:31project was almost more an excuse to to hire  me than I did much for that. I kind of feel
175. 17:38like our work on Python was a little subversive.  You know, we tended to spend a lot more time on
176. 17:44Python than maybe we should have. I went to  CNRI and since I had been a systems person,
177. 17:50I was responsible for setting up python.org. We  should have also grabbed python.com even though
178. 17:56we didn't think of Python as a commercial  enterprise because someone else got it and
179. 18:01uh used it to host a stupid porn site. For years  we had to warn people go to python.org, not .com.
180. 18:13It really was the perfect home, I think, for doing  all of that early Python development - and develop
181. 18:18a lot of the infrastructure around Python. We  had a meeting discussing making some kind of user
182. 18:24group or software organization where the Python  software activity was proposed. We need a little
183. 18:30bit of governance, a little bit of organization,  but we don't want too much because I don't think
184. 18:35any of us really liked bureaucracy. And then the  question came up about what we should call Guido
185. 18:40as part of that. I think it was me who suggested  that Guido be called the Benevolent Dictator and
186. 18:48then Barry suggested Benevolent Dictator for Life  and that became an internet meme or actually a
187. 18:55software meme. Benevolent Dictator for Life was..  you can find it on Wikipedia. Ultimately, you
188. 19:05know, Guido was the inventor of the language and  the final decider about yes, this is going to make
189. 19:11it in or this is not going to make it in. I think  I developed my own style during those first years.
190. 19:19All ideas are welcome, but I will choose which  ideas I believe are right to add to the language.
191. 19:27There's dozens and dozens of aspects of  Python today where somebody had a vision for,
192. 19:36if you just add this to Python, look at  all these amazing things that I can do.
193. 19:45[Music]
194. 19:48I'm one of the many people that came to Python,  not as a programmer, but as a scientist. Python
195. 19:53wasn't the first language I used for science. I  was doing satellite data processing and I used
196. 19:57Perl to do it and then a year later I would come  back and try to figure out what I'd done and I
197. 20:01had no idea. Python had the opposite experience. I  first started using Python in 1997. I just used it
198. 20:07to do some scripting, do some data processing with  some medical imaging data and I was like this is
199. 20:12nice. This is pretty cool. A year later, so 1998,  I came back to look at the code I'd written and I
200. 20:18could read it. I understood what I did. And that  was kind of for me the convincing data that said,
201. 20:23"Oh, I need to dig in on this language."  And there were a few really key people at
202. 20:28that time who were publishing or talking about  how to use Python for science. Not very many,
203. 20:32but a few. Konrad Hinsen was one of them, I  remember, and David Ascher and Jim Hugunin.
204. 20:38There's been a number of stages in the life  cycle of Python, when it just grew from a a core
205. 20:45language capability to a platform with a number of  different specialty modules and packages that made
206. 20:52it very very useful for certain user communities.  The first one was probably the numeric community,
207. 20:59the science and engineering folks that wanted  high performance matrix operations to be done.
208. 21:04Even though Guido wasn't a scientist himself,  wasn't necessarily deeply concerned with the
209. 21:09same things that we were concerned with as  scientists, he was open to the suggestions.
210. 21:16I think one of the things that people don't  appreciate about the Python language as much
211. 21:19now that it's so popular and mainstream  is how important it was that Guido was
212. 21:25open to new ideas and open to the needs of  people who are not traditional programmers.
213. 21:32I think he had a very clear-eyed quiet focus  on understanding some things about the future
214. 21:41that he built into the language. It was uniquely  suited to its environment. One of the reasons I
215. 21:46used Python and I think a lot of people used  Python at the time is that in the early 2000s,
216. 21:50the world was almost divided into here's an  open source ecosystem or you can go and use
217. 21:55Java which is actually a paid thing. I  was using MATLAB. I really didn't like
218. 22:00the fact that when I wrote code and I  wanted to share that code with others,
219. 22:02I was essentially telling people they had to go  buy a license for a software package before they
220. 22:07could even look at my or use my code. You  know, these are the days where you have to
221. 22:10like buy the developer tools or spend a couple  hundred bucks to get Visual Studio 6.0 and then
222. 22:16Python and then all of these libraries were open  source and then could be adapted, could be used
223. 22:21in a commercial project. I was sort of early  on convinced by community members that Python
224. 22:27being open- source and the particular way it was  open source was very important so that people
225. 22:33would feel comfortable using Python to make great  things that they would then be able to to sell.
226. 22:51[Music] In 1998, Blender was published on  the internet. The software was free and
227. 22:58then you could buy some things around it.  We were restricted to use open source. So,
228. 23:04you didn't have a lot of options. We were doing a  call on our website like what is the most popular
229. 23:10scripting language that we should add in Blender?  Perl was really big and Python was upcoming and
230. 23:18there were some others. And it was interesting  to see that everybody who was advising us to
231. 23:24use Perl as scripting language but they only  advertised how great Perl is. And the people
232. 23:32who came more from the Python background who  said well I think Python is a better choice.
233. 23:38They came with a balanced opinion. And they  said okay Perl will give you this and this
234. 23:43and this and Python will give you other things  and balancing all of it I would recommend you to
235. 23:49do Python and only for that reason I picked  Python just because the people were nicer.
236. 24:04It was honestly just a lot of enthusiasts. It was  a lot of people who just enjoyed the language,
237. 24:09enjoyed the people involved with the  language and it would just happen
238. 24:12to be typically motivating enough for  people to want to help out. Ultimately,
239. 24:16there's a sense of fun that comes across  and Guido coded that into the name of the
240. 24:21language influenced by Monty Python. It creates  a sense of community to have your little jokes.
241. 24:30Tim Peters is a famous early example of blending  both like the aesthetics of Python and humor. Tim
242. 24:37Peters was this mythical feature.. creature in  the world of Python, and no one had met him. He
243. 24:45was out there in the ether as a wise one to give  us advice on what to do with numerical processing.
244. 24:51Steve Majewski sent me some email telling me about  Python. So I started a email correspondence with
245. 24:58Guido about the language and about the design  and I got some of the pre-release code and played
246. 25:03with it and talked to him about design decisions  and liked it very much. Tim Peters was a Python
247. 25:10contributor and community member and mentor  to me from very early on. Tim was a channeler
248. 25:18of Guido. He had a really unique way of saying: I  think Guido's going to like this or I don't think
249. 25:24Guido is going to like that. There's a poem called  the Zen of Python which is a partially humorous,
250. 25:30partially serious poem about the the aesthetics of  of Python. I'm actually looking at the Usenet post
251. 25:39where Tim posted his first version of the Zen of  Python, although he called it the Way of Python.
252. 25:47And it was in response to some discussion  where people were complaining that they
253. 25:54didn't understand what was Python's philosophy.  They couldn't figure me out. And Tim had figured
254. 25:59me out and he put it basically in poetry. Here  is the Way of Python by Tim Peters from 1999.
255. 26:09Beautiful is better than ugly. Duh. Explicit  is better than implicit. Simple is better than
256. 26:15complex. Complex is better than complicated. And I  love that distinction. Flat is better than nested.
257. 26:23Sparse is better than dense. Readability counts. I  will say it counts for a lot. Special cases aren't
258. 26:29special enough to break the rules. Although  practicality beats purity. Errors should never
259. 26:36pass silently unless explicitly silenced. In the  face of ambiguity, refuse the temptation to guess.
260. 26:43I did not follow that in a very key piece of NumPy  and it's still a mess. There should be one and
261. 26:51preferably only one obvious way to do it. That's  a direct reference to Perl's motto of there's
262. 26:59more than one way to do it by the way. Although  that may not be obvious at first, unless you're
263. 27:03Dutch. Now is better than never. Although never is  often better than right now. If the implementation
264. 27:09is hard to explain, it's a bad idea. That's a  really good one. If the implementation is easy
265. 27:15to explain, it may be a good idea. Name spaces  are one honking great idea. Let's do more of
266. 27:21those. This is so Tim. Normally 'import this',  imports some useful library, which is very serious
267. 27:31business. Having an 'import this' that was just  a joke was a bit of lightness that we liked to
268. 27:38add. This is from a small hobbyist community,  right? Because if you're a corporate behemoth
269. 27:43making a a programming language, like someone's  probably going to tell you that you can't put a
270. 27:47silly poem in your in your programming language,  but Python's just made by dogs on the internet,
271. 27:52so they can put poems in their software. [Music]  The different ways to make money on the internet
272. 28:00are just beginning to emerge. Entrepreneurs are  putting their faith in a new medium to deliver
273. 28:05the big payoff. It's the dot-com bubble, all  this money pumping into Silicon Valley software
274. 28:11and internet stuff. Early 2000, I decided to  leave CNRI and with a few co-workers joined
275. 28:19little startup named BeOpen. While CNRI, I think,  was an amazing home for Python in the early days,
276. 28:27everything changes and everything evolves and  thinking, well if we're going to strike out on our
277. 28:31own this is probably the right time to do that.  That didn't work out. I'd say within maybe a year
278. 28:38or something. BeOpen was completely incompetent.  We spent the summer in blissful ignorance working
279. 28:45full-time on Python. We built and released Python  2 and within 5 months it was over. We were paid
280. 28:53our salaries every 2 weeks, and then suddenly in  late October we weren't. [Music] It was a moment
281. 29:02where the [ __ ] could have hit the fan. If all  the Python guys went their separate ways... Python
282. 29:08wasn't big enough to survive that at the time.  That was when I had a company Digital Creations
283. 29:14that later became Zope, which is an application  server, database server, index server, web server,
284. 29:19etc. Very large scale commercial quality  application written in Python. And if Python died,
285. 29:26you know who else would die? We'd die. The whole  platform was built on Python. So they were Python
286. 29:32experts. We need to go and make sure that Python  is secure for the future and that the team stays
287. 29:40together. So we negotiated an agreement with them  to join my company. That was an incredibly lucky
288. 29:49rescue. We felt like we really trusted them  and we believed in what they were doing and
289. 29:53what they wanted to do with Python and for Python.  This is a place that I'm proud of. In hindsight,
290. 30:00it was a really critical point in Python's life.  I think if we had chosen wrong or tried to say,
291. 30:06"Oh, well, we'll just strike out on our  own anyway." You never know. But I'm not
292. 30:11so sure that Python would have survived  that juncture in its life. Yeah. [Music]
293. 30:33[Music]
294. 30:33Python just kept sort of growing and  the community kept self-organizing.
295. 30:40One of the I think underlying themes of Python,  if you take it in its totality, is this sense of
296. 30:48grassroots movements. Things grow from the bottom  up and evolve to the point where everybody sort of
297. 30:56realizes, oh, we need a little bit more structure.  Oh, the Python Software Foundation. In my opinion,
298. 31:02the PSF, that damn thing was a success from day  one. That came out of one of the concerns we had
299. 31:10had with BeOpen actually. So CNRI had written  a license for Python and put its name on it and
300. 31:19BeOpen had copied a version of that license  and put the BeOpen name on it. There was a
301. 31:26concern that at some point I might accidentally  end up working for a company that tried to grab
302. 31:32ownership of Python. We realized that we really  need an organization that will be independent,
303. 31:39will not be beholden to any company and their  whims and their lawyers and will really keep
304. 31:47Python users as their first and foremost customer,  so to speak, in mind. [Music] And then the magic
305. 31:59happened when we got into PyCon, which is maybe  the third stool of the miracle of Python. [Music]
306. 32:17What has been your take away from PyCon US this  year? It feels like people are still willing to
307. 32:24go do hard things, work on things together, you  know. Wow. They still do have the passion to be
308. 32:29honest. So, what was it for you? Uh, for me most  definitely like meeting people I haven't met in
309. 32:35a while. That's cool. Yeah. And just pretty  much hanging out with them. You try to walk,
310. 32:40Yeah, to the other side. There's like 10  people that you want to see in between like
311. 32:45stop. I want to talk to you. Oh, I know. Pablo  and Yuri presented as bananas and then Lukasz
312. 32:52asked the question dressed as a banana.  And then Guido came in just as a banana
313. 32:56as well. Thank you for what you do for the  PSF. Please don't step down. Just keep doing
314. 33:01it forever. Raise your hand if you love the  web. Just raise your hand. Raise your hand
315. 33:08if you have done React front-ends. Raise your  hand if you would like to come back. To React?
316. 33:17No, come back to Python. All right. Good. For  the record, 100%. I begged my mother to take
317. 33:24me to PyCon and I met all my online heroes. He  was this distant Benevolent Dictator for life
318. 33:33who I'd been talking to over the internet, but  here he was in person and so I was thrilled to
319. 33:38be able to finally meet him. Then I  think everyone was aware that he was
320. 33:43a 15-year-old kid on the other end of the  terminal. When PyCon was held in Montreal,
321. 33:49the organizer asked me if I'd be willing to  give an opening address to the conference. Well,
322. 33:55I'm a slightly fast talker and I had some spare  time and I just used it as an opportunity to say
323. 34:02thank you. "I like to think of it as I came for  the language, but I stayed for the community. So,
324. 34:06I want to personally thank all of you for  making this such a wonderful place to be
325. 34:10and such a wonderful group of people to be  around. So, thank you". Luckily, it came
326. 34:15off well because people still quote it. I think  it kind of speaks to the core ethos of Python.
327. 34:21We try to develop this tool that people can use  for their needs to get their work done. But the
328. 34:28community is the true strength of Python. It's  not just the language, right? It's the people.
329. 34:36[Music] It was steadily growing every single  year. It was getting more and more downloads,
330. 34:43more and more people making use of it. It  became a more sophisticated language that was
331. 34:50fit for purpose for this new emerging programming  paradigm where the the web itself is the platform.
332. 34:57He told me an interesting number that that made me  realize that the Python user community was already
333. 35:04much bigger than I had estimated because  he said, "Guido, you don't know the sales
334. 35:10numbers for Python books, but I am Tim O'Reilly.  I published the two most important ones and I do
335. 35:17know the sales numbers and they are fantastic."  I don't know if he had that much enthusiasm in
336. 35:23his voice then, but he was sharing this new  to me important fact which made me realize,
337. 35:30oh my gosh! Python tends to get spikes when  certain specialties, I guess you could say,
338. 35:38pick the language up and decide that this is  a thing that they want to use in their area.
339. 35:44Everybody wanted to to do things with the  worldwide web and Python actually became
340. 35:49a really powerful platform for for doing not  just web pages but web servers and services and
341. 35:57suddenly we had an increase of people coming in  from the web world. [Music] I think Dropbox was
342. 36:07one of the early companies to start building  in Python and reaching millions of users.
343. 36:14My name's Drew and I'll be showing you a quick  tour of Dropbox, which is a new way to store and
344. 36:18share files online. Python had been largely viewed  as scripting language you run on the back end or
345. 36:24maybe in an academic setting or in a scientific  setting, but not really for like production
346. 36:29desktop software that you ship to millions of  people. But I just started writing in Python
347. 36:33and hoping that none of the roadblocks would  completely blow me up. You can develop programs
348. 36:38much faster in Python because it's a higher level  language and it's a very clean easy language and
349. 36:43so we could we could be a lot more competitive  than other companies who were writing with Java
350. 36:48or C++ or Perl. Even back then we were competing  against the Google's and Microsofts of the world
351. 36:54and you know our odds didn't seem very good. So  we needed all the help we can get. Google had this
352. 37:00big team like a hundred people, C++ programmers,  trying to do a video hosting site and they could
353. 37:06never keep up with this little thing over there  called YouTube and they went and looked and it was
354. 37:10just a couple of people writing Python. Python  ended up being a big force multiplier on our
355. 37:16effort and no other language that we considered  had anything close to that kind of capability.
356. 37:27[Music]
357. 37:28The fact that data science suddenly  became a thing that you had to do,
358. 37:32combined with the fact that Python was sort  of well positioned to do data science stuff
359. 37:39meant a huge amount of new Python users flowed in.
360. 37:50What was starting to happen in the 2009-2010  time frame, we were seeing more and more of
361. 37:56our consulting deals tied to using Python not as  really a replacement for MATLAB or engineering
362. 38:03type things, but using Python for doing data  processing. We didn't call it data science at the
363. 38:09time, but it was that kind of work, right? It was  modeling, predictive analytics, things like that.
364. 38:13This is the time of Hadoop, the time of big data,  the time of Spark, and people were doing Java at
365. 38:18scale. We were starting to do much more consulting  work in the financial industry and we'd walk into
366. 38:24these large very well-capitalized firms and they  were using Python to do business data processing.
367. 38:28I realized okay if it's a research group over  here doing some science research and they don't
368. 38:33have money and they use an open source thing  I get it but if you have JP Morgan right which
369. 38:38has a lot of money and they're picking up and  using these scientific tools and they're loving
370. 38:43it. I realized that we didn't just have a cheap  free alternative. We actually had something that
371. 38:48was innovative and that was doing something fairly  unique. Peter and I ended up leaving Enthought and
372. 38:55starting another company. We started as Continuum  Analytics actually and its vision was to scale
373. 39:02NumPy and Pandas to large data sets and large  clusters. We quickly ran into a simple problem,
374. 39:08very quickly in, but a very annoying problem.  Yes. people just couldn't even install the
375. 39:12software needed to run all this Python stuff. The  thing about the Python scientific and data stack
376. 39:19is that all of these different libraries, they're  very different than the web development libraries
377. 39:23because these data libraries are often backed  by a very large amount of complex C++, Fortran,
378. 39:28other kinds of software modules. And to build  those correctly takes some work. To build them
379. 39:34correctly on every operating system takes even  more work. And then to build them so that they
380. 39:38can actually be connected together, that requires  you to have this entire build system rationalized.
381. 39:44And so the very first thing we did was like make  a distribution of Python to make it easy to get
382. 39:48that installed. A lot of the data science Python  people, they didn't even use normal Python. They
383. 39:52used this Anaconda Python distribution. Sort of  a dumb joke that I came up with at one point.
384. 39:56It was basically Python for big data. So it's a  big snake. So Anaconda. After like a few years,
385. 40:03we kept going to conferences and going  to places and people wouldn't know who we
386. 40:07were. But the instant that we mentioned that  we make Anaconda, they would say, "Oh, yeah,
387. 40:11of course. I love Anaconda. I use it all the  time." Right? So, after that happened like the
388. 40:14thousandth time, we said, "Okay, maybe we should  rename the company Anaconda." Data engineering
389. 40:19became a discipline where Python was incredibly  entrenched. It was Python and R. And over time,
390. 40:26Python's data analysis libraries caught up and  plotting libraries and stuff started to catch up
391. 40:31to what R had. And then people realized, oh well,  with Python, I can work end to end with my data
392. 40:36versus R being very good at data analysis, but  not necessarily like the data collection stage
393. 40:42and various other stages. I remember going to  a Python conference at one point. I was like,
394. 40:46there's so many sciency people here like where  are all the web developers? This is all data data
395. 40:52pipelines and stuff. I don't want to be immodest  about this because it was a collective effort for
396. 40:55the whole community, but I think that what we  did at Anaconda, not only making the software
397. 40:59installer that made it one click for people to  just run, but also shephering the conferences
398. 41:04and building that community and and whatnot  was really critical. They were incredibly
399. 41:09important contributions. I think we actually made  Python super popular. I I think we did. [Music]
400. 41:24Well, I remember hearing from a friend of  mine that Guido van Rossum was just working
401. 41:28at Google. And that when I thought about  it, I'm like, wait, yeah, Guido, I guess,
402. 41:34works at a normal company and I can't remember  the exact circumstances, but I think either
403. 41:41I got an introduction to him or I just maybe I  think maybe just send him an email saying, "Hey,
404. 41:45um, big fan of your work." you know, it's like fan  mail basically. And he responded and we got lunch.
405. 41:52I mean, he's a hero of mine and he was totally  friendly and approachable. And at that talk,
406. 41:58it was revealed that Dropbox was entirely written  in Python, both the client and the server,
407. 42:04and that was exciting. He was interested in  what we were doing with Python. And I think
408. 42:08we were stretching the language and the runtime in  interesting ways. He approached me again and said,
409. 42:15"Hey, we would really like you to work for us."  And then to my surprise and delight, he ended up
410. 42:21joining the Dropbox team, which was super exciting  for all of us. I stayed there for 7 years. We did
411. 42:28more important stuff for Python than I did during  7 years at Google I think. I think he had a big
412. 42:35impact on the rest of the team. You know, here's  someone who's one of the icons of computing,
413. 42:39but you would never know it just from his, you  know, how he badges in to the office every day
414. 42:45and I think set a really great example in terms  of being super humble and curious and friendly.
415. 42:52You know, it's hard for me to think of someone  who has had more impact with lower ego. Uh,
416. 43:00let's start with the first Python license  plate. Sorry, that was my car when we lived in
417. 43:08Virginia. And there was a box on the application  - sorry, just grabbing the other one, too - uh,
418. 43:17do you want a custom license plate? So, I managed  to get Python in Virginia and in California,
419. 43:23Python was long taken. Like someone in Silicon  Valley probably has a Python license plate
420. 43:30somewhere. So, the best I could get was PY3K,  which I thought was pretty pretty cute anyway.
421. 43:40It became ever more popular and people  started thinking about flaws in the
422. 43:46language. There were a lot of ideas floating  around the core development community,
423. 43:51changes you could make to the language that  people thought would be major improvements.
424. 43:54The only problem with them was that they would  break old Python code that currently existed.
425. 44:00Some of that probably got away  from us a little bit in hindsight.
426. 44:06We released Python 3.0 in December of  2007. I remember because I was actually
427. 44:15interning under Guido at the time. We were  giving people roughly 5 years, I think,
428. 44:20to do the transition. And everyone thought  that was just extraordinarily generous. The
429. 44:25world would totally be on Python 3 in just a few  years. This is best for the language long term.
430. 44:30We're going to make these changes. The community  has always come along with us. We think they'll
431. 44:33come along this journey as well. We were very  naive about that and they they made us know it.
432. 44:43The community said, "No, not worth  it. Kill it. Don't do it. It's not
433. 44:47going to work." This included some very  notable individuals in the community.
434. 45:03So, please welcome our next speaker.
435. 45:09Hi, my name is Armin. You might be familiar with  some of the Python libraries that I wrote and
436. 45:14you might even maybe use some of them. The  Python 2 to Python 3 transition. Initially,
437. 45:19it happened very gradually over many years. There  was the version that everybody used and there was
438. 45:24sort of the version on the side that was built  that nobody used. People didn't want to put in the
439. 45:27effort because their code worked, you know, why  rewrite it. It's very hard to mix Python 2 and 3.
440. 45:37It's uh it's a it's a headache.  So I just didn't move and a lot
441. 45:41of people didn't move. It felt there's a  likelihood that it will not go anywhere.
442. 45:46It felt much more like a top-down decision than  any of the things in the past. And that top- down
443. 45:53decision, I think for some people felt like it  was not inclusive of the real pains of people who
444. 46:00had a long tale of use cases. I think I severely  underestimated how successful Python already was.
445. 46:08How many people had written so much code in  Python 2 that it would be hard for them to
446. 46:16sort of put the effort into translate to Python  3. I thought the most controversial thing by far
447. 46:23was that we wanted to change the way Unicode  was handled. A way to say these things are
448. 46:28strings and they're human consumable. They're  words that people want to read and write and
449. 46:32these things are bytes. These are things  that computers want to read and write.
450. 46:39We basically made all strings Unicode and we made  you use a new prefix 'b' for bytes. And this was
451. 46:46incredibly disturbing. I was implementing  a WSGI library at the time and one of the
452. 46:50things that you do when you implement WSGI,  is that you have to parse HTTP data. You have
453. 46:54to parse cookie data. And there there was  always a mixture of bytes and Unicode. And
454. 46:59it was very easy to work with on Python 2. And  it was incredibly annoying to work on Python 3,
455. 47:04at least initially. There really wasn't  a way to convert your Python 2 code to
456. 47:10Python 3. The tools didn't exist. I  started writing blog posts about it.
457. 47:14Why I think that in the way which Python 3  works right now, I just cannot see the move.
458. 47:31He was a valued community member. And at  the time I was actually quite surprised
459. 47:37that he was so viciously attacking Python  3. I was probably even more negative on it
460. 47:45than appropriate. But Python had such a good  run and I put so much energy into it. I was
461. 47:50part of like building out web frameworks  and and all these libraries. I felt like:
462. 47:54why are we ruining all of this with this move  that didn't even have that many benefits.
463. 47:59I had not appreciated how many people already  had enormous code bases or were on their way
464. 48:06to building up enormous code bases. The payback  for that was that I had to initiate the project
465. 48:13at Dropbox. We had a pretty long road to migrate  many millions of lines of Python 2 code. 5 million
466. 48:22lines of code in the server alone and another  million for the client. That is a very large
467. 48:28pile of code to transform. And so we had to  invent ways of sort of doing it in in pieces.
468. 48:55[Laughter]
469. 48:56A funny sticker I found in my pocket.  There's a built tool called Bazel.
470. 49:01It's sort of a frozen version of  Python 2. So here the sticker says
471. 49:07after re-education enemy of the people  Python denounced their bourgeois class,
472. 49:12renounced formalist notions like Turing  incompleteness and began a new life as
473. 49:18citizen Starlark. This is especially an  example of Benjamin Peterson's humor.
474. 49:27I had the fortune or the misfortune depending how  you see it to be the uh Python 2.7 release manager
475. 49:33for an entire decade. We had to adjust our plans  for Python 3 and for Python 2 so that there was
476. 49:41a longer series of releases where there were both  new Python 2 versions and new Python 3 versions.
477. 49:48What ended up happening is then the community was  like maintaining 2 and 3 and that created a lot
478. 49:53of burden. And so this was happening all during  the growth of NumPy. We made it work for Python
479. 49:592 and then I think Python 3.1 I think or 3.2,  we also made it work there but saw no adoption.
480. 50:11We knew it was going to take a long time.  Uh we knew it was going to be painful but
481. 50:17there is light at the end of the tunnel. The  community contributed stuff like lib2to3 and
482. 50:22other migration tools came out. Benjamin Peterson  wrote Six that really helped that transition. The
483. 50:29language itself became a little more backwards  compatible. I eventually advocated to bring
484. 50:34back the 'u' prefix on the strings and that  actually made it easier to write unified code
485. 50:39bases that target both Python 2 and Python  3. There was a point in time where we made
486. 50:43a very strong declaration, there will never  be a Python 2.8 and Python 3 is the future.
487. 50:53Python 3 just is a better language and it  is getting better over time. Python 2 on
488. 51:00the other hand is a fine language and it will  remain exactly what it is. For a long time,
489. 51:06there was a lot of measurement of how much  Python 2 usage is there, how much Python 3
490. 51:11usage is there. I think it was when Python was in  Portland, was roughly when we started to see like
491. 51:1650/50. Once 3.4 came out, it actually added some  features that were interesting to people like, oh,
492. 51:22I want that capability. So, to me, Python 3.4 was  really Python 3.0. And then 3.5 was an even bigger
493. 51:30milestone. That was the one that got widespread  adoption of Python 3. And by the time Python 3.5
494. 51:36came out, then it was clear that Python 3 would  work. And then you started to have companies come
495. 51:40out and give talks like Instagram's talk, which  I believe is also at Portland, was a big deal.
496. 51:48Yes, you heard it right. Instagram has been  running fully on Python 3 for a few months.
497. 51:54The whole process took I would say about 9 to  10 months. And I think Instagram was the largest
498. 52:00production deployment for Python at the time. She  gave a very good, very thorough talk explaining
499. 52:06why they did it. Two main things: typing and  asyncio. How they did it. We cleaned out all
500. 52:12the libraries that are not making the migration  from Python 2 to 3. How you should do it. Unit
501. 52:17test. What they had learned during the process. We  had actually some good performance gains as well,
502. 52:22after our migration. It felt as vindication  that at least there were large companies that
503. 52:30took the time to do it right and to convert all  their code to Python 3 and were able to do it
504. 52:36successfully. It was definitely well received.  We got a lot of people saying: we really wanted
505. 52:41to migrate as well, but we weren't getting the  kind of support that my team or my company was
506. 52:47giving us. And now I can bring this back to  my management. And that really gave people
507. 52:52the confidence in migrating to Python 3. It became  like a look, Instagram can do it. Why don't you
508. 52:59do it? Instagram's getting benefits, new features,  faster code. Why don't you also want that benefit?
509. 53:12[Music]
510. 53:18Projects undergo things like this. And Python  learned a lot from it and it did actually
511. 53:23eventually end up with Python 3 being in a pretty  good spot again that it was safe to use. And safe
512. 53:29to use sounds weird but like where I felt like I  can actually start new projects on Python 3 now
513. 53:34because we're back to a stable situation where  you're about as productive as I felt like I was
514. 53:40with Python 2. Hi, Anna, nice to meet you. I  really liked your talk. But it was basically
515. 53:44time that healed it. When Python 2.7 was declared  that there wouldn't be any more security fixes,
516. 53:50it became a security concern. That was sort  of the final push. People who donated their
517. 53:56work to do this migration in the wider Python  world and community eventually got the world
518. 54:01running on Python 3, but as I said like it  was a decade long process. And I don't think
519. 54:06anybody anticipated how much work it would be  to move to Python 3. I think Python's probably
520. 54:11too big to ever go through a transition like that  today. The community was maybe a little too big,
521. 54:19but not too big. I even still get angry  messages sometimes from people who are like,
522. 54:24"My code needs Python 2.7. You need to  provide it." And I tell them, "No." I
523. 54:31never wavered in my commitment to Python 3. It  was a lesson learned for sure. For a long time,
524. 54:37I regularly joked there would be a Python 4, but  the transition would be handled much better than
525. 54:43the transition to Python 3. I felt I owed that  to the community. And now, the mantra has changed
526. 54:50to there will never be even a Python 4. To be  honest, I don't know how many people in Python
527. 54:56even know about the whole transition because  that all ended in 2020. At least it's behind us.
528. 55:17[Music]
529. 55:18Okay, so now we get in the closet. So there's  the frisbee. This is the most unique merch I've
530. 55:28ever seen related to Python. Now, here's  a significant t-shirt. It says, "Python is
531. 55:35for girls." I received this anonymously  in the mail at Google. And to this day,
532. 55:45I don't know exactly what the intention of the  sender was. They never revealed themselves,
533. 55:50at least not to me. And I don't know if  there was an intention even.. But I sort
534. 55:55of realized at that point that there wasn't a  whole lot of women in the the Python community
535. 56:04and the community also discovered that  and fixed it for the Python conference.
536. 56:14This idea of being a place where people from  a wide range of backgrounds can learn how to
537. 56:20program, benefit from programming, find community.  There really was some snowballing adoption of
538. 56:26these ideas across languages at the time. Jessica  McKellar looks at the speakers at PyCon and says,
539. 56:34"There's only 3% that are women. What can we  do about this?" She was co-organizer of the
540. 56:40biggest Python meetup in the world in Boston.  She knew what she was talking about. It's not
541. 56:45rocket science. It's like, ask people if they're  interested in speaking and we just.. providing
542. 56:49like a little bit of support drove a huge shift in  the volume and demographics of the talks that were
543. 57:01submitted. And then 3 years later, like 35% of  the speakers are women. I forget what the number
544. 57:08was but I think it was almost even with male  speakers. PyCon... I had an old tweet about this,
545. 57:15so I just pulled this up. Says, "Hello from your  PyCon Diversity Chair." So, the percentage of
546. 57:19PyCon talks by women: 2011 was 1%, 2012, it was  7%, 2013, 15%, 2014 and 2015, 33%. And in 2016,
547. 57:3140%. Problems have solutions. That was a very  inspiring story to me also. Something that was
548. 57:38less inspiring to me was that at the language  summit there wasn't a single woman in the room,
549. 57:45and I don't believe we have a single female  committer to core Python. There wasn't ever
550. 57:54ill intent but unfortunately we fell into  the trap of not doing enough diversity
551. 57:59reach out to try to get more people to  come in. To be honest it was a lot of
552. 58:04white men. So Guido wanted to change that.  I want at least two female core Python devs
553. 58:12in the next year and I will try to train  them myself if that's what it takes. So
554. 58:18come talk to me. I believe one of the first  people who took up that call was Mariatta.
555. 58:30[Music] Even though I've always been  passionate about technology and programming,
556. 58:41I also felt it wasn't a welcoming environment for  women. My schoolmates who took computer science
557. 58:49are mostly men. Most of the professors were men  and at work most of my co-workers were men and I
558. 58:56just didn't really have examples or role models  of women being successful in the tech industry.
559. 59:06One of my co-workers went to PyCon. He  saw that at that conference there is
560. 59:12this community called PyLadies and  I felt like.. I want to meet them.
561. 59:32[Music]
562. 59:33So in 2015, that's when I decided I  want to go to PyCon. At this conference,
563. 59:40I got to see a lot of women speaking and  presenting. At my previous tech conference,
564. 59:48all speakers were men, all the keynote speakers  were men. So this was really different.
565. 59:55One of the keynote speakers at that same  conference was Guido van Rossum. Hello
566. 1:00:00everybody. Glad to see you all. I want to meet  and see the creator of the Python programming
567. 1:00:06language. "And I will try to train them  myself if that's what it takes". I heard
568. 1:00:10this speech. However, at that time, I  just didn't think that it's something
569. 1:00:16I could do. I wasn't even contributing to  the Python community. I didn't contribute
570. 1:00:22to open source at all. I just thought  I'm sure they will find women next year.
571. 1:00:35The year after I went back to PyCon US and Guido  van Rossum gave another keynote. We still don't
572. 1:00:41have two female core developers. I think at  that time that's something really clicked in me,
573. 1:00:47like I realized that there's something  within me that says: if nobody's doing it,
574. 1:00:52I will do it. I wrote to Guido van  Rossum and I asked him for help.
575. 1:01:00Somehow I felt really scared. I just didn't  feel like I deserved mentorship from Guido
576. 1:01:07van Rossum. So I really hesitated to send this  email to him. But in the end, I realized that I
577. 1:01:15want to try. Like I felt like this was a great  opportunity for me. I pressed the send button.
578. 1:01:27I got a reply. He was very friendly. He started  sharing resources about, you know, how to get
579. 1:01:33started contributing to Python. He offered to  do video chats. It's been really helpful to have
580. 1:01:40his moral support. We had a pretty sort of deep  mentorship relationship for a few years. Mariatta
581. 1:01:47learned the process from Guido and continued to  contribute more and more very much around tooling,
582. 1:01:52like she really found her niche in terms of trying  to help make our development process easier.
583. 1:01:58It was scary at first, like I just didn't feel  comfortable sharing my questions in public. I
584. 1:02:05didn't want people to know that I don't  know how to do certain things, you know,
585. 1:02:11like I didn't want people to see that I'm actually  not that good. But this is how it it works in open
586. 1:02:18source. And the community has always been kind  and helped me with my questions without saying
587. 1:02:27that you're stupid. Like I never felt like I was  being judged. I think it does start with Guido.
588. 1:02:41The fact that Guido was very active in bringing  more voices into Python absolutely made Python
589. 1:02:48better and stronger. Just the way the whole  community works and what we stand for,
590. 1:02:53thanks to Guido's leadership, is where I want to  be. And I think that's true for a lot of people.
591. 1:03:10About six, seven months after I  started reaching out to Guido,
592. 1:03:15I was recommended to become a Python  core developer. It really wasn't until
593. 1:03:214 months ago that for the first  time ever. There's a woman...
594. 1:03:31There's a woman who became a Python core developer  and earned that commit privilege. The first woman
595. 1:03:38ever to join our team and has continued to be  there and participate and try to help us and
596. 1:03:43always a very good advocate to try to improve  our diversity. Now that I've seen the camera,
597. 1:03:52I can't do it. My life really changed after  that. A lot of opportunities opened up just
598. 1:03:59because I'm a Python core developer. Hi, how are  you? [Music] Mariatta went on to give her own set
599. 1:04:10of talks about how important mentorship is. When  you don't have role models who you can relate to,
600. 1:04:19you start believing that you cannot do it. Thank  you. Hi everybody. Uh my name is Mariatta. I'm
601. 1:04:28a PyLady. I run PyLadies Vancouver. I also  help with the PyLadies con. 10 years ago,
602. 1:04:37I was also in this room for the  Pyladies luncheon at PyCon US 2015.
603. 1:04:41It was a life-changing event for me. It was at  that event that I felt for the first time in
604. 1:04:49my life as a developer, I felt like I belong in  tech because I see 100 women in the room with me.
605. 1:04:58I just want other women to have role models.  I hope I can help inspire them and help them.
606. 1:05:09These opportunities belong to you, too. You're  all the reason I do this because I know there's
607. 1:05:15still a lot to do and I know together we can  help each other and support each other. Okay,
608. 1:05:23I think that's it. Thank you so much everybody.  [Applause] I have been so consistently impressed
609. 1:05:33by the Python community's ability to have  respectful and and real reflection and
610. 1:05:41dialogue about how to steward itself and how to  move it forward in a way that I find so inspiring
611. 1:05:51and a vision of what is possible in other  communities. And I mean that really sincerely.
612. 1:06:06There were an ever growing number of sort of  scientific disciplines where all the code was
613. 1:06:13written in Python or a lot of it was prototyped  in Python at least. And then the next step was
614. 1:06:21machine learning packages which started with  TensorFlow and was later followed by PyTorch.
615. 1:06:28For a long time, Python was seen sort of as  a scripting language that you would just do
616. 1:06:35small things that you need to automate and not  production ready code and so it wasn't given a
617. 1:06:41ton of attention. And then now AI/ML has sort of  changed things where it is now a business driver.
618. 1:06:48It appeals to people who are not professional  programmers but need software to do a task.
619. 1:06:57There are some other languages like Rm you can do  some interesting stuff in but Python is really it
620. 1:07:03for a AI/ML these days. Python has probably  like centuries of engineering time that has
621. 1:07:09gone into code in the ecosystem specifically  on those topics. So if you're using Python,
622. 1:07:15you're able to instantly tap into those libraries.  Meta has PyTorch, which has a amazing Python API
623. 1:07:22and front-end people can interact with. There's  also SciPy, NumPy, Numba, Pandas. There's a
624. 1:07:28whole suite of really extensive tools. Trying to  build a language and a language ecosystem that
625. 1:07:35is amendable to all of that is non-trivial.  That actually took real design, real thought,
626. 1:07:39and real ecosystem collective innovation over the  course of decades. The back-end of those tools are
627. 1:07:46usually C or Fortran or some other native language  that's much faster, but would be difficult for
628. 1:07:54someone with an AI background to program in. And  it continues to be used for all that by all the
629. 1:08:03AI scientists, developing the models, and  the people training the models, people
630. 1:08:07using the models. And that brings yet another  increment of of growth to the Python community.
631. 1:08:17[Music] Python's a fantastic language  even if you're not a scientist. I just
632. 1:08:24don't think it would have risen to the level  of dramatic usage without the science first
633. 1:08:29data science and machine learning story. We  had a massive impact in the world. You know,
634. 1:08:33I think one testament to the future of Python  is, if you prompt any of the LLMs to do code
635. 1:08:39generation today. If it's front-end, it's going  to be JavaScript. If it's back-end data analysis,
636. 1:08:44it's going to be Python. That's  the future. And Python is a part
637. 1:08:49of that future. Lucky me that I that I chose  that language and not not a different one.
638. 1:08:58Hey, Jacob. Oh, hello, morning. Hope you don't  mind being filmed. Yeah. Ah, okay. Ida's following
639. 1:09:05me. All right. Yeah, she's shooting some b-roll  for the documentary, I think. Ah, nice. I first
640. 1:09:11used Python in 1997 or something like that.  Wow, that's really early days. So, it's got
641. 1:09:2130 frames of animation. Oh, wow. I told everyone  that he would hate getting the DSA, but they did
642. 1:09:26it anyway. Yeah. Sorry, the DSA? Distinguished  Service Award. I labeled him in the annual report
643. 1:09:32as like a leader in the Python community once and  he made me take it out. Yeah, we can see that. But
644. 1:09:37nevertheless, he was on the steering council for  five years. Uhhuh. And he was a leader. I got a
645. 1:09:43job from a Python conference back in 2017. I grew  up in India. I moved to Germany, changed my life.
646. 1:09:49So I think Python changed a lot of things for me  and I just wanted to say thanks to you for for
647. 1:09:54making this language. That's very sweet. Yeah. Do  you mind if I uh take a picture with you or that
648. 1:10:00you don't like it? I'm so sorry. I really don't  like it and especially not in busy places. Okay,
649. 1:10:06no problem. I get it. Okay. Find me in a dark  alley and I'll do it. Okay. Okay. I get it.
650. 1:10:14Ah, I always kind of looked at myself as the  bass player of Python, which is great. You know,
651. 1:10:20you can do a lot, you can have fun, and the  spotlight's not on you. And I love that. I can't
652. 1:10:26imagine having that kind of spotlight on you,  you know, for so many years. I heard a story from
653. 1:10:32an Apple VP of engineering, that Steve Jobs had  turned down their hire of Guido because he said,
654. 1:10:38he has his own following and we don't  want to have that kind of distraction.
655. 1:10:48Python has always been a language that changed.  We've never been one of those languages that says
656. 1:10:54if you write code against this version of Python,  it will run forever with any new version of the
657. 1:10:59language. We just simply have never been those  people. A PEP is an acronym for Python Enhancement
658. 1:11:08Proposal. That's a document that's written  by someone who wants to make a significant
659. 1:11:14change to Python. Basically, you're trying to  convince people that this is a good idea. It
660. 1:11:19would be sent for discussion for a mailing  list and then eventually Guido as the BDFL,
661. 1:11:24the Benevolent Dictator for Life, would decide  whether your change was going in or not. But
662. 1:11:31people always have worried that the next change  is going to be the one that changes the feel of
663. 1:11:40the language. It won't fit my brain anymore or it  won't be accessible anymore. I've actually seen
664. 1:11:47people storm out of rooms saying, "You're ruining  this language." Some people take it very, very
665. 1:11:52personally and seriously. And the Walrus operator  was no different. Yeah so, PEP 572, I don't know
666. 1:12:00who coined the term the Walrus Operator, but  if you kind of turn your head, it looks like
667. 1:12:04a little walrus. It was a new feature that was  proposed, but it was also a syntax change. The
668. 1:12:13Walrus operator lets you do assignments, meaning  put values into variables in places that you
669. 1:12:19previously could not. That was probably the most  contentious language change, oddly enough, because
670. 1:12:26it's kind of a minor thing, but it it there was so  much passionate feeling about whether that should
671. 1:12:31be added or not that it created kind of a bit of a  of a ruckus. When I first saw the Walrus Operator,
672. 1:12:37I didn't like it. It kind of felt a little  un-Pythonic to me, but then I started to play
673. 1:12:43with it. There was a particular use case that  I came up with in my own code that I was like,
674. 1:12:47"Oh, you know what? This is actually pretty  cool". And so then I was like, well, I it's
675. 1:12:53something that I won't use very often, but when I  need it, it's really great to have. And I turned,
676. 1:12:58you know, 180° and became a fan of it. But it was  very disruptive. [Music] Language features and
677. 1:13:06especially syntax can be a double-edged sword. So  one person's succinct, elegant code can be another
678. 1:13:13person's unreadable puzzle or a fusticated mess.  There were very strongly voiced predictions about
679. 1:13:21how bad it would be for Python. And then just  because Python was so much more popular at the
680. 1:13:27time, there are so many places where people talk  about Python. I mean Twitter, you know, at the
681. 1:13:31time. This is all happening in public, right? And  everyone can kind of with low effort throw their
682. 1:13:37hat into the ring and not everyone is going  to restrain themselves when they're doing it.
683. 1:13:45I started being more and more convinced  that PEP 572 was actually the right thing
684. 1:13:50to do. I was the BDFL and I accepted  the PEP. And the next morning I woke up
685. 1:14:04and I felt miserable
686. 1:14:09because of all the attacks  that had happened before.
687. 1:14:14I sat down at my computer, wrote a short email,  and hit send wherein I announced that I resigned
688. 1:14:23at BDFL. And that had an incredible impact  because nobody had expected I would resign
689. 1:14:33and certainly not that I would rage quit  over this issue, which essentially it was.
690. 1:14:55[Music]
691. 1:14:56Transfer of power. Now that PEP 572  is done, I don't ever want to have to
692. 1:15:02fight so hard for a PEP and find that  so many people despise my decisions.
693. 1:15:07I would like to remove myself entirely from  the decision process. I'm basically giving
694. 1:15:12myself a permanent vacation from being BDFL  and you all will be on your own. I am not
695. 1:15:19going to appoint a successor. So what are you  all going to do? Create a democracy, anarchy,
696. 1:15:25a dictatorship, a federation? I'll still be  here, but I'm trying to let you all figure
697. 1:15:31something out for yourselves. I'm tired  and I need a very long break. That was it.
698. 1:15:40Haven't read that.. That was I thought that  was pretty well written. I like to say he mic
699. 1:15:47dropped his way out. You know, he just was like,  I'm out. I'm done. You guys figure it out. So,
700. 1:15:51it was a complete and total shock to me.  Yeah. I just felt really sad about it. I
701. 1:15:58wish he would have retired under better  circumstances. I felt I needed to take a
702. 1:16:06step back to sort of recover myself. I think a  lot of people really felt for Guido, you know,
703. 1:16:13that he could be pushed to that point like  maybe we let him down to allow the vitriol
704. 1:16:26over this change to Python to push, you know,  somebody that we cared about to that point.
705. 1:16:35We all perceived that it was like a a momentous  occasion or like a significant change that
706. 1:16:41you would put a mark in your timeline in the  history book in the history of Python, right?
707. 1:16:46Because it had been run on the BDFL model for its  entire existence. So there was a little bit of
708. 1:16:52uncertainty like, can this work at all without  got Guido? It's kind of a parent saying, it's
709. 1:16:57time for you to grow up and figure it out on your  own. I'm not going to solve this problem for you.
710. 1:17:05The first order of business was deciding on  how to decide. And it turns out some people
711. 1:17:11have very strong opinions on voting systems.  Programmers love arguing about voting methods.
712. 1:17:15I remember thinking this isn't going to work. Lot  of strong voices. It's really difficult when you
713. 1:17:22have hundreds of core developers and thousands  of people online and millions of people in the
714. 1:17:28community and probably billions of dollars of  software that run on Python. You can't think
715. 1:17:35about that cuz it definitely gets overwhelming.  I will fully admit it was stressful. I had to go
716. 1:17:42to an ear, nose, and throat specialist because I  actually was starting to choke on my own throat
717. 1:17:46from the stress. It was not fun because if we  couldn't decide on how we were going to decide,
718. 1:17:52the project was done. We really approached it  like the nerdy engineers we are. In Python classic
719. 1:17:59fashion, we wrote a bunch of peps. We didn't want  to rush it because we knew we were going to have
720. 1:18:03to live with this potentially forever. But we also  realized the world was watching and did not like
721. 1:18:10the idea of not knowing whether this language was  going to be around in a year or not, because we
722. 1:18:14couldn't stop bickering among ourselves about how  to run ourselves. People sponsored different kinds
723. 1:18:20of models. I think there was one which was like,  let's pick another BDFL because that's what we
724. 1:18:23feel comfortable with. And then there was various  kinds of committees and councils and we ended up
725. 1:18:28settling on the five person steering committee.  Good morning PyCon! Thanks so much for joining
726. 1:18:34us. So, welcome to our panel with the Python  steering council. Because there's five people,
727. 1:18:40you don't have the same problem of like all  the stress and responsibility of making these
728. 1:18:45decisions falling on one person. I actually  felt that it made sense for me to be on that
729. 1:18:51first steering council given that there was  no longer feeling so burned out or attacked,
730. 1:18:56to provide some amount of continuity. "You know  how this goes when your kid goes off to college.
731. 1:19:02Some of you may have experience with that. you're  no longer directly involved in their lives maybe,
732. 1:19:07but you never stop worrying. And that's how  I feel about Python at the moment. And that's
733. 1:19:14why I [Applause] why I nominated myself  for the steering committee and here I am".
734. 1:19:25There were elections for the second year steering  council and at first I nominated myself and then
735. 1:19:31withdrew because there were enough other strong  candidates. They didn't need me and I was happy to
736. 1:19:37uh to sort of let go of it. [Music] The legitimacy  that came from the replacement springing up from
737. 1:19:56the community is essential to it being accepted.  It came from us, so it's accepted by us. It works.
738. 1:20:06I think it works for us now. There may be a time  in the future where it doesn't work or pieces of
739. 1:20:11it don't work. You have to be willing to evolve.  Change is the one universal constant of the
740. 1:20:17universe. And so things will always change and you  have to be ready and willing to adapt to change.
741. 1:20:31[Music]
742. 1:20:38It started out as a hobby project. Definitely.  I think I probably wanted to prove something
743. 1:20:43to myself more than anything. It exceeded my  wildest expectations and it actually continues
744. 1:20:50to do so. I constantly hear from people whose  lives were completely changed by Python and it
745. 1:21:00has created this incredible community which was  also a complete sort of bonus over everything
746. 1:21:08else I had anticipated. This community that  has like a massive conference every year and
747. 1:21:16smaller conferences in every continent I think  except Antarctica. There was one of the Python
748. 1:21:26conferences where there was a few thousand people  in the audience. I remember just looking around
749. 1:21:32and going, this is mind-blowing. That's the  point at which I knew we had achieved something
750. 1:21:40important, something that made a difference in  people's lives and in the world. Python turned out
751. 1:21:50to be much more successful than perhaps any of us  thought except maybe Guido and his team. I think
752. 1:21:57they always believed that this was the future.  And of course, it's now one of the most popular,
753. 1:22:02if not the most popular programming language  in the world. He spent two weeks in Christmas,
754. 1:22:06in like 1989, writing a programming language  because he thought it would be fun. Like who
755. 1:22:10would have imagined that that would be such a life  altering and also world altering thing to have
756. 1:22:16done. I think it's been great for the world. I  think it's made people feel like they can get into
757. 1:22:23programming when they never thought they could.  I think it made programming accessible to people.
758. 1:22:28It's literally part of kindergarten through grade  12 education around the world. And is on Mars
759. 1:22:36thanks to being used as a scripting language to  process the parachute of the Perseverance rover
760. 1:22:41landing. Like it's hard to think of anything  that hasn't been touched by some Python code
761. 1:22:45somewhere. It's mind boggling. It's an important  language. It's a popular language. It's a vibrant
762. 1:22:52and relevant language today as much as it was in  the past and I believe as much as it will be in
763. 1:22:58the future. Programmers 30 years from now will  be like, "Oh yeah, you know, I I'm still using
764. 1:23:03Python." Just like today, people are still using  Fortran and C and C++ and those are old languages
765. 1:23:09you know too in the scheme of things. Scheme  - Lisp another old language. So yeah. [Music]
766. 1:23:35[Music]
767. 1:23:40Now I keep thinking of like weird  jokes. There's this library in
768. 1:23:43Python called Pickles because pickle is a  funny word I guess. The documentation used
769. 1:23:49to have a footnote about how you should  imagine Guido and Jim sniffing pickles.