00:00hi everyone welcome to the a 6nz podcast
00:02today's episode is all about the end and
00:05new beginning of programming featuring
00:07guests chris wanstrath co-founder and
00:09former CEO of github and a six in c
00:12general partner Peter Levine who's on
00:13the board of and led her investment in
00:15github the conversation was recorded
00:17previously at our annual ASIC since a
00:19summit event only this Q&A turns the
00:21tables with Chris asking Peter all the
00:23questions around his thesis about the
00:25end of programming with the glimpse
00:27ahead of what comes next
00:28hi everyone hello thank you for joining
00:31us today i'm chris by the way this is
00:34Peter you weren't if you're confused
00:36I'm the github person this is the
00:38Andreessen person but we've actually
00:40known each other since 2012 we've been
00:43working together for a long time
00:44we're ten years old and we first decided
00:47to raise investment in 2012 after we
00:49were around four years old because we
00:51wanted to branch out from our own
00:53personal networks of software developers
00:56from programmers we wanted to get into
00:58the world of business we wanted -
00:59brilliant - anyone who was interested in
01:02building software and and Driessen has
01:04been a huge part of that
01:05but Peter in particular has been there
01:07since 2012 since the very first meeting
01:09that we had so I say that not to say
01:12that Peter is this great guy because he
01:14is but to tell you that he has
01:15interviewed me so many times that I'm
01:17very excited to interview him for once
01:19so this is a big moment for me I come
01:22from this world of really product
01:24oriented consumer focus software
01:28developers building things for other
01:29software developers and peter has helped
01:31as a software developer himself me and
01:34the company really expand what we're
01:36thinking in terms of customers and in
01:37terms of you know our approach to
01:39business so this was an opportunity for
01:42us I feel to talk about some of the
01:45things we agree on in terms of the next
01:47ten years and some of the things we
01:48don't agree on I don't think either of
01:50us knows the truth of what's gonna
01:52happen in ten years from now but I think
01:53it's somewhere in the middle I have to
01:55ask one question you know we're here to
01:57talk about the the beginning and the end
02:00of programming how many programmers are
02:02out there in the world we think there
02:03are 20 million people employed that are
02:05writing code but the idea of what a
02:07programmer is someone who writes
02:09software in their free time or even
02:10professionally that's not counted in a
02:13it's way more than 20 million it's way
02:15more than 40 million it's growing every
02:17day and I think what that means
02:19isn't that the reporters are wrong or
02:21Gartner is wrong or read monk is wrong I
02:23respect this firms a lot I think really
02:24what that means is the term and the sort
02:27of definition is evolving and changing
02:29we also see our customers people that
02:32used to use JIRA people that are in the
02:33marketing team that want to update just
02:35some copy on a website they're
02:37increasingly moving to something like
02:38github and they're they're also giving
02:40us feature requests specifically for
02:42these teams of people that are not the
02:44core you know sort of code collaborate
02:46ship developer pipeline so I think the
02:49answer to question is 20 million
02:51full-time 40 hours plus developers but
02:54don't be fooled by that number there are
02:57way more people who have their fingers
02:59in development than the numbers then
03:01github can even tell and it's growing so
03:04the more you can open that up I think it
03:05the better it is for everyone so anyway
03:06what I want to talk about though is not
03:0820 million or 40 million developers we
03:11talk at github a lot about the next
03:13hundred million developers so I what do
03:16we need to do to ensure people in the
03:17future can become developers for me it's
03:20a lot more of how do we lower the
03:21barrier to entry how do we get more
03:23people on board how do we make
03:25development this thing that a billion
03:27people want to use I think for Peter he
03:29sees it almost I don't put words in his
03:31mouth sees a different route to the next
03:33billion developers and I think you see
03:35it more from data than simplifying tools
03:38making tools easier lowering the bread
03:40entries so originally I was thinking
03:42about this talk as titling it the end of
03:46software development but then again
03:48maybe a little much so it's really how
03:50do we go from 20 or 30 million
03:52developers to a billion developers so in
03:55order to get to a billion developers I
03:57want to change the definition of what a
03:59developer and what a programmer is right
04:01so when we think about sort of data as
04:04the input into a system we often think
04:07about right now like you know facial
04:10recognition is easier done better done
04:12by a computer than done by a human right
04:15and I started to think about how data as
04:18an input to a system could actually
04:21change the world of programming such
04:24that everyone becomes a programmer the
04:26reason why I think that programming ends
04:28as we currently know it is the
04:31declarative language is that we've used
04:33since the beginning of time beginning of
04:36compute time has all been based on
04:38if-then-else if this do that do this
04:41right and it gets very complicated but
04:42it's all if-then-else in basically as a
04:46programmer you have to know what you're
04:48asking for you tell the computer what
04:50you want it to do as opposed to the
04:53reverse is I want to look at data and I
04:56want to instruct the outcome as opposed
04:59to a priori knowing what the input might
05:01be so let me give you an example a
05:04football coach becomes a programmer and
05:08the one had a program a football game
05:11and said quarterback throws the ball to
05:13the receiver if this then that and these
05:17it's an indeterminant problem there's
05:19too many moving pieces like the program
05:22would be very very complex to just run
05:26one play okay just like if you were
05:28doing facial recognition or voice
05:30recognition is very complicated to do
05:33on facial recognition so imagine now we
05:38have all the tools to go and do this
05:40imagine that through repetitive looking
05:42at football plays the system actually
05:44learns and it learns what the right
05:47movements are and all of that and then
05:50what does the coach do the coach comes
05:51in and optimizes that model to say you
05:54know what the system got it a little
05:56the quarterback should move ten yards
05:58back whatever and the coach in
06:01conjunction with the system actually is
06:04a programmer because the coach actually
06:06informs the data and they become a in my
06:10new definition of the world a programmer
06:12and you can take this now across any
06:14field whether you're a lawyer an
06:16accountant a salesperson and we can
06:19watch the best behaviors we can watch
06:21the best behaviors and best human
06:24interactions let's say I'm trying to fix
06:26a transmission well I can wear goggles
06:28toward it and I have the best expert
06:30going and fixing a transmission and that
06:33person would then go back to that data
06:35and of course we need the right tools so
06:37what are the tools for us to go
06:39manipulate this data won't be a
06:41spreadsheet it will be some visual
06:44interface where I as an auto mechanic
06:47can go in and modify the output of what
06:50the system is doing and in effect we
06:52create a set of expert rules based on
06:55what the human does and then imagine
06:57that we go roll out those rules to
06:59everyone else who has to go fix
07:01transmissions or become a salesperson or
07:03a lawyer or an accountant or football
07:05coaches and in my mind by using data and
07:08having folks optimize the data they
07:11actually become the next generation of
07:14programmers not trained as programmers
07:16they won't even know the program really
07:18is they're all manipulating data in the
07:20context by which they are familiar with
07:23their own expertise so that's what I
07:26think happens well so I guess my
07:28question there is my father is not a
07:30computer guy but very much a car guy my
07:32father too he loved cars he works on
07:35them and if I can give him all the
07:38information at his fingertips to fix the
07:40car I think it makes his journey much
07:42better but if I were to give him all the
07:44information available to fix the
07:46diagnostic program that was running on
07:49you know let's say he has all the data
07:51available to him is he able to actually
07:53make those changes so one side of it is
07:56yes I have all this data available to me
07:57the other side of it I think is do I
07:59have the tools to make these changes yes
08:01I can understand but can I actually
08:02affect the world is there a second sort
08:05of industry or a second tier of you know
08:08data becomes the most important thing
08:10that says yes now everyone has access to
08:12what's going on but how do we actually
08:14affect that or do you see those things
08:16I mean I think it's very intertwined you
08:18know your father I've actually met your
08:19father he brought ice cream to a party
08:22he didn't eat her ice cream one time if
08:25you ever from Cincinnati please go to
08:26Graeter's ice cream it's the world's
08:28best ice cream it's worth bringing to
08:29California I mean imagine your dad is
08:33fixing some part of the car and that has
08:36been done by someone else and imagine if
08:38somebody else has optimized effectively
08:40created the exact precise user manual
08:43for how to fix a particular part of the
08:45car imagine if your dad now was wearing
08:47goggles and it looked at what he was
08:49doing and the system was helping him to
08:54do a better job in fixing certain parts
08:56of the car you take the experts who are
08:58the programmers in their field and they
09:01then push that knowledge off to everyone
09:03else and maybe your dad's an expert in
09:05certain parts of this and he can
09:06participate in it it's the open source
09:08of data right everybody manipulating
09:10this to get better solutions so then for
09:12the current open source developers
09:14people that today spend their time
09:17fixing bugs sort of trying to make this
09:20piece of software better not because
09:22they're making money off of it because
09:24they want to become a developer who has
09:25this name what does this do for them
09:27does this make them become data
09:29scientists the current crop of
09:32open-source developers or do they now
09:34partner more with data scientists in the
09:36future you know I think that data
09:38science will become the new academic
09:40approach in computer science it'll be
09:43less coding and more about data science
09:45new algorithms for data science new
09:48approaches to understand the world
09:50around us so you know we can call this
09:53software 2.0 which is moving from code
09:56to data and I think many people will
09:59need to literally upgrade their skill
10:01sets to know much more about data than
10:04they will about code my opinion don't
10:06people just suck at interpreting data so
10:09like I think about fake news I think
10:11about like the lies that we all see
10:13going to the internet and I think about
10:15programming like if we're trying to get
10:16a billion developers don't we feel like
10:18people in general including myself are
10:20bad at data but okay at like telling
10:22people and machines what we want
10:24so let me parse the question in a couple
10:26of ways when we know what we're looking
10:27for we're actually very good at that I
10:30can write a query and say look for you
10:33know all numbers you know greater than
10:3510 when you know what you're looking for
10:37we're actually very good at that what
10:39we're not good at is looking for things
10:42that we don't know that we're not
10:43looking for right and so I think what's
10:47going to happen and we're already
10:48starting to see this is
10:50systems in the future will actually help
10:52us see data in ways that we can't see it
10:57right now so let me give you an example
10:59a real example there's a company that
11:01select data off at traffic signals and
11:03to see how many red and green lights
11:06that occur in a traffic environment and
11:09the flow of traffic that's the input so
11:12that I would argue we could code that
11:13but the system actually uncovered the
11:17fact that there's a correlation between
11:19the lights not working and the
11:22maintenance schedule for the light
11:23itself that was something that the human
11:26in your sense right terrible at finding
11:30things that we don't know that we're not
11:31looking for because you don't know but a
11:33computer through looking at data and
11:36looking at patterns actually help to
11:38determine in that situation a new thing
11:41that the human wasn't looking for it to
11:43begin with and that's where I would
11:44argue that code breaks down code will
11:47say the number of times that you know
11:49the current code if light goes on
11:51counter goes up by one right but if
11:54you're not looking at a maintenance
11:55schedule there or whatever how would you
11:57ever know to look at a maintenance
11:58schedule right so that's an example of
12:00kind of taking correlated data and using
12:03that through a machine that I believe
12:06will start to really you know move us
12:09again from this notion of intent you
12:12know find something to this notion of
12:14using data to find correlations that
12:17actually we may not have been looking
12:18for I mean to some people the fake news
12:21is real news yeah that's right a more
12:23arbitrary but certainly you know the
12:25input of data I believe is going to be a
12:26super big deal so then I guess part of
12:29that is one of the things that we first
12:31talked about ever when we met was that
12:34the X Window System yes the UNIX
12:36windowing system X Windows so I've used
12:38that for a long time it came out of the
12:4080s one of the big first in my opinion
12:43successful free software projects that
12:45brought sort of this idea of open source
12:48and free software to consumers and it
12:50wasn't the best but it was there and it
12:51was an option and it was great so I
12:53think that is all to say that there was
12:56once a period of time where everyone
12:57everyone believed that codes intrinsic
13:00data was in being kept private and there
13:02was there was value in
13:04and like if I add this piece of code and
13:06you didn't I had something that was
13:08worth something and maybe you were in a
13:11position where while the industry felt
13:13that way you know some of the hippies
13:15that you're not wearing sandals today
13:17but some of the hippies you either went
13:18to war are you were t-shirt yeah shoes
13:21in the 80s where you were hanging out
13:23they were really into free software
13:25before a lot of us I think realized that
13:26there's something real there it's not
13:28just about hippies it's about there's a
13:29practical value do you feel like data's
13:31in a similar spot it's our valuable
13:33asset there's intrinsic value into
13:35keeping data secret right we once felt
13:37that about code I don't know if I think
13:39that data being open is going to be
13:41something that works but I think the lie
13:44people once felt that about code yeah do
13:45you think that we're gonna want to pay
13:47tomorrow pin yes I think we do you know
13:49there's this phrase out there data is
13:51the new oil right and I have this vision
13:53that we all become our own oil well and
13:56we are going to dispense data and there
14:00are now ways where people can get paid
14:03for producing data right so instead of
14:05contributing my data to some centralized
14:08place I can actually monetize that data
14:11my data to the extent that I want to or
14:14I can contribute it examples might be if
14:16I have a camera on the front of my car
14:18and I'm going through streets and I'm
14:20recording visual information I can
14:23choose to maybe offer that to some
14:27service that's going to make maps better
14:28but maybe I get paid for that or if I
14:31have let's say our talk here if somebody
14:33finds it valuable maybe we get paid for
14:36it right and so I think there will be a
14:38lot of cases where data becomes the
14:41purview of each individual and we will
14:44choose how we want to go dispense it and
14:46I think in certain cases our data and
14:48what we produce will become valuable to
14:51consumers and there'll be a market for
14:53data and rather than all our data going
14:57to centralized companies that are you
14:59know hidden inside of organizations we
15:02all will have control over our data and
15:04I think to have a market and to have
15:06economics around that data is really
15:08interesting so I think that's kind of
15:10the way it takes shape in terms of what
15:13and how people own it and control it
15:15yeah that definitely resonates with me
15:18I've seen a lot of history repeats
15:20itself but the timescale is always what
15:23you know is the most different to me you
15:25know free software really started in
15:2780's j-dub came out in 2007 and for me
15:31github was very much influenced by the
15:32open source movement but I think a lot
15:33of people open-source didn't become part
15:36of their vocabulary until maybe later
15:37thirty years after free software so with
15:40data do you see a similar timeline I
15:42mean in some ways it's already happening
15:44there are companies out there who look
15:46at salesforce productivity looking at
15:49the most productive salespeople in an
15:51organization learn what they do modify
15:54that and then go give that information
15:56to so where everyone gets assisted by
15:59this new technology a lot of that stuff
16:02is starting to happen we may not think
16:04about it as the data economy or this
16:08change in programming but it is starting
16:11to happen as we have these new models
16:13where we can process massive amounts of
16:16information and to where the outcomes
16:18really become I'd say digital assistance
16:21to humans to actually make them better
16:25whether it's the football player the
16:26auto mechanic the accountant the lawyer
16:28or the salesperson each of those can be
16:31better based on observing the experts in
16:34that field and I would argue with
16:36starting to happen and sure it'll take a
16:37long time and all that but I definitely
16:40see it so my final question is hopefully
16:42relevant of people here it seems like
16:44this is starting to happen like your
16:46blog post I think from a year ago on
16:48edge computing yeah you felt it was
16:51starting to happen so maybe to you a lot
16:53of what's occurring in this conversation
16:55feels obvious but to others it's very
16:57new what should I think about if I'm
16:59like a CIO or any executive at a company
17:01and I'm hearing this is anything I
17:02should do differently in the next years
17:03I think I should think about or is it
17:05just like trust my team because this is
17:07happy you know in all of these sort of
17:10my end of things that I've been thinking
17:12about whether it was the end of cloud
17:14computing last year the end of
17:16programming you know want to hear
17:17everything right yeah but it starts a
17:20new thing and what has been interesting
17:22is when you start out with these I like
17:25when I start to talk about this
17:27literally everyone looks at me because
17:29I've been through this now like I'm
17:31crazy like are you nuts program is gonna
17:33be around forever and and that and as I
17:36start to think about it in my own mind
17:38and put some concrete examples and bring
17:40things together it actually starts to
17:43make sense even to me I mean I thought
17:45about this I couldn't really describe it
17:47now you know kind of over time you get
17:49better at sort of thinking about the
17:50examples in that so what I would leave
17:52for all of you is what does it mean in
17:56your organization to utilize data I mean
17:59I gave the Salesforce example that's a
18:01very real example so where else can we
18:04model the experts in your environment
18:07where can we take those you know kind of
18:10that data and then how do we provide you
18:13know what framework and what rules would
18:16you go and look at as CIOs on where you
18:19might optimize things based on people
18:21doing expert work in your environment
18:23and I will bet you if you start just
18:25thinking about this because I did I went
18:27through this exact exercise you will
18:30realize that every function within your
18:33company has the opportunity to be
18:35data-driven has the opportunity to be
18:38iterative and cognitive based on the
18:40work that we're all doing to try to
18:42optimize things and once you start
18:45thinking about this then it becomes
18:46reality and so I'll leave you with that
18:49and next year we can revisit this I
18:51guarantee you it will start to
18:54proliferate through the industry in the
18:56same way that some of the other things
18:58are awesome well thank you so much
18:59computer thanks everyone