## Intro [00:00] One of the things that we did in the early days of 2014 was kind of just looked around and said, you know, I think we have an opportunity to make a couple of choices that will be bold and aggressive that give us a shot to rein relevance with developers of the world. And if we don't, we're going to be on an iceberg that's going to slowly melt. And at some point, we're going to be swimming. And so we kind of had a set of meetings in spring of 2014 and kind of wrote on the whiteboard, let's be bold. And came up with three big things that we said, okay, what can we do to become more relevant? The first one was we actually introduced a community edition of Visual Studio. Decision number two is let's open source.net and let's make it crossplatform. We want the community to contribute code and do it under the right license and put it on GitHub. And then decision number three was as much as we love Visual Studio, the IDE, let's also recognize increasingly web developers and those that are not using a compiled language are looking for something that's much more of a lightweight code optimized editor. And there was a great project that had been started. It was basically a web-based editor written in Node, written in TypeScript, but it ran in a VSS online. It was great technology, but at the time people weren't really looking to write code in a browser. And the three different decisions we made, we kind of made all those decisions, I think, in about an hour and a half. Of the three decisions, they were all very big. The ones that we remember the most would be VS Code and then the open sourcing of.NET. Microsoft was founded 50 years ago in 1975. So, how has the company changed and how it builds developer tools across these five decades? Today I sat down with Scott Guthrie who has been with Microsoft for 28 years, created the first prototype version of ASP.NET and is currently the executive vice president for cloud and AI. In this episode, we talk about few decades of Microsoft and how Windows became a success in part thanks to shipping programming tools like quick or MFC to help devs build programs on top of the OS. How and why Visual Basic, C, ASP.NET, and Visual Studio were created. the time when developers paid thousands of dollars per year to access quality documentation over MSDN. How Microsoft decided to embrace open source and create VS Code as an open source project and many more topics including what Scott is excited about looking ahead. This is a rare conversation with Scott who has been with Microsoft for over more than half his lifetime. And I hope you enjoy the stories and details that he shares with us. So Scott, welcome to the podcast. It's great to be here. Thanks for having me. It is so nice to to meet again, especially that we met a long time ago, 10 years ago, and that I also worked at Microsoft. But today, I'd ## Microsoft’s early years building developer tools [02:25] like to talk just go back to way to the beginning, 50 years ago. It's crazy to say that that's when Microsoft was founded. And when I think about Microsoft, my first memory is is Windows, but actually it started with developer tools, right? Yeah. The very first product that Microsoft built and and really, you know, not just the first product, but for for many years, we were just a developer tools company. And so it was Microsoft Basic for the Alter and uh Bill and Paul Allen built it while Bill was still in Harvard and uh you know the legend is they flew out to Albuquerque, New Mexico, which is where um the the company that built the Altter was. the computer no longer exists obviously and um and they brought a tape uh of Microsoft Basic a paper tape and they actually plugged it in and loaded it and it worked and it was actually they they built the the app basic entirely without access to the computer and uh it was kind of a the hit product for that that particular computer and then they expanded other platforms and then eventually operating the first dollars they made by selling basically a developer tool, a compiler, right? So like you could you could write basic code and it would compile to machine code, right? That that was the product, right? I think it was an interpreter at that time. The interpreter. Yeah. But but basically, yeah, I mean it was it was literally and and back then that was how you kind of used computers is often you were a programmer and you actually, you know, I guess today it would be modern equivalent be scripting. Uh and you um you know, you built your own logic and your own applications. And so yeah, it's it's you know, that's one of the things that's been cool is that the company started as a developer tools company. And when you look at our what we build today, whether it's VS Code, whether it's GitHub, whether it's Azure, we remain a developer tools company. If you want to build a great product, you have to ship quickly. But how do you know what works? More importantly, how do you avoid shipping things that don't work? The answer, Statig. Static is a unified platform for flags, analytics, experiments, and more. Combining five plus products into a single platform with a unified set of data. Here's how it works. First, static helps you ship a feature with a feature flag or config. Then, it measures how it's working from alerts and errors to replays of people using that feature to measurement of topline impact. Then, you get your analytics, user account metrics, and dashboards to track your progress over time, all linked to the stuff you ship. Even better, Static is incredibly affordable with the super generous free tier, a starter program with $50,000 of free credits, and custom plans to help you consolidate your existing spend on flags, analytics, or AB testing tools. To get started, go to stats.com/pragmatic. That is satsig.com/pragmatic. Happy building. This episode is brought to you by Cinch, the customer communications cloud trusted by thousands of engineering teams around the world. If you've ever added messaging, voice, or email into a product, you know the pain. Flaky delivery and platform stack with middlemen. Cinch is different. They run their own network with direct carrier connections in over 60 countries. That means faster delivery, higher reliability, and scale that just works. Developers love Cinch for its single API that covers 12 channels, including SMS, WhatsApp, and RCS. Now is the time to pay attention to RCS, rich communication services. It's like SMS but smarter. Your brand name, logo, and verify check mark all inside the native messaging app. Built by Google, now rolling out with Apple and major carriers. RCS is becoming the messaging standard. Cinch is helping teams go live globally. Learn more at cinch.com/pragmatic. That is sch.com/pragmatic. And just stepping forward from from from basic this was 1975 but in in the 90s or ## How Microsoft’s developer tools helped Windows succeed [06:15] actually like mid 80s Microsoft started to become big because of Windows. And as I looked into the history I was pretty surprised to see that as Microsoft developed Windows 1.0 and then you know 3.3.0 and 3.1 was a big big breakthrough. They always developed programs to build on top of Windows. So they did uh quick basic Microsoft C and quick C. Do you think, you know, having talked with with early Microsoft employees and just knowing knowing the answer, was this an intentional strategy or or might have this thing actually helped Windows become as big as it did? Oh, I think it absolutely helped Windows become as big as it is. I mean, at the end of the day, if you're building an operating system or you're building a cloud platform, you know, your success is very much you, no one buys a platform by itself. They buy it for the applications that run on top. And so if you don't have developers, you know, building those applications, you don't you don't have a business. And so I think definitely in the early days with Windows, you know, part of the whole thesis was, you know, build great tools, make it easy to build applications that are great, help developers be successful. And if their apps are successful, we'll sell more Windows. And it's similar today with Azure in terms of, you know, one of the things I covered in my keynote was chat GPT, which runs, you know, entirely on Azure. And you know it's like another great example of you know how do we make uh in this case a company OpenAI successful and enable them to build an amazing app and if if they are successful and they're running on our platform we're successful too and so it's it's a model that I think um is very much core to the DNA of the company. Yeah. And then in the in the 90s as Windows became more popular and you know there was these tools Microsoft did something really ## Microsoft’s first tools were built to allow less technically savvy people to build things [08:00] really interesting. They built some tools that allowed maybe not as technical developers to build stuff visual basic Microsoft access later Microsoft front page and I'm now having a little deja vu of of what we're seeing today but just reflecting on that time and what what you've seen you know visual basics seems to still have a massive impact when I when I worked at a bank at JP Morgan you know like building like software that was used by professional traders some of the traders loved using visual basic that they as nontechnical people could you know write their own programs. Yeah, Visual Basic at the time was was absolutely revolutionary because I remember I was uh I was in high school and I was a developer on the Mac. I mean system 7 had just come out. It was like a long time ago on the Mac. Um you know pre you know OS 10 and everything and you know building a guey application was a lot of work. you know, if you wanted a menu, if you wanted buttons, if you wanted a dialogue, I mean, you were writing a lot of code and it was it was often very errorprone code because if you got anything wrong, it would crash. And um there were no visual designers, you know, it was pretty much you kind of had to do everything from scratch. And and Visual Basic came along and it was sort of you could drag and drop buttons, you know, visually lay them out, double click, write a couple lines of code, hit F5, and it ran. And it really transformed development. And um you know we ultimately took that same paradigm and used it with access for databases. We also had a product which is one of the ones I used. I think it's the first thing I ever got paid for was a product called Visual Fox Pro which is another database product that was graphical. Um and then uh and then extended it to things like Visual J++ and then uh you know ultimately the Visual Studio family including C, C++ uh and others. And so, you know, it's that that notion of, you know, again, it was cort of like how do we take the friction out of what developers needed to do and provide um great developer tools to do it. And as much as you know, the the drag and drop aspect of Visual Basic was revolutionary. The other thing that was revolutionary was just the speed with which you could run and uh it had a pioneered a feature called edit and continue. So you could even have the application running, modify a couple lines of code, and then push the button again, and without having to kind of rebuild or recompile or relaunch, you could actually see the edits immediately take effect. And that was also kind of transformative. I, you know, today we, you know, with JavaScript, we we're we're sort of used to that kind of paradigm of make a change, hit save, and it just works. But, you know, in the 90s, that was it was radical um in a big way. Can can can we just rewind back time because I feel there's a bit of a similarity like back then in the 90s before visual basic and and access and fox came about there was you know no way for like non super technical people who knew you know you learn the syntax to write programs and these things just allowed honestly everyday people like my ## A case for embracing the technology that’s coming [11:00] my my dad who was not a developer he used he put together a bunch of stuff with with access later Fox pro and today we're in we're in a similar thing But I I'd like to know like do you remember what the what the mood was like? Were developers saying like oh no this is the end of development. We're now you know like we're not going to have work. Was there any of this and then what what happened as a result of this cuz cuz now we have like 20 or 30 years of looking back of how it transfor and the industry. There probably was some you know questioning in terms of like oh is this going to you know mean anyone can be a developer. I think is often the refrain whether it's you know that era whether it's web development whether it's you know co-pilots and AI assistants now you know I I remember that same push back when Java came out and net came out it was sort of like ah you know no one no real developer would ever use garbage collection oh you know it's really should be you know malakan free um and uh you know I think the history of development and the history of computing is really around how do we continue to optimize the productivity And um you know I think what makes a developer successful or have impact you know is not syntax and it's not unnecessary writing of code that can be automated. It's it's often the logic. It's identifying a problem. It's the creativity and it's and it's often the problem solving um to make it work. And I I think developers that embrace the new technology and um build on top of it end up being so much more productive which ultimately means that they have better careers or more impactful and and so you know I always encourage people that kind of don't don't assume your value is the syntax that you're familiar with you know assume the value is is a higher level thing that you can leverage and you know I think it's true both for developer tools and languages um you know especially in the context of AI now but I think It's also true in terms of of cloud platforms. You know, it's I talked in my keynote this morning at build about um uh Chad GPT and you know, if Chad GBT had been built 10 years ago, you know, or or certainly 20 years ago, a company would need to build their own data centers, they would need to build their own operating system, they would need to build their own deployment orchestrator, they need to build their own Google to do like 20 or 30 years exactly what Google had to do. and and the fact that they you know they were able to leverage things like Cosmos DB and our Azure Kubernetes service you know they they man you know they now have an app that's the fastest growing in history with 500 million weekly users and um one of the stats I put in the keynote was they have a total of 12 people on their infrastructure team that manages all their Kubernetes and compute infrastructure and you know that type of productivity is stunning and um so you know I do think you know I would encourage if you're a developer embrace the technology and the the productivity that's coming and um uh it only makes you more successful and have more impact and there's more than enough problems that that are still left to solve. So a after visual basic one of the you know big years was 1997 I guess for two reasons. Was that the year that I remember that you joined Microsoft? I did I joined Microsoft full-time in 97. ## Why Microsoft built Visual Studio and .NET [14:11] I was actually an intern in 96. So uh but I joined out of college I think I graduated like May of 97 and I joined Microsoft on June 16th and obviously in like in the history books we don't look back on on you joining specifically but it was Visual Studio released and of course later Visual Studio became this really really powerful the bullet tools I I I started my career working on Microsoft technologies and and Visual Studio. So back back then like you were at Microsoft as an intern before this came out uh before Visual Studio came out. What was the story behind why why did Microsoft decide to double down? There were developer tools like there was I think Microsoft had their their quick C there like there were things that kind of like compared to Visual Studio looked like very kind of early versions but there were things like that. H how did this whole development come about and then why did Microsoft decide to just double and triple down on it? Well, I think one of the big things that happened, you know, really and we kind of uh really started the project in 98 was both the real emergence of Visual Studio as a tool and then also net and they kind of you know in in 2000 we unveiled kind of net and visual studio.net and had together at the PDC conference in July of 2000 if I remember. And um you know the impetus behind that was you know Visual Basic had been hugely successful in the mid to late 90s. And um but there was still this gap in between where you had Visual Basic or Visual Fox Pro and then you had on the right hand side in the more advanced category C++ and you know C++ was um and still is a a key programming language but it and and there was MFC right the MFC libraries MFC libraries which which I I guess you can't talk about Windows development with it wasn't just C++ NFC made it like so productive. Yeah. And and there was also a thing called ATL uh which is another library. I haven't thought of that in a long time. But yeah, so we had we had a couple different framework libraries, but it's still, you know, if you use C++, MFC made a huge difference, but it it it wasn't as productive as it could be. And there was this sort of gap in the between. And part of why we created .NET was to sort of say, okay, can we create this common language runtime that could handle VB, could handle a C++ like language, and could handle a bunch of languages in between. And then you know could we also avoid each language having to have its own programming framework. And so like MFC only worked with C++ it didn't work with VB. You know the VB designer didn't work with C++. And you know instead could we build a common set of developer tools whether it was debugging whether it was around visual design whether it was around code optimization profiling etc. um IntelliSense and and leverage it and and that's where really .NET and Visual Studio really uh took root and and what basically we came up with common language runtime. I I worked started a project called ASP.NET with with Mark Anders the two of us um and uh and that was a web framework. Um and what when did that start compare to .NET? So like did net come first or was ASP.NET parallel to to all of this? Sort of parallel. I mean you basically aspet started I mean I kind of wrote the original prototype over Christmas 97 to 98. Wow. Um and uh and it was the prototype I wrote was you know I used some C++ some JavaScript and some Java. It was kind of a you know it was more the idea as opposed to there was no code that was actually reused but it was you know it was this idea of like okay could you use classes could you use objects and could you have language productivity that allowed you to kind of work very quickly and um so I started you know we started Mark and I started showing it to a lot of people internally and got a lot of excitement around it sort of in parallel at the same time frame the common language runtime got started it wasn't called net um but it was um uh it was called core uh was the original I think code name and um but they were kind of building a runtime that could do languages. They didn't have libraries, but you know, they had languages. And then Visual Studio was trying to figure out, okay, we have the VB IDE, we have the VC VC, Visual C++ IDE, we have this Java IDE. How do we merge that? Yeah. Jar, right? J. And then um it all kind of came together. Uh and so in in the process of 98, these three teams sort of found each other and we started working together. And um 99 we built a whole bunch of stuff including Windows libraries and uh for guey and for other things. And then ultimately um we were supposed to release it to the world. I think it was like the equivalent of build we called the PDC back then. Yeah. I think it was supposed to be February or March of 2000 and we were late. Um and so we slipped it to July and uh July 2000 was when we kind of unveiled it at um a big event much like the build event we're doing today and um showed the world like languages, frameworks and tools all working together and that was kind of really the unveiling of .NET and and tools and it was it was pretty critical for us our success in the 90s or the 2000s you know in the same way that VB really helped drive Windows client.NET NET really helped Windows Server and SQL Server and uh you know really introduced Microsoft to a generation of server programmers. Yeah. And I think one of the most iconic you know videos and now memes is Steve Balmer yelling developers, developers, developers, you know, sweating in the ## Steve Ballmer’s speech about .NET [19:54] t-shirt. And usually that's the only part that gets quoted and people think, yo, you know, Microsoft was all about developers. But when you watch the whole thing, which which I did, you know, the whole thing that he said, he starts at, okay, what's a $64,000 question, what the hell are we supposed to do with .NET, Steve? And then he goes developer. So, as I understand, this was about before the .NET release about how to reach developers with .NET like can can you bring a little bit more behind the meme because I feel there's way more depth uh in in in this, you know, statement. Yeah, I mean I I wasn't at that event, so I don't have all the context, but I think I think the main point that Steve was trying to get across and um you know, you see it if you ever watch the video is just the passion he had around developers and and his main point was just developers really matter. And so when people say why are you doing all this? It's because of developers and and developers developers developers. And I think that that is kind of the the just a critical nature which is again if you want to build a platform, if you want to have an ecosystem, you have to have developers and and you know ultimately developers are the ones that both build the most interesting solutions and they also push the platform and apps the hardest. And um yeah, I think that that was his main goal was just to get across to the audience just how passionate he was around that. It wasn't about money. It wasn't about press. It's a really around are you winning the hearts and minds of developers. And I think that's a big part of why you know Microsoft Build is special is um you know even take today's keynotes. It's you know tons of demos lots of hands-on live demos you know hands-on lots of labs. You know it really is an opportunity for developers to get together and it's it's not about sort of chess pumping. It's more around here's what you can do and how do we you know have a a good conversation and dialogue around you know what can we do better at Microsoft but also how can we make you successful. Yeah. And so net was huge when it launched with Visual Studio hand in hand but a third part of of why I think it was really successful at the time was C itself. Where did C come from? Was it in oh it must have come from inside Microsoft but was it ## The origins of C# and Anders Hejlsberg’s impact on Microsoft [22:04] before net? Was it during net? cuz cuz I do remember that C# kept evolving as well with new new features well hand inhand with with net link for example was was a good example where it's both the language feature but you needed framework support as well well you know the the real uh genius behind C is Anders Hollisburg at least especially in the early days and you know Anders is still at Microsoft he's still building languages and he also is responsible for TypeScript um and uh before Microsoft Anders worked at Borland, which is a name that most people, if you're not my age, reme don't remember, but it it was another it was an iconic developer tools company in the the early 90s, late 80s, and uh built some amazing tools. One of which was called Turbo Pascal. Yep. And Anders was the guy who wrote Turbo Pascal. And he he wrote it uh originally from Denmark. I think he wrote it in when he was in Denmark and sold it or licensed it to Borland. And you know, part of what made Turbo Pascal revolutionary, this even came out before I think VB Visual Basic was it was just lightning fast. And so you could literally on a PC with 256K of RAM. Um, you know, it had an editor, it had a debugger, and if you run, you know, in a few seconds, your your Pascal app would would work. And you know, he added good language features into Pascal and and really built that. And you know, we were very fortunate. He joined Microsoft um in the mid 90s um along with a bunch of Borland uh employees and really brought both that developer ethos. He kind of really helped rejuvenate the developer ethos at the time and then also just a language sensibility. And you know, I've worked with Anders now for 25 plus years, and he's just he's absolute genius in terms of understanding both what to add into a language and what not to. Um, and there's a real aesthetic where it's easy to kind of just throw in the kitchen sink into a language, you know, but how do you make these things make sense? How do you you know you mentioned link which was a uh a language query technology and you know I think part of what made the elegance of link at the time so great was it built on generics which was also built into the language which then composed into the runtime and generics in C# was very powerful. Yeah. And it was you know at the time it was a very it was a big differentiator versus Java which didn't have generics um when it when C# first introduced it. and just the way that Anders kind of saw over multiple generations. Okay, we're going to add generics into the CLR. we're gonna add it into the language and then the next version we're gonna come up with link and um you know it was just it's it's been sort of a mastery to see and I think versus other languages there's great continuity throughout it versus you know different languages I think sometimes have taken kind of a left turn and or right turn and not always a kind of linear progression and they've taken detours and um I think it's been great to see the way Anders is both with C# sharp and then also with TypeScript has kind of had a vision that he's built on top of that really has a bunch of consistency and um sort of common direction over time and then in the early and mid and late 2000s you the Microsoft ecosystem was ## The 90’s Microsoft stack, including documentation, debuggers, and more [25:29] really special in the sense when when I worked at one of my first companies we we were paying you know,000 2,000 $3,000 per developer to access Visual Studio together with with C ASP.NET net uh get getting access to the the different software like SQL server and IIS so we could develop and use them because the licenses were very expensive otherwise and to MSDN library now I just want to pause on the MSN library because I've never really seen a company do this before or after where this was before Stack Overflow and and even before having good things on the internet we we would pay to access really really good documentation early on it was sent out on CDs because they were so Do you remember like how or why did Microsoft get this idea of like let's just invest in documentation especially because as a developer I'm going to be honest, you know, like it's usually one of the last things I I I come around to. Yeah, it's it's I you know I feel like I'm dating myself on this podcast, but you know I think for a lot of listeners, you know, the idea of like buying documentation sounds weird these days. Yes. But back then back then it was pretty revolutionary and part of it was you know back like when I joined Microsoft or was an intern you know the internet was still very very new and there wasn't like a search engine out there that was very good and this was all pre Google prebang and you know HTML was was still pretty rudimentary you know it's uh I remember up until about 2002 2003 you know a lot of sites did not even rely on JavaScript or or at most used very very minimal JavaScript because some browsers didn't support it. You know, this is pre-CSS. And so as a result, you know, the idea of searching for documentation and uh reading it in HTML in a browser was just not done. And so, you know, the the original impetus, I think, between MSDN and the '9s was this uh subscription service. And you got every quarter, every month, you got the CD. Yeah, you got you got a lot of CDs. I remember it was, you know, 50 CDs sometimes. Yeah. You know, it would have all the updated applications. So, it have all the Microsoft operating systems, databases, developer tools. I think even Office at one point was included. Yeah. More and more things came into and then you had like tons and tons of CDs with the documentation and so you could kind of install it and then you could search and um you know that coupled with IntelliSense which is or or you know statement completion which is another thing that I think a lot of people take for granted now. And then the other thing that I think we also take for granted is is debuggers. Um, you know, I remember when I was in university, you know, the the debugging experience was often print f statements and or a command line debugger that was very rudimentary. You know, you could kind of dump your your symbols or registers. Um, but it wasn't like debugging today. And so, you know, as a result, it you know, the the the OSDN was for a different time. I guess it's a little quaint but um it evolved over time and and part of it also became this notion of a subscription service where your apps were always up to date and so I think at some point the documentation just turned on to be to be on the internet but um you know the notion of the apps was still but I think one of the my takeaways is is is that when I think back like I was at a startup like a small company 50% company I it was actually in Hungary you know we we didn't have that much like revenue or income compared to the US and my employer still paid that $1,000 or $2,000 per developer per year because the developers were so much more productive using this whole Microsoft stack. And I know things have changed, but to me that point in time is just a bit of a reminder that there is a big premium on just being so much more productive than anything else cuz back then, you know, everything was already together. I think we take these days it's it's so like everything that we're amazed about having documentation of working debugger you know like software that you can use for free like for for database software like that that was server back then but it was all all there and I I wonder if if you know we'll move this notion that like a company that everything's so much better and built an ecosystem that just worked it was really valuable it probably applies today as well like maybe not in you know like a multi,000 subscription but but some something is there like like there's I I just vividly remember how like no one forced anyone to use Microsoft and and it was not about Microsoft it was just about like there was this thing it just were working faster especially when we were building either the first websites cuz ASP.NET was still you know there was a point where 25% of all sites were ASP.NET based on various things after launch it just went up on this thing and I assume this might this must do you have theories on why that was I mean your ASP.NET was your baby a little bit. ## How productivity has changed over the past 10 years [30:17] Well, I think you know the general thing I I think it's true for pretty much every field in fact uh when it comes to technology is you know people like things that make them more productive and let them do more faster cheaper. Yeah. I think that's it's always worth reflecting because like on just you know imagine 10 or 15 years ago what the world looked like. you know, if you were going to an event like this, you would either rent a car or, you know, you'd stand in line for a long taxi and whether it would get you there, whether you get ripped off, no, you know, now you have Uber or Lyft, you know, similarly coffee um uh or a coffee maker, you know, you'd go and find a store and try eight of them and you'd search the job, you know, now you can order on Amazon, it'll show up same day or next day delivery. And so, you know, I do think we kind of overlook the productivity that's changed in just 10 years on a regular basis. It's not baseline, right? Yeah. When you step back and think about what the world looked like back then, even five or 10 years ago, it's completely different. And I think that's going to be very much true for AI. And um when we look at the kind of productivity that something like GitHub copilot has or um you know other AI assistance tools provide you know it's it's going to be like that print f and the debugger kind of story I told earlier or you know it's going to be also kind of a a quantum leap in terms of productivity and you know at the end of the day if if you're getting more productive and saving time you know companies and developers are willing to pay for it if If the developer can make more money because we see from the past like very hard money. Yeah. This episode is brought to you by modal the cloud platform that makes AI development simple. Need GPUs without the headache. With modal just add one line of code to any Python function and boom, it's running in the cloud on your choice of CPU or GPU. And the best part, you only pay for what you use. With sub-second container start and instant scaling to thousands of GPUs, it's no wonder companies like Sunno, Ramp, and Substack already trust Modal for their AI applications. Getting an H100 is just a pip install away. Go to modal.com/pragmatic to get $30 in free credits every month. That is m o d.com/pragmatic. Now, I just wanted to touch on uh just briefly on on Windows Phone. Windows Phone was I I was a Windows phone developer. In fact, I think that's where we connected on one of the events when Microsoft was about to launch Windows Phone. There was pre-launch events. And to this date, ## Why Gergely was a fan of Windows Phone—and Scott’s thoughts on why it didn’t last [32:50] when people ask me what I think about I development on iOS, development on Android, I tell them the best development experience I've had was on Windows Phone. It so Windows Phone, many listeners will have not used it for obvious reasons, but I was a huge fan. Not again not because of of Microsoft or anything but it had some really forward-looking features like live tiles which are now kind of a given everywhere but it was years ahead of it and the develop development experience was so smooth. We had that first class debugger we had the simulator that just worked and you know in in the end Windows Phone is now history but you were it you were there when Windows Phone was born throughout its rise and then in the end when Microsoft decided to discontinue it. What are some learnings that you have on on building platforms and and delighting developers and what works and and and what doesn't? Yeah, it's uh there are some great tools. I mean, I I I ran the the Windows Phone development tool team at one point amongst other things, but but so I was involved on the developer tools side for Windows Phone. I I I didn't work on the Windows Phone team itself, but you know, we kind of took net and C and um Silver Light technology and ZAML and and Visual Studio and I think to your point I think we had some really great tools that were pretty different were really good um I think that if there were two or three lessons from Windows phone for me at least from a development perspective I think one was which is true I think in um technology in general which is uh if you're not not number one you got to be number two um because it's it's really hard to have be number three and number four in the market and to ever catch up and and so in some ways the Windows phone project started after the iPhone was released and um or you know the the modern Windows phone that you like and I think Windows phone would have had a shot other than Android came out and came out beforehand um a short amount of time and and you know I think it was really a question of whether Android or Windows Phone was going to be the the the number two in the market and I think if we'd been a year earlier maybe it would have turned out differently. Um I think the other thing is is the need to kind of connect with developers broadly and um that means you you know we supported Windows phone windows as a development experience but it didn't work on the Mac and and if you think back to 2009 2010 if you went to a developer conference huge number of people were using Macs they were using open source and so you know basically saying okay you got to install Windows on your Mac and use it in order to build an app, you know, was a huge issue. Um, and then I think also at that time, you know, designers were exclusively Mac and to some extent today they still they still are. But, um, and so, you know, that that combination was also headwind. Um, you know, and and and that combination I think was was the reason why ultimately it didn't get the escape velocity. And that's that's why I think speed matters in platform shifts. Well, and that was true for Windows. That was true um you know even in the context of Azure when we first launched Azure I think we were number seven in the market on cloud providers and thankfully we're able to become number two. Yeah. And let's talk about Azure because uh after Windows phone maybe even during Windows Phone you you I as I understand you were a big part of of of the Azure even creating Azure. How did Azure start and how especially I think we just need to remind ourselves that this was back then where Windows was everywhere. The internet was just I guess starting up. Some visionaries might have seen it but it was not that obvious that it will be as big as it it would be. Microsoft was making so much money from Windows licenses and selling developer tools. you know, CDs were arriving on on the mail and then, you know, there was some big investment, lots of people starting to work on this thing called Azure, ## Lessons from working on (and fixing) Azure under Satya Nadella [36:43] which which didn't now now we know it's big, but back then it wasn't like what what was your conviction and and how how did this whole project get off the ground, especially as you said, you were number seven when you started. Well, you know, I think we introduced Azure to the world in the 2008 PDC. So, the equivalent of build, so in 2008, um, and I think it it went general availability in 2010. So the 2008 was kind of a preview and you know back then in 2010 cloud was still very new. I mean Amazon was the leader. Um and but there were lots of I call it hosting companies you know Rackspace and Joyant and and a whole bunch of companies you know that that probably are less familiar name-wise today that um uh different companies had cloud solutions and and but they were really kind of more hosting solutions. Um and you know Azure did have uh when it came out it pioneered this idea of platform as a service meaning kind of some higher level services that made it easier for developers to build solutions but you know I would say the platform is you know it had some usability issues in in that time frame and it also uh did not support Linux or open source at all and the tooling was not great. Um, and you know, I'd say by 2010 when it went general availability, it was it it wasn't doing super well in the market. Hence, I think we were like number seven or number eight or number six. I can't remember what it was. But beginning of 2011, I think Satia took over what was in the server and tools business where Azure lived. And um, I think it was about a three or four weeks later, two or three weeks later, he kind of wandered into my office and said, "Hey, um, how would you like to work on Azure?" Um and uh at the time I was in the developer division, you know, for some reason I said yes. Um and um at the time, you know, a lot of people, you know, emailed me and they're like, "What are you doing? This is career suicide." You know, um you know, this is not going to go anywhere. And so, you know, after we announced it and I got lots of mail from people telling me I'd made a huge mistake. I there was a brief moment I was like, "Well, maybe I made a mistake. I don't know." But you know this idea of having kind of this cloud-based computer that you could build platform you know build solutions on top of and run them at scale and dramatically change the curve of uh productivity as well as success for startups and small companies and big companies um you know was enticing and uh and so yeah I think I started in 2011 you know one of the first things we did was uh that's kind of gone down in lore was I kind of was worked on the project for maybe 60 days, played with the product a lot and said, "Gosh, we have a lot of a lot of things to fix." And so we had sort of this offsite where we brought together all the senior leaders and architects into a room and I went to a Safeway uh and bought these sort of Visa credit cards that were like prepaid and gave one to each table and we mixed the teams up and we said okay you're going to build an app and we have two days to build an app together starting with signing up and um you know half the people couldn't figure out how to sign up and you know half the people struggle to get the tools installed the documentation was out of date, it didn't work and you know the idea was just you know can you build a hello world app and um it was an eye openener I think to a lot of people of like oh my I thought my part was great but no one can use it and um you know we used that offsite at the end I kind of went to the whiteboard and said okay let's list all the things we need to fix over the next year and we kind of created a punch list I just starting as a new user who wants starting with a website and sign up start with documentation let's start with the developer tools, you know, hey, we probably need to support open source. Yeah, we created this punch list and um 12 months later, we relaunched Azure in 2012, started to get some good traction and as part of that, we also supported Linux. We supported VMs, which that was a huge change. I still remember didn't do and um and then in 2014, we renamed Azure to be Microsoft Azure back then. Before that, it was Windows Azure. And then as part of that we also very much focused on businesses because we realized uh Amazon really owned the consumer startup space and which was the biggest market in cloud at that time and um you know I think one of the things that was part of my lessons from watching the Windows Phone project was if you simply try to do exactly what your competitor does and they're ahead of you it's hard to catch up. you know, pick something, pick a beach head that is small enough to win, big enough to matter. And we sort of said, let's be the cloud for modern business was kind of the tagline. It was a hybrid, take advantage of cloud, you know, use cloud, connect to the existing enterprise you already have. And we were able to kind of build differentiation in Azure that um went after a segment that was still pretty small at the time and Amazon wasn't great at. And that was the key thing that kind of took us from like number seven to number two. was people said, "Okay, I I get why I might use Amazon and Azure and um it helped us get some of that escape velocity and put us in second place." And then we've kind of grown every year since then and and now we have ton you know hence chatbt and we have lots of other now we're going much broader market but it helped us in 2014 to 2016 or 17 that we were kind of great in hybrid cloud and Amazon kind of didn't want to go after that space and and as a result that really helped us accelerate and become relevant and this is probably a good lesson for even like startups small startups who are in a very crowded space let's say with with AI like you know look at the places that not might be a bit Yeah, more opportunity. If you can find an underserved market where people are desperate for a solution, that's a great place whether you're a startup or you're an established company to kind of build a new product. Uh if all you're doing is building the same thing that your competitor is doing, but you're coming out later without as much share, it's really really hard the history of computing to catch up. So you mentioned open source and a big shift in the 2010s ## Codeplex and the acquisition of GitHub [42:50] was open source. Microsoft did have a bit of a it felt to me a half-hearted attempt with something called codeeplex. A lot of people will not remember it. It was basically like hey host your stuff on codeeplex. You can choose your license or you can update your license. But most most projects were like like non-permissive license. You could see the source code but the license was just not there. And then something happened. So the weird thing was that Microsoft did have this solution CPLEX and it was popular with Microsoft. One of my old companies used it. And then Microsoft started to just like first embrace GitHub. Obviously there was a acquisition of GitHub but even before I think Microsoft already started to do a lot more permissive open source I think the Ajax toolkit might have been one of the first ones to go out there. What happened then? Like was this coming from internally from below? Was it was it you know like like leaders like yourself saying all right we should just you know like just be a lot more serious about open source like proper open source. Yeah I think it was it was very much a a a cultural shift in the company. Um and then also business model shift. I mean I think ultimately there you know was philosophical things about open source but I think a lot of Microsoft's early hesitation around open source was really around business model. Um and I think it's true today even I think if you if you look at any company and what they're doing you know start with what is their business model and often what they do is driven by the business model. And so if you are um social media, you care a lot about advertising because that is your business model. And so when you think about privacy or you think about um data protection, you know, your position on that matters a lot as to whether you can monetize and you know pay the rent and you know similarly I think if you are a hardware company, you care a lot about the hardware and your your business model is really around selling the hardware and so you care about you know a different set of things and you know we were at the time in the early 90s or the late 90s and and early 2000s you know, most of our money came from commercial software licenses. And so the idea that like you could buy one copy and give it to everyone, you know, for free for free was kind of a scary thing at that time. And so, you know, I think so part of it part of our kind of concern around open source was really grounded frankly in business model. And then I think there was also kind of a misunderstanding of what is open source and um you know back then there was sort of GPL and you know if you checked in one line of code would it you know the license mean that like you just gave away all your IP and so there was there was still in those early days there was a little bit of misunderstanding. It was before, you know, GPL kind of modified the license a little bit and clarified a little bit what it meant. But I think there was also some fear, uncertainty, and doubt that that a lot of companies, not just Microsoft, kind of had around it. And so, you know, that was the '9s. I think what changed in the the late 90s and I probably played a big role in this was just recognizing but if you're a new developer trying to say you need to pay us money for everything whereas something else and and that you can't see the code you can't help contribute to the code you can't participate in the development you know you're swimming against the tide and uh you know at the end of the day if if you are a new or an old developer in in in that time frame you kind of want to be able to see the code you want to be able uh to contribute. Um and so you know we took some early steps and it was a little bit um you know you mentioned the Ajax toolkit I think was the first thing we open source it I think it was I talked to some folks at Microsoft so and I remember another big step we did was we added jQuery um going back memory lane uh which was at the time one of the most popular I think actually prototype was the most popular and jQuery was the up and coming because it was really streamlined it overtook it in no time it overtook it u but we kind of added it into the ASP.NET template inside Visual Studio and that was also considered groundbreaking at the time. I mean in hindsight this all sounds trit I guess but at the time it was a big thing because suddenly we were taking licenses I think they're both like it was MIT or BSD license and incorporating into our commercial product and it was you know it was a bit of a we crossed a bridge or a Rubicon and people said okay I guess the sky didn't fall in and maybe we could do more and so you know it was it was a bit of a journey to kind of take the company through it but um probably my team in particular helped a lot with that and then the other thing that really changed in the the the kind of early 2000s or the 2010, 2011, 2012 was just recognizing that also in the cloud world, you could be very successful if all your customers only use Linux and open source. Yeah. Because the business model did also change. And now people are paying for compute, for for resources, for services, for you know, all the added abstractions if you needed. And and we also then recognized at the time that if you didn't really uh have developers uh hearts and minds, you know, they weren't going to pay any attention to the cloud platform you put out or the or the tools you put out or the databases you put out. And so, you know, we kind of went through this shift over a couple years where I think both culturally we kind of brought the company along of like, hey, this is not scary. This is good. And then also business model wise we had permission to kind of experiment and take risks where you could say okay let's make it free or let's make it open source and which which leads us into visual studio code you know in 2015 if you would have asked me like you know what is unlikely to happen I would have said like that people use more of Microsoft's IDE which was Visual Studio again you still had to pay a bunch of money it was amazing I came from that world but when I moved to new companies they were like let's use open source stuff let's use free stuff let's use we use atom or sublime or maybe We we we might have paid for Jet Brains, but people usually like to just, you know, use a worse experience, but it was open source, hackable. And then out of seemingly nowhere, VS Code came out, which is it's free to use. You know, there's no like tricky things around it. And it just felt again a bit uncredible to me because just knowing how Microsoft used to base their all developer business on pay for for quality tools, ## 2014: Three bold projects to win the hearts of developers [48:52] it just seemed impossible. How did that happen inside? You know, what was the thinking? Now, of course, most developers use VS Code or a fork of VS Code. Yeah. So, I'd been out of the developer division when I took over Azure. So, from 2011 to say 2014, um I hadn't been in in the developer division. I've been focused on Azure. And then when Saté took over as CEO in 2014, um the same day I kind of took over his old job. Um and that included DevDev. Um and so that kind of came back uh into my um world. And you know, I think one of the things that we did in the early days of 2014 was I kind of kind of just looked around and said, you know, I think we have an opportunity to make a couple of choices that will be bold and aggressive that give us a shot to reinvance with developers of the world. And if we don't, we're going to be on an iceberg that's going to slowly melt and at some point we're going to be swimming. And so, you know, we kind of had a a set of meetings in, you know, in 20 spring of 2014 and kind of wrote on the whiteboard, let's let's be bold and came up with kind of three big things that we said, okay, let's what what what do we what what can we do to become more relevant? The first one um was we actually introduced a community edition of Visual Studio. So, previously if you wanted to like use Visual Studio and take advantage of the features that customer, you know, that developers loved, you had to pay a minimum of about $1,000 to do it. And we said, "Okay, let's make it free." And, you know, it was it was scary for a lot of people at the time, but we said, "Look, you know, it's we can make it for small projects, for startups, for, you know, independent developers that want to build something and let's make the full feature set available." And so, that was, you know, decision number one. Um and uh decision number two is let's open source.net had um and let's make it crossplatform with mono and uh well it was it was less with mono we actually took our our base library sorry mono was before yeah mono was before yeah because there's no alternative and now there's no there's no need for mono and we said let's let's open source it and do it right not meaning just open source where yeah where where we can contribute code but like no true open source we want the community to contribute code and do it under the right license and put it on GitHub uh And um yeah, that was big decision number two and we and make it and have a great Linux and Macport um and make them first class. And so that was kind of decision number two. And then decision number three was um as much as we love Visual Studio, the IDE, let's also recognize increasingly web developers and those that are not using a compiled language are looking for something that's much more of a lightweight code optimized editor. And there was a a great project um that had been started. It was basically a web-based editor uh written in Node, written in in Typescript. Uh but it it you ran it in a a VS online, I think is what we called it. Oh, yeah. VS online. Yes. Um and uh it was great technology but you know at the time people weren't really looking to write code in a browser and people would say it's a great it's great if I want to edit five or six lines of code but like it's not a you know it doesn't have a debugger it doesn't have IntelliSense it doesn't you know it wouldn't scale for a large project and you know we said what what if you know I think I kind of said why don't you take this and is there a way that you could package it up in a you know Mac and Windows and Linux you shell and add file system support. And why don't we take this open source debugger we're about to do in .NET and and port it and make it work with it and and really, you know, don't have a project system was kind of one of the mantras and really focus on streamlining editing and and let's make it open source. Um, and you know, the three different decisions we made and and we kind of made all those decisions I think in about an hour and a half. Um, so it was it was a good it was a good meeting. That was a good one. Um, and you know, if honestly if you asked like what was the risk probability of each one succeeding, I thought the first one which is making the tools free was going to definitely drive more developer usage and whether it was going to destroy the Visual Studio revenue, I don't know. But I said like let's take that risk. I think the .NET being open source I thought would help and it certainly did. Um, and then the VS Code one was the most I would say um, speculative where I thought like like I think this could help, but I don't know. There's a lot of other editors out there. There was Adam, there was Sublime, there was um, gosh, a couple other ones that were only Mac and then later new ones are like Neil VMS, etc. Yeah. And so, you know, it was but it was uh, you know, of the three decisions they were all very big. I think the ones that we remember the most would be VS Code and then the open sourcing of .NET. And ironically, the one that I thought would have the biggest impact was probably the one that had the least impact, although it helped a tremendous amount. Yeah, it just it comes to tell you you you can never know even from the inside. But but you know basically I think later in that year we kind of launched announced all three and um you know that really I think helped us with our next kind of I'll call it rejuvenation with the developer community and uh in some ways also the VS code I remember doing events uh in 2015 2016 with a lot of the GitHub team which was then an independent company and um you know VS Code also was that bridge that I think earned us the credibility to talk to GitHub and to be sort of in the developer community at large and um you know was was was the bridge that kind of started the conversation around ultimately GitHub becoming part of Microsoft um which you know if you if you went back to 2010 it would impossible impossible it would be the most crazy suggest solution suggestion ever and then frankly I think it beyond the fact that the GitHub team wouldn't want to have been bought um or be part of Microsoft I think that the developer community would have laughed immediately and um and yet we you know fast forward to 2017 I think was when we did the acquisition people were still had some concern but they said no let's tell you what you've earned the right we're going to give you a chance. Yeah, there there's a lot of skepticism as well. I agree. But but yeah, and then now it brings us today actually to build where Copilot, you know, built by the GitHub team, we know the the AI assistant with a bunch of advanced capabilities is now open source as well together with with VS Code. Yeah. So, so, so now, now that we're here, just like looking ahead, you know, we we are where we are with lots of exciting new tools, lots of new ways to work, and I think we're going to figure out their agents are are the hot ## What Scott’s excited about in new developer tools and cloud computing [55:40] thing. Personally, what are you excited about in terms of developer tools and also cloud when you look at may these be projects, may these be directions or things that need to be figured out? I think the thing I get excited most about is um certainly on the developer tool side. I do think this notion of having an agent that works with us and that's you know some of the demos we showed today and yesterday of you know how can you assign an issue in GitHub to your co-pilot um I think I think a lot of us have have used whether it's GitHub copilot whether it's chatgbt or or n65 copilot or cursor you know people are used to kind of a request response model where it's you you type something you get a response immediately and that's super powerful so like that's not going way. But this notion as the models get richer that you can just assign a task to effectively a co-orker that is AI that can do something over maybe 10, 15, 20 minutes and then assign it back to you. I think that is super profound. And so the ability to say, "Hey, take this Figma or take this screenshot that I've I've sketched out, you know, turn it into a nice HTML with CSS or, you know, create for me you know, a Kubernetes deployment file and a basic microser architecture that has these five dimensions and it's connected to these three resources and make it scalable and secure like assign it to the the co-pilot, grab a coffee, come back and get something back. Um, you know, that that I think is is is really profound. And then how do you extend that so that in addition to development a lot of kind of what you might think of as the operation tasks or s sur hey is there anomaly detection in my logs you know is the performance dropping why you know can you ask it what changed in the environment what changed in the codebase what's changed in terms of the user behavior you know how do you have an AI agent that can kind of assist you in that you know I think that's really about kind of giving every developer superpowers And um you know it's kind of like the Marvel Iron Man. You know suddenly you have a a suit that gives you kind of these amazing superpowers. You know I think to some extent these types of co-pilots are going to do that. And then when you combine it with a cloud like Azure where you know if you're if you've got a great idea and you're going into you know business as a startup or whether you're a big company you know the ability to run it in 70 regions around the world and um you know certainly right now when you think about whether it's tariffs or whether it's um geopolitical you know this like okay how do I meet every country's local residency requirements on data you cloud is the way to do that. Um similarly if you think about you know how do you target the growing markets in the world cloud is the way you do that and and then when you start thinking about you know how do I actually build my own AI application and take advantage of the latest models how do I use diffusion or fine-tune my own model that's based on say my own data or do post training you know that that's the cloud's going to be the way you do that and so I get excited in terms of when you take these sort tools. You know, tying it back to the beginning of the conversation, going back to MFC and Visual Basic. You know, I think this is the same level of kind of exponential jump, but probably even more profound in terms of the impact because it's not just development, it's also runtime. And then it's that feedback loop of okay, now I get something out there. I'm learning from my user behavior. Let me improve it. And um that's the thing that always motivates me is is can you ultimately make someone more successful and um if you can allow them to take an idea and run with it faster and do it faster, better, cheaper, you know, you can change their life and you can bring to life this great idea. And if you can do that at scale, it's it's a fun journey. And as as closing, hearing about like AI colleagues can be a little scary for ## Why Scott thinks AI will enhance productivity but create more engineering jobs [59:50] as as engineers. I think it's a similarly big shift as as back if you think back in the '9s. Oh, anyone can program with like dragging and dropping and doing similar stuff that I'm doing right now. What would your advice be for software engineers today like you know mid-level or experienced engineers who who want to be the you know standout engineers of tomorrow on on how to approach this and how to think about the fact that now they can a lot of the work that they have been valuable for until now can potentially be offloaded. It's a little bit you see what I mean that kind of messes with your mind a bit. It does. I mean I think the thing I'd probably say to encourage is like I'm a big student of history. I like technology history I' I've been part of and I study sometimes but I also like to look back and you know the last 500 years of history and read lots of biographies and you know history has a way of repeating itself um and you know like in just going back to the development community for a while like I remember in the you know early 2000s people would come to developer conferences and they would literally have t-shirts that said intelligent intellisense rots the mind. Oh, because like no real developer in sense was a really good autocomplete auto. Yeah. Like like no real developer would use autocomplete was kind of the statement at the time. Um you know real developers you know knew the name of the method and the parameters the classes. Yeah. And um you know seemingly I think type safety uh whenn net was introduced you know c++ developers were like type safety is for wimps. I mean like you know just all you need is a void star pointer and you know if if you got it wrong you'd crash but you know like who needs type safety you know and and similarly I think um you know even when JavaScript emerged in HTML 5 you know people would say like well you can't really a real app with HTML and JavaScript we had the thing it's not a real language JavaScript I remember this one and so you know all these parallels I kind of and then going back to you know the days of debuggers you know real developers don't use debuggers um uh or profilers. And so it's you know I' I've heard these things before where people say you know you can and similarly with cloud I've certainly heard lots of people say well you know I can run my private cloud cheaper and faster and better than you can. It's like okay you know most of those companies are not in business anymore but um but sure go for it. And so I I do think um you know that would be my only kind of history lesson I'd pass on which is you know generally these things are are frightening sometimes because they do make it look like wow could this automate me? Could this replace my job? But ultimately if you embrace the productivity you suddenly discover you can do a lot more and you can you can deliver a lot more. And that is really what makes you valuable. It's it's not whether or not you know the name of the method and are proud that you remember the 18 parameters to it. It's it's you know what does the method do and and does the application calling the method accomplish some value. And so to the extent that we can leverage AI to kind of make developers more successful. I think we we in the limit it's going to create more jobs not less. And I think ultimately it means those jobs are going to be higher paying because the business value you're delivering to the company or the startup or going up you your app that you're building becomes even greater. So you know that you know it's it's not always a perfectly linear progression but I think that's if I look again back in history garbage collection intellisense debuggers type safety um even open source you know it's it's those are all things that were proven right um and the naysayers were proven wrong um over time. Yeah and I guess it's just good to remind that open source I think from the outside we would not think it's scary but from the inside like Microsoft it was and it turned out pretty good it worked out pretty well. So, thank you very much, Scott. This was so much fun. This was a lot of fun. Thanks so much for having me. Thanks to Scott for his trip down memory lane and for his advice on how software engineers can keep growing in an industry that changes as fast as software engineering does. For more details about Microsoft's developer tools evolution, check out our deep dive into pragmatic engineer linked in the show notes below. If you've enjoyed this podcast, please do subscribe on your favorite podcast platform and on YouTube. A special thank you if you also leave a rating on the show. Thanks and see you in the next