>> Announcer: From Sand Hill Road in the heart of Silicon Valley, it's theCUBE, presenting the People First Network, insights from entrepreneurs and tech leaders. (bright electronic music) >> Everyone, welcome to this special CUBE conversation. We're here at Sand Hill Road at Mayfield Fund. This is theCUBE, co-creation of the People First Network content series. I'm John Furrier, host of theCUBE. Our next guest, Yuvi Kochar, who's the Data-centric Digital Transformation Strategist at GameStop. Variety of stints in the industry, going in cutting-edge problems around data, Washington Post, comScore, among others. You've got your own practice. From Washington, DC, thanks for joining us. >> Thank you, thanks for hosting me. >> This is a awesome conversation. We were just talking before we came on camera about data and the roles you've had over your career have been very interesting, and this seems to be the theme for some of the innovators that I've been interviewing and were on the People First is they see an advantage with technology, and they help companies, they grow companies, and they assist. You did a lot of different things, most notably that I recognized was the Washington Post, which is on the mainstream conversations now as a rebooted media company with a storied, historic experience from the Graham family. Jeff Bezos purchased them for a song, with my opinion, and now growing still, with the monetization, with subscriber base growing. I think they're number one in subscribers, I don't believe, I believe so. Interesting time for media and data. You've been there for what, how many years were you at the Washington Post? >> I spent about 13 years in the corporate office. So the Washington Post company was a conglomerate. They'd owned a lot of businesses. Not very well known to have owned Kaplan, education company. We owned Slate, we owned Newsweek, we owned TV stations and now they're into buying all kinds of stuff. So I was involved with a lot of varied businesses, but obviously, we were in the same building with the Washington Post, and I had front row seat to see the digital transformation of the media industry. >> John: Yeah, we-- >> And how we responded. >> Yeah, I want to dig into that because I think that illustrates kind of a lot what's happening now, we're seeing with cloud computing. Obviously, Cloud 1.0 and the rise of Amazon public cloud. Clearly, check, done that, a lot of companies, startups go there. Why would you provision a data center? You're a startup, you're crazy, but at some point, you can have a data center. Now, hybrid cloud's important. Devops, the application development market, building your own stack, is shifting now. It seems like the old days, but upside down. It's flipped around, where applications are in charge, data's critical for the application, infrastructure's now elastic. Unlike the old days of here's your infrastructure. You're limited to what you can run on it based on the infrastructure. >> Right. >> What's your thoughts on that? >> My thoughts are that, I'm a very, as my title suggests, data-centric person. So I think about everything data first. We were in a time when cloud-first is becoming old, and we are now moving into data-first because what's happening in the marketplace is the ability, the capability, of data analytics has reached a point where prediction, in any aspect of a business, has become really inexpensive. So empowering employees with prediction machines, whether you call them bots, or you call them analytics, or you call them machine learning, or AI, has become really inexpensive, and so I'm thinking more of applications, which are built data-out instead of data-in, which is you build process and you capture data, and then you decide, oh, maybe I should build some reporting. That's what we used to do. Now, you need to start with what's the data I have got? What's the data I need? What's the data I can get? We were just talking about, everybody needs a data monetization strategy. People don't realize how much asset is sitting in their data and where to monetize it and how to use it. >> It's interesting. I mean, I got my computer science degree in the 80s and one of the tracks I got a degree in was database, and let's just say that my main one was operating system. Database was kind of the throwaway at that time. It wasn't considered a big field. Database wasn't sexy at all. It was like, database, like. Now, if you're a database, you're a data guru, you're a rock star. The world has changed, but also databases are changing. It used to be one centralized database rules the world. Oracle made a lot of money with that, bought all their competitors. Now you have open source came into the realm, so the world of data is also limited by where the data's stored, how the data is retrieved, how the data moves around the network. This is a new dynamic. How do you look at that because, again, lagging in business has a lot to do with the data, whether it's in an application, that's one thing, but also having data available, not necessarily in real time, but if I'm going to work on something, I want the data set handy, which means I can download it or maybe get real-time. What's your thoughts on data as an element in all that moving around? >> So I think what you're talking about is still data analytics. How do I get insights about my business? How do I make decisions using data in a better way? What flexibility do I need? So you talk about open source, you think about MongoDB and those kind of databases. They give you a lot of flexibility. You can develop interesting insights very quickly, but I think that is still very much thinking about data in an old-school kind of way. I think what's happening now is we're teaching algorithms with data. So data is actually the software, right? So you get an open source algorithm. I mean Google and everybody else is happy to open source their algorithms. They're all available for free. But what, the asset is now the data, which means how you train your algorithm with your data, and then now, moving towards deploying it on the edge, which is you take an algorithm, you train it, then you deploy it on the edge in an IoT kind of environment, and now you're doing decision-making, whether it's self-driving cars, I mean those are great examples, but I think it's going down into very interesting spaces in enterprise, which is, so we have to all think about software differently because, actually, data is a software. >> That's an interesting take on it, and I love that. I mean I wrote a blog post in 2007 when we first started playing with the, in looking at the network effects on social media and those platforms was, I wrote a post, it was called Data is the New Development Kit. Development kit was what people did back then. They had a development kit and they would download stuff and then code, but the idea was is that data has to be part of the runtime and the compilation of, as software acts, data needs to be resident, not just here's a database, access it, pull it out, use it, present it, where data is much more of a key ingredient into the development. Is that kind of what you're getting at? >> Yes. >> Notion of-- >> And I think we're moving from the age of arithmetic-based machines, which is we put arithmetic onto chips, and we then made general-purpose chips, which were used to solve a huge amount of problems in the world. We're talking about, now, prediction machines on a chip, so you think about algorithms that are trained using data, which are going to be available on chips. And now you can do very interesting algorithmic work right on the edge devices, and so I think a lot of businesses, and I've seen that recently at GameStop, I think business leaders have a hard time understanding the change because we have moved from process-centric, process automation, how can I do it better? How can I be more productive? How can I make better decisions? We have trained our business partners on that kind of thinking, and now we are starting to say, no, no, no, we've got something that's going to help you make those decisions. >> It's interesting, you mentioned GameStop. Obviously, well-known, my sons are all gamers. I used to be a gamer back before I had kids, but then, can't keep up anymore. Got to be on that for so long, but GameStop was a retail giant in gaming. Okay, when they had physical displays, but now, with online, they're under pressure, and I had interviewed, again, at an Amazon event, this Best Buy CIO, and he says, "We don't compete with price anymore. "If they want to buy from Amazon, no problem, "but our store traffic is off the charts. "We personalize 50,000 emails a day." So personalization became their strategy, it was a data strategy. This is a user experience, not a purchase decision. Is this how you guys are thinking about it at GameStop? >> I think retail, if you look at the segment per se, personalization, Amazon obviously led the way, but it's obvious that personalization is key to attract the customer. If I don't know what games you play, or if I don't know what video you watched a little while ago, about which game, then I'm not offering you the product that you are most prone or are looking for or what you want to buy, and I think that's why personalization is key. I think that's-- >> John: And data drives that, and data drives that. >> Data drives that, and for personalization, if you look at retail, there's customer information. You need to know the customer. You need to know, understand the customer preferences, but then there's the product, and you need to marry the two. And that's where personalization comes into play. >> So I'll get your thoughts. You have, obviously, a great perspective on how tech has been built and now working on some real cutting-edge, clear view on what the future looks like. Totally agree with you, by the way, on the data. There's kind of an old guard/new guard, kind of two sides of the street, the winners and the losers, but hey, look, I think the old guard, if they don't innovate and become fresh and new and adopt the modern things that need to attract the new expectations and new experiences from their customers, are going to die. That being said, what is the success formula, because some people might say, hey, I'm data-driven. I'm doing it, look at me, I'm data. Well, not really. Well, how do you tell if someone's really data-driven or data-centric? What's the difference? Is there a tell sign? >> I think when you say the old guard, you're talking about companies that have large assets, that have been very successful in a business model that maybe they even innovated, like GameStop came up with pre-owned games, and for the longest of times, we've made huge amount of revenue and profit from that segment of our business. So yes, that's becoming old now, but I think the most important thing for large enterprises at least, to battle the incumbent, the new upstarts, is to develop strategies which are leveraging the new technologies, but are building on their existing capability, and that's what I drive at GameStop. >> And also the startups too, that they were here in a venture capital firm, we're at Mayfield Fund, doing this program, startups want to come and take a big market down, or come in on a narrow entry and get a position and then eat away at an incumbent. They could do it fast if they're data-centric. >> And I think it's speed is what you're talking about. I think the biggest challenge large companies have is an ability to to play the field at the speed of the new upstarts and the firms that Mayfield and others are investing in. That's the big challenge because you see this, you see an opportunity, but you're, and I saw that at the Washington Post. Everybody went to meetings and said, yes, we need to be digital, but they went-- >> They were talking. >> They went back to their desk and they had to print a paper, and so yes, so we'll be digital tomorrow, and that's very hard because, finally, the paper had to come out. >> Let's take us through the journey. You were the CTO, VP of Technology, Graham Holdings, Washington Post, they sold it to Jeff Bezos, well-documented, historic moment, but what a storied company, Washington Post, local paper, was the movie about it, all the historic things they've done from a reporting and journalism standpoint. We admire that. Then they hit, the media business starts changing, gets bloated, not making any money, online classifieds are dying, search engine marketing is growing, they have to adjust. You were there. What was the big, take us through that journey. >> I think the transformation was occurring really fast. The new opportunities were coming up fast. We were one of the first companies to set up a website, but we were not allowed to use the brand on the website because there was a lot of concern in the newsroom that we are going to use or put the brand on this misunderstood, nearly misunderstood opportunity. So I think it started there, and then-- >> John: This is classic old guard mentality. >> Yes, and it continued down because people had seen downturns. It's not like media companies hadn't been through downturns. They had, because the market crashes and we have a recession and there's a downturn, but it always came back because-- >> But this was a wave. I mean the thing is, downturns are economic and there's business that happens there, advertisers, consumption changes. This was a shift in their user base based upon a technology wave, and they didn't see it coming. >> And they hadn't ever experienced it. So they were experiencing it as it was happening, and I think it's very hard to respond to a transformation of that kind in a very old-- >> As a leader, how did you handle that? Give us an example of what you did, how you make your mark, how do you get them to move? What were some of the things that were notable moments? >> I think the main thing that happened there was that we spun out washingtonpost.com. So it became an independent business. It was actually running across the river. It moved out of the corporate offices. It went to a separate place. >> The renegades. >> And they were given-- >> John: Like Steve Jobs and the Macintosh team, they go into separate building. >> And we were given, I was the CTO of the dotcom for some time while we were turning over our CTO there, and we were given a lot of flexibility. We were not held accountable to the same level. We used the, obviously, we used-- >> John: You were running fast and loose. >> And we were, yes, we had a lot of flexibility and we were doing things differently. We were giving away the content in some way. On the online side, there was no pay wall. We started with a pay wall, but advertising kind of was so much more lucrative in the beginning, that the pay wall was shut down, and so I think we experimented a lot, and I think where we missed, and a lot of large companies miss, is that you need to leave your existing business behind and scale your new business, and I think that's very hard to do, which is, okay, we're going to, it's happening at GameStop. We're no longer completely have a control of the market where we are the primary source of where, you talk about your kids, where they go to get their games. They can get the games online and I think-- >> It's interesting, people are afraid to let go because they're so used to operating their business, and now it has to pivot to a new operating model and grow. Two different dynamics, growth, operation, operating and growing. Not all managers have that growth mindset. >> And I think there's also an experience thing. So most people who are in these businesses, who've been running these businesses very successfully, have not been watching what's happening in technology. And so the technology team comes out and says, look, let me show you what we can do. I think there has to be this open and very, very candid discussion around how we are going to transform-- >> How would you talk about your peer, developed peers out there, your peers and other CIOs, and even CISOs on the security side, have been dealing with the same suppliers over, and in fact, on the security side, the supplier base is getting larger. There's more tools coming out. I mean who wants another tool? So platform, tool, these are big decisions being made around companies, that if you want to be data-centric, you want to be a data-centric model, you got to understand platforms, not just buying tools. If you buy a hammer, they will look like a nail, and you have so many hammers, what version, so platform discussions come in. What's your thoughts on this? Because this is a cutting-edge topic we've been talking about with a lot of senior engineering leaders around Platform 2.0 coming, not like a classic platform to... >> Right, I think that each organization has to leverage or build their, our stack on top of commodity platforms. You talked about AWS or Azure or whatever cloud you use, and you take all their platform capability and services that they offer, but then on top of that, you structure your own platform with your vertical capabilities, which become your differentiators, which is what you take to market. You enable those for all your product lines, so that now you are building capability, which is a layer on top of, and the commodity platforms will continue to bite into your platform because they will start offering capabilities that earlier, I remember, I started at this company called BrassRing, recruitment automation. One of the first software-as-a-service companies, and I, we bought a little company, and the CTO there had built a web server. It was called, it was his name, it was called Barrett's Engine. (chuckles) And so-- >> Probably Apache with something built around it. >> So, in those days, we used to build our own web servers. But now today, you can't even find an engineer who will build a web server. >> I mean the web stack and these notions of just simple Web 1.0 building blocks of change. We've been calling it Cloud 2.0, and I want to get your thoughts on this because one of the things I've been riffing on lately is this, I remember Marc Andreessen wrote the famous article in Wall Street Journal, Software is Eating the World, which I agree with in general, no debate there, but also the 10x Engineer, you go into any forum online, talking about 10x Engineers, you get five different opinions, meaning, a 10x Engineer's an engineer who can do 10 times more work than an old school, old classical engineer. I bring this up because the notion of full stack developer used to be a real premium, but what you're talking about here with cloud is a horizontally scalable commodity layer with differentiation at the application level. That's not full stack, that's half stack. So you think the world's kind of changing. If you're going to be data-centric, the control plane is data. The software that's domain-specific is on top. That's what you're essentially letting out. >> That's what I'm talking about, but I think that also, what I'm beginning to find, and we've been working on a couple of projects, is you put the data scientists in the same room with engineers who write code, write software, and it's fascinating to see them communicate and collaborate. They do not talk the same language at all. >> John: What's it like? Give us a mental picture. >> So a data scientist-- >> Are they throwing rocks at each other? >> Well, nearly, because the data scientists come from the math side of the house. They're very math-oriented, they're very algorithm-oriented. Mathematical algorithms, whereas software engineers are much more logic-oriented, and they're thinking about scalability and a whole lot of other things, and if you think about, a data scientist develops an algorithm, it rarely scales. You have to actually then hand it to an engineer to rewrite it in a scalable form. >> I want to ask you a question on that. This is why I got you and you're an awesome guest. Thanks for your insights here, and we'll take a detour into machine learning. Machine learning really is what AI is about. AI is really nothing more than just, I love AI, it gets people excited about computer science, which is great. I mean my kids talk about AI, they don't talk about IoT, which is good that AI does that, but it's really machine learning. So there's two schools of thought on machine. I call it the Berkeley school on one end, not Berkeley per se but Berkeley talks about math, machine learning, math, math, math, and then you have other schools of thought that are on cognition, that machine learning should be more cognitive, less math-driven, spectrum of full math, full cognition, and everything in between. What's your thoughts on the relationship between math and cognition? >> Yeah, so it's interesting. You get gray hair and you kind of move up the stack, and I'm much more business-focused. These are tools. You can get passionate about either school of thought, but I think that what that does is you lose sight of what the business needs, and I think it's most important to start with what are we here trying to do, and what is the best tool? What is the approach that we should utilize to meet that need? Like the other day, we were looking at product data from GameStop, and we know that the quality of data should be better, but we found a simple algorithm that we could utilize to create product affinity. Now whether it's cognition or math, it doesn't matter. >> John: The outcome's the outcome. >> The outcome is the outcome, and so-- >> They're not mutually exclusive, and that's a good conversation debate but it really gets to your point of does it really matter as long as it's accurate and the data drives that, and this is where I think data is interesting. If you look at folks who are thinking about data, back to the cloud as an example, it's only good as what you can get access to, and cybersecurity, the transparency issue around sharing data becomes a big thing. Having access to the data's super important. How do you view that for, as CIOs, and start to think about they're re-architecting their organizations for these digital transformations. Is there a school of thought there? >> Yes, so I think data is now getting consolidated. For the longest time, we were building data warehouses, departmental data warehouses. You can go do your own analytics and just take your data and add whatever else you want to do, and so the part of data that's interesting to you becomes much more clean, much more reliable, but the rest, you don't care much about. I think given the new technologies that are available and the opportunity of the data, data is coming back together, and it's being put into a single place. >> (mumbles) Well, that's certainly a honeypot for a hacker, but we'll get that in a second. If you and I were doing a startup, we say, hey, let's, we've got a great idea, we're going to build something. How would we want to think about the data in terms of having data be a competitive advantage, being native into the architecture of the system. I'll say we use cloud unless we need some scale on premise for privacy reasons or whatever, but we would, how would we go to market, and we have an app, as apps defined, great use case, but I want to have extensibility around the data, I don't want to foreclose any future options, How should I think about my, how should we think about our data strategy? >> Yes, so there was a very interesting conversation I had just a month ago with a friend of mine who's working at a startup in New York, and they're going to build a solution, take it to market, and he said, "I want to try it only in a small market "and learn from it," and he's going very old school, focus groups, analytics, analysis, and I sat down, we sat at Grand Central Station, and we talked about how, today, he should be thinking about capturing the data and letting the data tell him what's working and what's not working, instead of trying to find focus groups and find very small data points to make big decisions. He should actually utilize the target, the POC market, to capture data and get ready for scale because if you want to go national after having run a test in... >> Des Moines, Iowa. >> Part of New York or wherever, then you need to already have built the data capability to scale that business in today's-- >> John: Is it a SaaS business? >> No, it's a service and-- >> So he can instrument it, just watch the data. >> And yes, but he's not thinking like that because most business people are still thinking the old way, and if you look at Uber and others, they have gone global at such a rapid pace because they're very data-centric, and they scale with data, and they don't scale with just let's go to that market and then let's try-- >> Yeah, ship often, get the data, then think of it as part of the life cycle of development. Don't think it as the old school, craft, launch it, and then see how it goes and watch it fail or succeed, and know six months later what happened, know immediately. >> And if you go data-centric, then you can turn the R&D crank really fast. Learn, test and learn, test and learn, test and learn at a very rapid pace. That changes the game, and I think people are beginning to realize that data needs to be thought about as the application and the service is being developed, because the data will help scale the service really fast. >> Data comes into applications. I love your line of data is the new software. That's better than the new oil, which has been said before, but data comes into the app. You also mentioned that app throws off data. >> Yuvi: Yes. >> We know that humans have personal, data exhaust all the time. Facebook made billions of dollars on our exhaust and our data. The role of data in and out of the application, the I/O of the application, is a new concept, you brought that up. I like that and I see that happening. How should we capture that data? This used to be log files. Now you got observability, all kinds of new words kind of coming into this cloud equation. How should people think about this? >> I think that has to be part of the design of your applications, because data is application, and you need to design the application with data in mind, and that needs to be thought of upfront, and not later. >> Yuvi, what's next for you? We're here in Sand Hill Road, VC firm, they're doing a lot of investments, you've got a great project with GameStop, you're advising startups, what's going on in your world? >> Yes, so I'm totally focused, as you probably are beginning to sense, on the opportunity that data is enabling, especially in the enterprise. I'm very interested in helping business understand how to leverage data, because this is another major shift that's occurring in the marketplace. Opportunities have opened up, prediction is becoming cheap and at scale, and I think any business runs on their capability to predict, what is the shirt I should buy? How many I should buy? What color should I buy? I think data is going to drive that prediction at scale. >> This is a legit way that everyone should pay attention to. All businesses, not just one-- >> All businesses, everything, because prediction is becoming cheap and automated and granular. That means you need to be able to not just, you need to empower your people with low-level prediction that comes out of the machines. >> Data is the new software. Yuvi, thanks so much for great insight. This is theCUBE conversation. I'm John Furrier here at Sand Hill Road at the Mayfield Fund, for the People First Network series. Thanks for watching. >> Yuvi: Thank you. (bright electronic music)