 It's time to value, it's elapsed time, it's compressing that elapsed time, it's accelerating business outcomes, which compared to the TCO stuff is like telephone numbers, isn't it? Oh it is, and part of what you're seeing is the new technologies in open source are really breaking down walls of spark technology, really started to look at how they could help solution developers bring analytics into their applications quickly and within that timeframe, time to value and cost functions. So it's not a data person thinking about how they use analytics or an analytics person thinking it's a solution. And it's back to your point, all these new workloads need to do this and I think the spark ecosystem really started to hone in on that and offer a good value proposition. Would it be fair to characterize this acceleration in the use of analytics on two fronts? One is how agile you are in improving the analytics and then the other is how fast you can apply it at the time of transaction? Absolutely, and we're seeing this in the way some of the business models are done that sometimes before you use spark it would take a week to update these, data scientists involved, but then you need to apply it quickly. So this is part of our Demantic group, works with retailers online, looking at a billion records, would take them about a week to put the data database, do the things about you said get the algorithm right and then deploy it. Now with Hadoop they're looking at doing that in a few hours. That's kind of this continuous evolution of if I've got a breakthrough, I've found something, I need to deploy it fast. I can't wait a week to go do that. We were talking earlier about something that's sort of on the edge of not quite mainstream but experimental notebooks where you bring together different analytic types in a not necessarily a programmer medium, although you can drop into that, but sort of almost like accessible like a spreadsheet. That's true. And to what extent does that democratize the ability to build these predictive models? Oh, huge. And if you looked at notebooks two years ago, two and a half years ago, there were nothing on GitHub. Now there's probably, I think the last count we looked was over 200,000. So you now you have people coming out of college, graduates that know how to use notebooks. And notebooks for folks that don't know they're a web based interactive collaborative way. It's a wiki with cells so you can run code and see visualizations. Think about as a live analytics document. And that's fascinating because the types of questions business are asking are very open-ended. I think I see something, you sit with me data scientist and show me. Am I right? And if it is, I'll tell you where to go next. So our idea of an application isn't just inputs and outputs anymore. It's a spreadsheet of what ifs in many cases that then you find the business value for. So is your job essentially to be a trend spotter and then figuring out how to create value from those trends or try to identify where IBM can create value? Trend spotter is a good way to put it, but we put a lot of skin in the game. And so internally when we start to spot these, we'll socialize it, but who cares? Some smart people see stuff all the time. You're not just an observer. That's right. Customers. So we go out and we get customer examples. And if I can get maybe a dozen customer examples, then we know this is a real trend that's going to have some weight behind it. If I can't, then I have to tell folks, this was interesting, but it doesn't look like the adoption rate is going to be something we interested in. So I wonder if we could take two examples. And it may not have been directly in your area, but just as an observer of IBM, there's two major things going on. One is cloud. And the economics of cloud are clearly changing. Provisioning is becoming free. It's automated. It's like software economics. And then cognitive. Now, I feel as though IBM was more reactionary to cloud, but very proactive with cognitive and essentially inventing an entire new industry where cloud is like, all right, you know, it's kind of hard times and hardware. Steve Mills never liked hardware anyway. I always joke to him. Hardware, mainframe hardware is good. The rest of it, forget it. Storage in x86, but we joke all the time. But my point is, those were some tectonic shifts in the industry that you had to respond to. Whereas cognitive, I'd say you're creating a whole new industry. What role did people like yourself play in creating those opportunities? Well, let's see. The first of a kind that came out of the Jeopardy project was a research project. And my team took that as the first one and then took the next 14 months and commercialized it to give it and then created the Watson division. So my team had all the analytic skills, the big data skills, and this was a great project. Wonderful to partner with research and go do that. And learn about kind of the way that we could think about cognitive, the different areas, the different ways to look at data. And help the division get going. We're also the group that started Bluemix. So we partnered with others and I've been very small teams. It was called Initially Cloud OE and then Bluemix. You know, less than 20 people. And it caught such a wave in IBM and with customers. It was, you know, beyond what we ever expected. Now the whole team, like you say, reaction, we had to grow. It was just phenomenal on what was happening around it from a cloud standpoint. But back to my point, the way we socialize with both internal lecturing is lots of demos. Lots of ways to get customer reaction to say, you know, are we doing the good thing or did we miss the mark? The one thing that the capabilities of the cognitive computing services, Watson, looks very alluring. But it's hard even to someone who's moderately technical to picture what that API platform looks like. Can those be exposed in these more accessible notebooks the way you're exposing the spark analytical and other capabilities? Absolutely. In fact, we use notebooks when we develop Watson to hone the analytics down. And that was kind of our first experience with early, early notebooks on, you know, this is something we think could be productive for other people. So that was four years ago. And it was just a, you know, Python was doing some interesting things with it. But, yeah, it's, you can think about natural language processing. You know, how do I use a notebook and be able to call out to the different APIs to go do that or Watson learning? You know, how do I do that so I can train it better? My notebooks are very good for helping explore the way that you want to do training and documenting it. So businesses, others can see and understand or create new notebooks off the ones that you already have. So it's got to be rewarding to see the outcomes affect, you know, the human condition. Let's talk about Watson a little bit more. So, you know, it's like I said, we've been torturing ourselves for 15 years about ringing more costs out of IT. And the industry's done a pretty good job, but it just gets kind of rote. Look at something like Watson and specifically in healthcare. How do you see what's your vision rod as to how Watson, other, you know, parts of the ecosystem, maybe other cognitive systems affect the human condition? What's your vision there? Well, part of this is, you know, we're very happy to help, you know, a small team get Watson started and where it is today. We, again, spend a lot of time talking with oncologists and others about what was their vision for utilizing Watson. And, you know, the huge amounts of information, new journal articles and things, but also they wanted to be able to, you know, as we're doing, create an API so you can create new applications. Now, I think that's the exciting part of Watson is, how do I look at, you know, new vertical applications that IBM can't create? And, you know, you want to open up that ecosystem to do that. So, you know, I think that's the important part of looking at this corpus of information that Watson's got for healthcare and then having people explore, like, you know, the latest ones on, you know, meat. You know, here's what it causes, but, you know, I really read the article and said, well, it's 50 grams or more. Well, it's, you know, I need more information. All of us have the time to go read all this and have, you know, figured out what it is, or does Watson help us do those type of things when we're making a meal? Well, it strikes me that Watson was sort of remarkable in the first application in jeopardy, but it was also intimidating in terms of trying to think, well, as a corporate customer or whatever, or as an independent software vendor, how would I take advantage of that platform or service? And if you're exposing it now at something that's like a workspace, it would seem like you're breaking the fundamental bottleneck in terms of getting democratizing access to this. I think that's right. And part of it was the initial system performance was so key on the question and answer side of it. And as Mike talked about this morning, now they're looking at a lot of different solution spaces. So now I can break out the natural language processing interface. I can bring out the Watson machine learning the cognitive pieces to it. And so it's a good maturation over time of what, you know, driven by the types of applications people want to do. How do we break those APIs out slowly? And you saw it might be very careful on, it's not thousands, it's not waves of APIs, it's 30-some, and then they're going to grow it to 60-some. That seems like a small number. And the reason we want it small is how do we find that right sweet spot for developers to do the right value but not get swamped into much details? And how far are you in terms of testing the stability and the richness of the APIs with developers? All the time. And is that your group? No, it's not my group. That's Rob High's group and in the Watson group to do that. They're the ones testing it. We're one of the consumers now. We're working with customers on that as we put it together. People say, gee, can we put some Watson analytics into this? Well, great. How do we want to utilize that as a new service inside of an application? Okay. So, Rod, if you had to create a hall of fame of IBM innovations, it'd be a big wall, I guess. But what are some of the ones that stand out for you? Oh, geez. Well, you know, the superstar, one of the superstars in my mind is Fred Brooks. Fred Brooks? Yes. And I think Fred is still teaching in Chapel Hill if I'm not mistaken. And he wrote another book? I think he probably wrote another book. And designed. Yeah, he's just, one is he's so down to earth. I've met him a number of times, but he's so insightful too. You know, he just has a way to get that BB King note of getting right to the point and understanding the technology and communicating value points in it. So, to me, you know, Fred set the stage for a lot of the computing we're doing and trying to be very pragmatic but visionary at the same time. So, the system 360, the OS 360 is number one. I think so. And it's still... Don't you think? I mean, sort of got it all started. It got it started and it's still alive today. People keep saying, well, it can't last and guess what? It seems to have a pretty good... Virtualization. That's right. You know, it's still there. It's still there. That's right. You know, it's... DB2 or the relational... Relational came around quite a long time after its research started. That's one that's interesting where we innovated. We didn't take advantage of it. Oracle did a much better job than we did at that. Yeah. So, I think that, you know, probably my Hall of Fame starts with Fred Brooks on that. And then where do you think Watson will fit in that spectrum? I think Watson's going to be extremely high. And I think the way we've harnessed cognitive, how we're thinking about the future of how computers interact and gather information, you know, building these ease of use applications with it, robots, we saw today. You know, there's things... I'm getting ready to open source to how do we hook JavaScript into Spark because robots are a lot of being done in JavaScript. How do I do those things so people can, you know, be able to use machine learning, cognitive, and other things? I mean, robots they can't learn doesn't seem very interesting. Are there... You know, we saw that we saw Pepper up on stage today. Yes. What kinds of things are humans going to be able to do that robots won't be able to do? Climb stairs, I suppose. It's probably hard for a robot to climb stairs right now. And he's saying, no, just stay tuned. We didn't think that autonomous cars were a possibility a decade ago. Five years ago. No. No, I... You know, with... As Mike was talking about, and I think, you know, Bob was talking about kind of this, you know, dark web. There's lots of information out there that we're not really doing a lot with. Nest, you know, being able to adjust to different rooms and turn on. You know, we can collect information. I think that's where we want the intelligence to go is to help us. Like Watson, when we talked to doctors, Watson said, you know, I want a customer, I want a relationship with my patient. I don't want a relationship with Watson. Watson's there to help me. So I need to talk to you, you know, that personal interaction, look you in the eye to kind of, you know, what's your condition. Then I'll look to Watson to go help me with this. And I think computers will get smarter as they can, you know, look at what you're doing. It's like, you know, different types of moves that you're in, or if looking at your complexion to decide if you might have some oncoming disease or some, how do we help doctors do that and how do we help ourselves? Amazing times, Rod. We'll have to leave it there. Thanks very much for coming back on theCUBE. Really a pleasure. Thank you. Great to talk to you. All right. Keep it right there. We'll be back with our next guest right after this. Live from IBM Insight 2015. Right back.