WEBVTT 00:00:00.000 --> 00:00:02.560 align:middle line:84% The following content is provided under a Creative 00:00:02.560 --> 00:00:03.920 align:middle line:90% Commons license. 00:00:03.920 --> 00:00:06.360 align:middle line:84% Your support will help MIT OpenCourseWare 00:00:06.360 --> 00:00:10.000 align:middle line:84% continue to offer high quality educational resources for free. 00:00:10.000 --> 00:00:13.280 align:middle line:84% To make a donation or view additional materials 00:00:13.280 --> 00:00:17.240 align:middle line:84% from hundreds of MIT courses, visit MIT OpenCourseWare 00:00:17.240 --> 00:00:20.577 align:middle line:90% at ocw.mit.edu. 00:00:20.577 --> 00:00:22.160 align:middle line:84% IRVING WLADAWSKY-BERGER: Last week, we 00:00:22.160 --> 00:00:28.000 align:middle line:84% started talking about execution in the marketplace, 00:00:28.000 --> 00:00:40.520 align:middle line:84% and we talked about e-business and how a major part of what 00:00:40.520 --> 00:00:47.440 align:middle line:84% we did in IBM to help formulate a strategy that we could execute 00:00:47.440 --> 00:00:56.380 align:middle line:84% was to leverage our base, which included base of talent, 00:00:56.380 --> 00:01:01.320 align:middle line:84% the base of clients, existing products with the new stuff. 00:01:01.320 --> 00:01:07.420 align:middle line:84% And that the way we framed it, and it was a very good framing, 00:01:07.420 --> 00:01:07.980 align:middle line:90% I think. 00:01:07.980 --> 00:01:10.840 align:middle line:84% Well, it turned out to be pretty successful in the marketplace, 00:01:10.840 --> 00:01:16.960 align:middle line:84% is for IBM, the internet, World Wide Web, and e-business 00:01:16.960 --> 00:01:21.600 align:middle line:84% meant leverage the internet to make what you 00:01:21.600 --> 00:01:24.200 align:middle line:90% have that much more valuable. 00:01:24.200 --> 00:01:25.420 align:middle line:90% And that's our framing. 00:01:25.420 --> 00:01:28.620 align:middle line:84% Other people would have framed it different, 00:01:28.620 --> 00:01:31.420 align:middle line:84% saying, what you have goes away, get rid of it. 00:01:31.420 --> 00:01:33.020 align:middle line:90% That's a fine way to frame it. 00:01:33.020 --> 00:01:35.440 align:middle line:84% You have to convince your customers. 00:01:35.440 --> 00:01:44.480 align:middle line:84% And we talked about some of the key aspects 00:01:44.480 --> 00:01:48.600 align:middle line:84% of organizing this strategy, the market strategy, which 00:01:48.600 --> 00:01:52.360 align:middle line:84% is balance between proprietary and open, in-house versus 00:01:52.360 --> 00:01:53.840 align:middle line:90% partnerships. 00:01:53.840 --> 00:01:56.400 align:middle line:84% What are going to be your key offerings? 00:01:56.400 --> 00:01:59.630 align:middle line:84% How are you going to measure your progress? 00:01:59.630 --> 00:02:04.410 align:middle line:84% And those are very key organizational things. 00:02:04.410 --> 00:02:07.570 align:middle line:84% But we also talked about a very important subject, 00:02:07.570 --> 00:02:10.289 align:middle line:84% that when you're talking about a disruptive 00:02:10.289 --> 00:02:15.130 align:middle line:84% technology, like the internet, with a very 00:02:15.130 --> 00:02:19.650 align:middle line:84% big potential market-- and remember, 00:02:19.650 --> 00:02:24.050 align:middle line:84% we talked about the goddamn internet is just 00:02:24.050 --> 00:02:26.490 align:middle line:90% this incredible system. 00:02:26.490 --> 00:02:29.810 align:middle line:84% And as you know, if somebody said, well, where in this stack 00:02:29.810 --> 00:02:31.650 align:middle line:90% does the internet play? 00:02:31.650 --> 00:02:36.410 align:middle line:84% What makes it so disruptive is that it plays up and down. 00:02:36.410 --> 00:02:45.470 align:middle line:84% The more you go across, how do you focus your efforts? 00:02:45.470 --> 00:02:49.750 align:middle line:84% If you have something that complicated, if you just said, 00:02:49.750 --> 00:02:52.730 align:middle line:84% well, I want to go take on the world 00:02:52.730 --> 00:02:56.790 align:middle line:84% and I want to go and transform applications, no, no. 00:02:56.790 --> 00:02:58.650 align:middle line:90% Take it easy. 00:02:58.650 --> 00:03:00.530 align:middle line:90% We may get there. 00:03:00.530 --> 00:03:05.090 align:middle line:84% But the first question is, what are you going to do first? 00:03:05.090 --> 00:03:08.770 align:middle line:90% How do you get to first base? 00:03:08.770 --> 00:03:10.430 align:middle line:90% You know what I'm talking about. 00:03:10.430 --> 00:03:15.130 align:middle line:84% Let's get there one step at a time. 00:03:15.130 --> 00:03:22.490 align:middle line:84% And we will talk today about customer-based experimentation. 00:03:22.490 --> 00:03:26.870 align:middle line:84% And first, I'll talk about the work of Tom Key. 00:03:26.870 --> 00:03:30.090 align:middle line:84% And then we'll talk about Eric von Hippel's work. 00:03:30.090 --> 00:03:34.370 align:middle line:84% And Eric was really nice in saying he'll join us at 2:00 00:03:34.370 --> 00:03:38.090 align:middle line:84% o'clock, which is wonderful, because he will then join us 00:03:38.090 --> 00:03:41.450 align:middle line:90% in the discussions. 00:03:41.450 --> 00:03:47.370 align:middle line:84% I can't say enough about experimenting in the marketplace 00:03:47.370 --> 00:03:50.690 align:middle line:84% as a way to figure out what to do. 00:03:50.690 --> 00:03:55.890 align:middle line:84% If you're just doing pure technology invention, 00:03:55.890 --> 00:03:59.090 align:middle line:84% clearly you do that in the lab, and clearly you 00:03:59.090 --> 00:04:01.250 align:middle line:90% experiment in the lab. 00:04:01.250 --> 00:04:03.250 align:middle line:84% And if you're doing products, you 00:04:03.250 --> 00:04:06.450 align:middle line:84% would clearly prototype-build, et cetera. 00:04:06.450 --> 00:04:14.410 align:middle line:84% But innovation, especially now, more and more, and what 00:04:14.410 --> 00:04:16.050 align:middle line:84% we're talking about in this class, 00:04:16.050 --> 00:04:18.890 align:middle line:84% is how do you take stuff from the lab 00:04:18.890 --> 00:04:21.510 align:middle line:84% or from somebody's lab-- don't have to be your lab. 00:04:21.510 --> 00:04:25.350 align:middle line:84% Could be MIT's lab, Stanford labs, anybody's labs-- 00:04:25.350 --> 00:04:28.250 align:middle line:90% take it into the marketplace? 00:04:28.250 --> 00:04:32.430 align:middle line:84% Now, the marketplace is a wonderful concept, 00:04:32.430 --> 00:04:34.970 align:middle line:90% but it's a cosmic concept. 00:04:34.970 --> 00:04:37.830 align:middle line:84% So you need to start narrowing down, 00:04:37.830 --> 00:04:42.610 align:middle line:84% what the hell do you mean by the marketplace? 00:04:42.610 --> 00:04:47.610 align:middle line:84% In other words, you have to tell me where is first base, 00:04:47.610 --> 00:04:51.210 align:middle line:84% so that we know where you're heading. 00:04:51.210 --> 00:04:57.460 align:middle line:84% And a critical part of finding first base 00:04:57.460 --> 00:05:01.060 align:middle line:90% is marketplace experimentation. 00:05:01.060 --> 00:05:05.780 align:middle line:84% Working very, very closely with clients. 00:05:05.780 --> 00:05:14.780 align:middle line:84% Now, in the past, marketplace experimentation 00:05:14.780 --> 00:05:17.500 align:middle line:90% was just too expensive. 00:05:17.500 --> 00:05:20.020 align:middle line:90% There was very difficult to do. 00:05:20.020 --> 00:05:23.520 align:middle line:84% First, it was very difficult to build anything in the lab, 00:05:23.520 --> 00:05:26.180 align:middle line:90% then to build a prototype. 00:05:26.180 --> 00:05:31.300 align:middle line:84% Remember, most of the channels to market, 00:05:31.300 --> 00:05:35.260 align:middle line:84% most of the dealings with clients-- not most, all of them 00:05:35.260 --> 00:05:36.920 align:middle line:90% were people oriented. 00:05:36.920 --> 00:05:40.260 align:middle line:84% There was no technology to help us do that. 00:05:40.260 --> 00:05:42.960 align:middle line:90% So it was just very expensive. 00:05:42.960 --> 00:05:45.140 align:middle line:90% So you had to be very careful. 00:05:45.140 --> 00:05:48.140 align:middle line:84% And often, the market experimentation 00:05:48.140 --> 00:05:53.340 align:middle line:84% was another way of you announce your product 00:05:53.340 --> 00:05:56.700 align:middle line:84% and you hope and pray the marketplace likes it, 00:05:56.700 --> 00:05:58.580 align:middle line:90% or your service. 00:05:58.580 --> 00:06:00.800 align:middle line:90% But this is changing. 00:06:00.800 --> 00:06:04.500 align:middle line:84% Companies now have an opportunity 00:06:04.500 --> 00:06:11.260 align:middle line:84% to innovate and experiment in a whole new way. 00:06:11.260 --> 00:06:14.680 align:middle line:84% What are some of the things companies can do to experiment? 00:06:14.680 --> 00:06:17.760 align:middle line:90% Well, here is a long list. 00:06:17.760 --> 00:06:21.880 align:middle line:84% First, there is the internet itself, which as you know, 00:06:21.880 --> 00:06:26.420 align:middle line:84% is the world's biggest platform for innovation, experimentation, 00:06:26.420 --> 00:06:28.260 align:middle line:90% and collaboration. 00:06:28.260 --> 00:06:33.860 align:middle line:84% Then is the fact that IT systems are so inexpensive. 00:06:33.860 --> 00:06:38.100 align:middle line:84% I don't know if you know, but the amount 00:06:38.100 --> 00:06:42.780 align:middle line:84% of money that, let's say, brand-new ventures need 00:06:42.780 --> 00:06:46.780 align:middle line:84% to get going now is significantly less 00:06:46.780 --> 00:06:49.540 align:middle line:90% than during the dot com days. 00:06:49.540 --> 00:06:52.960 align:middle line:84% And a big part of that is that during the dot com days, 00:06:52.960 --> 00:06:55.980 align:middle line:84% in order to get computers, they needed to buy all this stuff, 00:06:55.980 --> 00:06:59.820 align:middle line:84% and that stuff was much, much more expensive. 00:06:59.820 --> 00:07:05.220 align:middle line:84% Today, computers, much to the chagrin of companies like IBM, 00:07:05.220 --> 00:07:07.580 align:middle line:90% are dirt cheap. 00:07:07.580 --> 00:07:11.620 align:middle line:84% And especially to get going-- and you can get software 00:07:11.620 --> 00:07:15.380 align:middle line:84% for practically nothing if you use LEMP and LAMP, 00:07:15.380 --> 00:07:17.840 align:middle line:90% and other open source stacks. 00:07:17.840 --> 00:07:22.860 align:middle line:84% So the cost of getting going is much less expensive. 00:07:22.860 --> 00:07:26.660 align:middle line:84% Even supercomputing, which before you had-- it 00:07:26.660 --> 00:07:28.660 align:middle line:90% was very expensive to access. 00:07:28.660 --> 00:07:30.780 align:middle line:90% You can do that much better. 00:07:30.780 --> 00:07:32.200 align:middle line:90% And on and on and on. 00:07:32.200 --> 00:07:36.740 align:middle line:84% So there is a whole set of new tools 00:07:36.740 --> 00:07:41.740 align:middle line:84% that very much enable you to experiment. 00:07:41.740 --> 00:07:49.150 align:middle line:84% And Tom Key makes the point that companies 00:07:49.150 --> 00:07:53.470 align:middle line:84% need to take that into account the opportunity 00:07:53.470 --> 00:07:58.130 align:middle line:84% for experimentation in their basic structure. 00:07:58.130 --> 00:08:00.710 align:middle line:84% That is, it's not like I go with the way I've been doing it 00:08:00.710 --> 00:08:06.250 align:middle line:84% for the last 50 years, and I'll bolt experimentation at the end. 00:08:06.250 --> 00:08:08.630 align:middle line:84% That's not the right way to do that. 00:08:08.630 --> 00:08:14.270 align:middle line:84% He makes the point that you want, from the beginning, 00:08:14.270 --> 00:08:18.230 align:middle line:84% to organize your efforts in terms 00:08:18.230 --> 00:08:21.510 align:middle line:90% of extensive experimentation. 00:08:21.510 --> 00:08:23.990 align:middle line:90% What is an example? 00:08:23.990 --> 00:08:27.550 align:middle line:84% Well, if you have a huge team that 00:08:27.550 --> 00:08:31.150 align:middle line:84% needs to do something for production, which maybe is what 00:08:31.150 --> 00:08:36.070 align:middle line:84% you need, maybe you want to have a small black belt 00:08:36.070 --> 00:08:41.030 align:middle line:84% team that does a rapid prototype that you can even 00:08:41.030 --> 00:08:44.830 align:middle line:84% start making available as an alpha or a testbed 00:08:44.830 --> 00:08:49.670 align:middle line:84% to start testing out the ideas, and then incrementally refine 00:08:49.670 --> 00:08:52.430 align:middle line:84% what you build so that you can start 00:08:52.430 --> 00:08:55.430 align:middle line:90% experimenting very quickly. 00:08:55.430 --> 00:09:00.590 align:middle line:84% I have to tell you, in the old days, people could bring, 00:09:00.590 --> 00:09:03.910 align:middle line:84% say at IBM and other companies, product plans that 00:09:03.910 --> 00:09:06.590 align:middle line:84% may have taken, I don't know, three to five years. 00:09:06.590 --> 00:09:12.310 align:middle line:84% Today, a plan that three to five years-- 00:09:12.310 --> 00:09:14.990 align:middle line:84% it's not that it would be a joke. 00:09:14.990 --> 00:09:18.830 align:middle line:84% And if you're building a new semiconductor fab, 00:09:18.830 --> 00:09:20.950 align:middle line:84% that's a reasonable plan because you're 00:09:20.950 --> 00:09:23.147 align:middle line:90% building a semiconductor fab. 00:09:23.147 --> 00:09:25.230 align:middle line:84% But let me tell you, if you're building a software 00:09:25.230 --> 00:09:30.790 align:middle line:84% product or an application or a new service, three years, 00:09:30.790 --> 00:09:32.150 align:middle line:90% it's ridiculous. 00:09:32.150 --> 00:09:35.430 align:middle line:84% The market will have changed several times. 00:09:35.430 --> 00:09:39.170 align:middle line:84% So you have to think what you plan to do. 00:09:39.170 --> 00:09:44.390 align:middle line:84% I advise people, tell me what you can get done in six months. 00:09:44.390 --> 00:09:45.950 align:middle line:90% Well, I need-- no, no. 00:09:45.950 --> 00:09:48.790 align:middle line:84% I know you need more time, but what can you 00:09:48.790 --> 00:09:51.910 align:middle line:90% get done in six months? 00:09:51.910 --> 00:09:53.850 align:middle line:90% And then do that quickly. 00:09:53.850 --> 00:09:58.010 align:middle line:84% And then six months after that, and six months after that, 00:09:58.010 --> 00:09:59.690 align:middle line:90% you keep iterating. 00:09:59.690 --> 00:10:03.230 align:middle line:84% So it really changes the whole way 00:10:03.230 --> 00:10:05.230 align:middle line:84% you look at what you're building. 00:10:05.230 --> 00:10:07.650 align:middle line:90% It doesn't work in every case. 00:10:07.650 --> 00:10:11.590 align:middle line:84% I don't mean this is universal, but I'd 00:10:11.590 --> 00:10:14.770 align:middle line:84% be very worried if somebody says, 00:10:14.770 --> 00:10:17.010 align:middle line:84% Irving, it's three years or nothing. 00:10:17.010 --> 00:10:18.770 align:middle line:90% I'd be very, very worried. 00:10:18.770 --> 00:10:19.510 align:middle line:90% Please. 00:10:19.510 --> 00:10:21.093 align:middle line:84% AUDIENCE: What you're just describing, 00:10:21.093 --> 00:10:24.110 align:middle line:84% it seems pretty obvious in the IT and software industries. 00:10:24.110 --> 00:10:25.133 align:middle line:90% How universal is it? 00:10:25.133 --> 00:10:27.050 align:middle line:84% IRVING WLADAWSKY-BERGER: Well, let me ask you, 00:10:27.050 --> 00:10:32.430 align:middle line:84% who have experience in other industries, drug research 00:10:32.430 --> 00:10:33.610 align:middle line:90% in pharmaceutical? 00:10:33.610 --> 00:10:39.310 align:middle line:84% There's a lot of work about testing new pharmaceuticals 00:10:39.310 --> 00:10:45.280 align:middle line:84% much, much faster with all kinds of new technologies. 00:10:45.280 --> 00:10:47.760 align:middle line:84% Do you all want to talk about your experience 00:10:47.760 --> 00:10:49.840 align:middle line:90% in your industries? 00:10:49.840 --> 00:10:51.960 align:middle line:84% AUDIENCE: In the drug industry, hand skills 00:10:51.960 --> 00:10:54.142 align:middle line:84% are to get out the drug are very-- 00:10:54.142 --> 00:10:55.600 align:middle line:84% the product plan would be something 00:10:55.600 --> 00:10:59.960 align:middle line:84% at least 10 years ahead, but the emphasis today 00:10:59.960 --> 00:11:03.440 align:middle line:90% is on failing really fast. 00:11:03.440 --> 00:11:07.080 align:middle line:84% And I don't know if this is connected 00:11:07.080 --> 00:11:09.400 align:middle line:90% with how fast we learn. 00:11:09.400 --> 00:11:11.840 align:middle line:84% But it seems like if you iterate and you iteration 00:11:11.840 --> 00:11:14.800 align:middle line:84% fail and takes three years, you tend 00:11:14.800 --> 00:11:17.858 align:middle line:84% to forget as an organization, what drove your iteration. 00:11:17.858 --> 00:11:19.900 align:middle line:84% Whereas if it's about six months or three months, 00:11:19.900 --> 00:11:20.840 align:middle line:90% you tend to keep that-- 00:11:20.840 --> 00:11:22.632 align:middle line:84% IRVING WLADAWSKY-BERGER: That's my feeling. 00:11:22.632 --> 00:11:27.440 align:middle line:84% And are pharmaceuticals making progress in doing that? 00:11:27.440 --> 00:11:28.920 align:middle line:90% AUDIENCE: They're trying. 00:11:28.920 --> 00:11:31.040 align:middle line:84% However, the problem is every iteration 00:11:31.040 --> 00:11:33.040 align:middle line:90% costs a huge amount of money. 00:11:33.040 --> 00:11:34.657 align:middle line:90% So the question is between-- 00:11:34.657 --> 00:11:36.240 align:middle line:84% IRVING WLADAWSKY-BERGER: I understand. 00:11:36.240 --> 00:11:37.480 align:middle line:84% AUDIENCE: --balancing money and your knowledge. 00:11:37.480 --> 00:11:39.120 align:middle line:84% IRVING WLADAWSKY-BERGER: I understand. 00:11:39.120 --> 00:11:42.120 align:middle line:90% How about other industries? 00:11:42.120 --> 00:11:46.440 align:middle line:90% I mean, in retailing. 00:11:46.440 --> 00:11:50.620 align:middle line:84% I heard this talk in June, the CEO of Metro. 00:11:50.620 --> 00:11:54.120 align:middle line:90% It's a big, big store. 00:11:54.120 --> 00:11:57.660 align:middle line:84% And we'll hear from Chris McCormick in a couple of weeks. 00:11:57.660 --> 00:11:58.900 align:middle line:90% We can ask him that. 00:11:58.900 --> 00:12:02.960 align:middle line:84% Seventy percent of new products on the shelf fail. 00:12:02.960 --> 00:12:05.180 align:middle line:84% By fail, I mean, customers don't buy them. 00:12:05.180 --> 00:12:08.080 align:middle line:84% It doesn't mean-- the product may be wonderful products. 00:12:08.080 --> 00:12:13.080 align:middle line:84% It's just if you define success as customers 00:12:13.080 --> 00:12:15.400 align:middle line:84% buy it as opposed to they are pretty 00:12:15.400 --> 00:12:21.520 align:middle line:84% and customers should buy it, then 70% fail. 00:12:21.520 --> 00:12:28.780 align:middle line:84% And this may get back to one of the questions that he mentions. 00:12:28.780 --> 00:12:29.280 align:middle line:90% Oops. 00:12:29.280 --> 00:12:36.120 align:middle line:90% 00:12:36.120 --> 00:12:40.880 align:middle line:84% If something is going to fail, if it's inevitably 00:12:40.880 --> 00:12:46.810 align:middle line:84% going to fail, the earlier you find out, the less-- 00:12:46.810 --> 00:12:49.630 align:middle line:84% in fact, if something fails quickly, 00:12:49.630 --> 00:12:52.250 align:middle line:90% it's a learning experience. 00:12:52.250 --> 00:12:56.690 align:middle line:84% If something fails after you've invested tens of millions 00:12:56.690 --> 00:13:03.850 align:middle line:84% of dollars and years, it's a goddamn problem. 00:13:03.850 --> 00:13:07.230 align:middle line:84% So in some ways, what we're talking about is, 00:13:07.230 --> 00:13:11.690 align:middle line:84% is there a way to do this so that whatever doesn't 00:13:11.690 --> 00:13:15.130 align:middle line:90% work you find out quickly? 00:13:15.130 --> 00:13:17.250 align:middle line:84% Because maybe what you're then doing 00:13:17.250 --> 00:13:20.750 align:middle line:84% is maybe weeding out a lot of the efforts, 00:13:20.750 --> 00:13:24.850 align:middle line:84% and you can concentrate on those parts that are most promising. 00:13:24.850 --> 00:13:27.350 align:middle line:90% Maybe that is what we're doing. 00:13:27.350 --> 00:13:32.690 align:middle line:84% But I know there are a whole new drug sequencing technologies 00:13:32.690 --> 00:13:36.430 align:middle line:84% that even in wet labs allow testing of lots of stuff. 00:13:36.430 --> 00:13:40.930 align:middle line:84% Of course, genomics sequencing is coming way down. 00:13:40.930 --> 00:13:47.210 align:middle line:84% Simulations are enabling to view designs 00:13:47.210 --> 00:13:51.650 align:middle line:84% at a much faster way than ever before supercomputing simulation 00:13:51.650 --> 00:13:54.010 align:middle line:90% and so on. 00:13:54.010 --> 00:13:56.610 align:middle line:84% But my feeling, and I really want 00:13:56.610 --> 00:13:59.570 align:middle line:84% to hear about your industries, is 00:13:59.570 --> 00:14:02.330 align:middle line:84% this may be one of the biggest innovation 00:14:02.330 --> 00:14:04.110 align:middle line:90% areas for every business. 00:14:04.110 --> 00:14:10.590 align:middle line:84% How do you reframe the way you do things to take this approach, 00:14:10.590 --> 00:14:16.250 align:middle line:84% or start with something and then start refining and incrementing 00:14:16.250 --> 00:14:17.990 align:middle line:90% with the biggest opportunities? 00:14:17.990 --> 00:14:23.010 align:middle line:84% And it's not clear what to do, but do that. 00:14:23.010 --> 00:14:26.570 align:middle line:84% Now, this gets back to some of the things 00:14:26.570 --> 00:14:28.730 align:middle line:90% we talked about last week. 00:14:28.730 --> 00:14:30.530 align:middle line:84% Somebody said, oh, Irving, you don't I 00:14:30.530 --> 00:14:32.890 align:middle line:84% have to build this whole infrastructure. 00:14:32.890 --> 00:14:38.370 align:middle line:84% I said, well, what's wrong with the internet and open source? 00:14:38.370 --> 00:14:42.010 align:middle line:90% Well, I don't like Linus. 00:14:42.010 --> 00:14:44.610 align:middle line:90% Linus is a fine guy. 00:14:44.610 --> 00:14:51.890 align:middle line:84% And so the more you leverage what exists-- 00:14:51.890 --> 00:14:54.870 align:middle line:84% an open source has the nice characteristic 00:14:54.870 --> 00:15:00.730 align:middle line:84% if you can use it as a platform, that the price is right 00:15:00.730 --> 00:15:02.110 align:middle line:90% when you are starting out. 00:15:02.110 --> 00:15:05.030 align:middle line:84% It may not be great for scaling later. 00:15:05.030 --> 00:15:08.710 align:middle line:84% You may need to do things for scaling. 00:15:08.710 --> 00:15:11.610 align:middle line:90% But to get going-- 00:15:11.610 --> 00:15:16.430 align:middle line:84% And if you use as much existing components as possible, 00:15:16.430 --> 00:15:20.130 align:middle line:84% obviously you don't want to steal anybody's IP, 00:15:20.130 --> 00:15:22.390 align:middle line:84% you just start way ahead of the game. 00:15:22.390 --> 00:15:24.310 align:middle line:84% You don't have to start from the bottom. 00:15:24.310 --> 00:15:26.350 align:middle line:90% You start from quite a bit. 00:15:26.350 --> 00:15:32.690 align:middle line:84% If you partner with others that have a base you can build on, 00:15:32.690 --> 00:15:38.420 align:middle line:84% a platform that you can leverage and then build on top of that, 00:15:38.420 --> 00:15:42.420 align:middle line:84% you can just go to market that much faster. 00:15:42.420 --> 00:15:46.120 align:middle line:84% But it's an interesting way of thinking. 00:15:46.120 --> 00:15:51.580 align:middle line:84% I know at IBM, I think I have mentioned that. 00:15:51.580 --> 00:15:55.600 align:middle line:84% I would say 10 years ago, people would think, 00:15:55.600 --> 00:15:58.500 align:middle line:84% well, you don't think I'm good enough 00:15:58.500 --> 00:16:01.580 align:middle line:84% to build the whole thing, that I have to go build it 00:16:01.580 --> 00:16:09.060 align:middle line:84% on Java, which comes from Sun, or I have to go acquire Lotus. 00:16:09.060 --> 00:16:11.700 align:middle line:84% Well, you're good, you're wonderful. 00:16:11.700 --> 00:16:16.540 align:middle line:84% But it will take you three goddamn years, maybe five. 00:16:16.540 --> 00:16:21.400 align:middle line:84% And we have zero promise that after these 3 to 5 years, 00:16:21.400 --> 00:16:23.140 align:middle line:90% you'll be where they are today. 00:16:23.140 --> 00:16:24.620 align:middle line:90% Zero. 00:16:24.620 --> 00:16:29.100 align:middle line:84% So from a business point of view, that sounds dumb. 00:16:29.100 --> 00:16:34.500 align:middle line:84% So the more you'd say, OK, what's a base that lets me-- 00:16:34.500 --> 00:16:36.580 align:middle line:90% it's a base that's not-- 00:16:36.580 --> 00:16:41.500 align:middle line:84% that is an enabler for your innovation 00:16:41.500 --> 00:16:43.240 align:middle line:90% that lets you get going? 00:16:43.240 --> 00:16:45.780 align:middle line:84% The faster you find an existing base, 00:16:45.780 --> 00:16:48.500 align:middle line:90% the faster you can get going. 00:16:48.500 --> 00:16:53.280 align:middle line:84% And I think that attitude permeates businesses these days. 00:16:53.280 --> 00:16:55.980 align:middle line:84% That's why initial investments are 00:16:55.980 --> 00:16:57.680 align:middle line:84% so much lower than they used to be, 00:16:57.680 --> 00:17:01.140 align:middle line:84% because they expect everybody to use a base. 00:17:01.140 --> 00:17:05.319 align:middle line:84% Any comments on this point from all your experiences? 00:17:05.319 --> 00:17:13.780 align:middle line:90% 00:17:13.780 --> 00:17:19.740 align:middle line:84% Now, the third point that Tom Key makes is to-- 00:17:19.740 --> 00:17:23.500 align:middle line:90% 00:17:23.500 --> 00:17:25.300 align:middle line:84% it's the same point, essentially, 00:17:25.300 --> 00:17:28.020 align:middle line:90% we just made before. 00:17:28.020 --> 00:17:33.660 align:middle line:84% By low fidelity, he means the early designs don't have 00:17:33.660 --> 00:17:35.800 align:middle line:90% to be the ultimate designs. 00:17:35.800 --> 00:17:41.260 align:middle line:84% They can be course designs that will show later that you 00:17:41.260 --> 00:17:43.160 align:middle line:90% will keep incrementing later. 00:17:43.160 --> 00:17:45.340 align:middle line:84% So it really keeps-- it's another way 00:17:45.340 --> 00:17:50.540 align:middle line:84% of saying pick approaches to design 00:17:50.540 --> 00:17:55.020 align:middle line:84% that enables you to incrementally improve them 00:17:55.020 --> 00:17:59.000 align:middle line:84% as you get information and feedback from the marketplace. 00:17:59.000 --> 00:18:00.540 align:middle line:90% That's what he's saying. 00:18:00.540 --> 00:18:03.900 align:middle line:84% And that's another way of saying that. 00:18:03.900 --> 00:18:13.900 align:middle line:84% And then finally, and again, we just discussed that. 00:18:13.900 --> 00:18:20.980 align:middle line:84% Be very clear about what your unique innovation is. 00:18:20.980 --> 00:18:25.180 align:middle line:84% What is really new in what you're trying to do? 00:18:25.180 --> 00:18:28.140 align:middle line:84% That's what you should focus your energy. 00:18:28.140 --> 00:18:33.110 align:middle line:84% That's what you should take the risks. 00:18:33.110 --> 00:18:39.510 align:middle line:84% Now, if probabilities of failures 00:18:39.510 --> 00:18:50.910 align:middle line:84% multiply, and p1 times p2 times p3 quickly get very small. 00:18:50.910 --> 00:18:56.470 align:middle line:84% So part combine new and traditional 00:18:56.470 --> 00:19:05.670 align:middle line:84% is if you build something new on top of existing components that 00:19:05.670 --> 00:19:11.070 align:middle line:84% are well understood, where the risk is minimal, 00:19:11.070 --> 00:19:15.790 align:middle line:84% you're just increasing your overall probability of success. 00:19:15.790 --> 00:19:19.950 align:middle line:84% And later on, once the whole thing is better, 00:19:19.950 --> 00:19:23.150 align:middle line:84% one of the first things you might do for release 2 00:19:23.150 --> 00:19:28.230 align:middle line:84% is improve this component and get a better one. 00:19:28.230 --> 00:19:30.870 align:middle line:84% But now you have a stable product 00:19:30.870 --> 00:19:33.110 align:middle line:90% and you know where it fits. 00:19:33.110 --> 00:19:37.310 align:middle line:84% One way that I have thought about it often is that when 00:19:37.310 --> 00:19:39.170 align:middle line:90% you have a project-- 00:19:39.170 --> 00:19:41.830 align:middle line:90% 00:19:41.830 --> 00:19:44.390 align:middle line:84% think of a project and think of-- 00:19:44.390 --> 00:19:50.630 align:middle line:84% I have this mental image, lots of moving parts in the project. 00:19:50.630 --> 00:19:53.670 align:middle line:84% Well, if you have too many moving parts, 00:19:53.670 --> 00:19:58.030 align:middle line:84% the damn thing can never come together. 00:19:58.030 --> 00:20:01.390 align:middle line:84% And if the thing doesn't come together, 00:20:01.390 --> 00:20:04.870 align:middle line:90% unfortunately, it falls apart. 00:20:04.870 --> 00:20:11.190 align:middle line:84% So you want a kind of core around which the moving part can 00:20:11.190 --> 00:20:16.630 align:middle line:84% settle down so that you are incrementally growing, 00:20:16.630 --> 00:20:20.230 align:middle line:84% and then come back and start replacing the core. 00:20:20.230 --> 00:20:23.250 align:middle line:84% And that's a way of reducing risk. 00:20:23.250 --> 00:20:25.150 align:middle line:90% Please. 00:20:25.150 --> 00:20:29.670 align:middle line:84% AUDIENCE: It seems like you're anticipating early information 00:20:29.670 --> 00:20:31.670 align:middle line:90% and fail early. 00:20:31.670 --> 00:20:36.470 align:middle line:84% You're saying, put the 80% solution out there. 00:20:36.470 --> 00:20:39.030 align:middle line:84% And you have to be careful with that, I think, 00:20:39.030 --> 00:20:44.030 align:middle line:84% because if you put out there an 80% solution but everybody 00:20:44.030 --> 00:20:46.370 align:middle line:84% doesn't know it's the 80% solution, 00:20:46.370 --> 00:20:49.190 align:middle line:84% or if it's really the 70% solution 00:20:49.190 --> 00:20:51.250 align:middle line:84% and everybody gets pissed off at you, 00:20:51.250 --> 00:20:54.540 align:middle line:84% and then even if you improve it, you have lost the credibility. 00:20:54.540 --> 00:20:56.290 align:middle line:84% IRVING WLADAWSKY-BERGER: I agree, totally. 00:20:56.290 --> 00:20:57.850 align:middle line:90% You have to be very careful. 00:20:57.850 --> 00:20:59.430 align:middle line:90% AUDIENCE: There's a line there-- 00:20:59.430 --> 00:21:00.950 align:middle line:84% IRVING WLADAWSKY-BERGER: Totally. 00:21:00.950 --> 00:21:01.450 align:middle line:90% No, no. 00:21:01.450 --> 00:21:04.930 align:middle line:84% But remember, if we knew where the line is exactly, 00:21:04.930 --> 00:21:08.510 align:middle line:90% business wouldn't be business. 00:21:08.510 --> 00:21:11.110 align:middle line:90% No, no. 00:21:11.110 --> 00:21:18.150 align:middle line:84% I think a lot of the innovation is picking the line. 00:21:18.150 --> 00:21:21.150 align:middle line:90% And I don't know-- 00:21:21.150 --> 00:21:24.550 align:middle line:84% I have a lot of emphasis on design in everything 00:21:24.550 --> 00:21:27.650 align:middle line:84% that I keep talking about, and the marketplace. 00:21:27.650 --> 00:21:29.620 align:middle line:84% And a lot of the reason-- and by the way, 00:21:29.620 --> 00:21:33.320 align:middle line:84% I mean design and the marketplace for the engineers, 00:21:33.320 --> 00:21:38.280 align:middle line:84% not for the MBAs that are doing whatever MBAs do. 00:21:38.280 --> 00:21:42.200 align:middle line:90% MBAs need to do-- 00:21:42.200 --> 00:21:43.460 align:middle line:90% we need to count money. 00:21:43.460 --> 00:21:46.000 align:middle line:90% We need to do a lot of stuff. 00:21:46.000 --> 00:21:53.200 align:middle line:84% But the kind of guidance about how 00:21:53.200 --> 00:21:58.520 align:middle line:84% to do a design that is suitable for the market you're after, 00:21:58.520 --> 00:22:02.680 align:middle line:90% where the 80% you do is correct. 00:22:02.680 --> 00:22:08.040 align:middle line:84% That has to be intertwined with your overall design. 00:22:08.040 --> 00:22:12.100 align:middle line:84% And if you are the design manager, chief architect, 00:22:12.100 --> 00:22:14.320 align:middle line:84% and if you have very good intuition about it, 00:22:14.320 --> 00:22:17.400 align:middle line:84% you can do a much better job than if you 00:22:17.400 --> 00:22:20.520 align:middle line:84% have to listen to what somebody else who's been out 00:22:20.520 --> 00:22:22.920 align:middle line:90% there tells you. 00:22:22.920 --> 00:22:27.840 align:middle line:84% But it's a huge part of the innovation in my mind. 00:22:27.840 --> 00:22:29.600 align:middle line:84% AUDIENCE: But also failing does not 00:22:29.600 --> 00:22:31.400 align:middle line:84% mean having to put it out to the public. 00:22:31.400 --> 00:22:34.207 align:middle line:84% Like a drug, I wouldn't test it on a human being. 00:22:34.207 --> 00:22:36.040 align:middle line:84% IRVING WLADAWSKY-BERGER: Yeah, I understand. 00:22:36.040 --> 00:22:36.960 align:middle line:90% AUDIENCE: Or cars. 00:22:36.960 --> 00:22:38.210 align:middle line:90% IRVING WLADAWSKY-BERGER: Yeah. 00:22:38.210 --> 00:22:39.880 align:middle line:90% Also, it may be simulation. 00:22:39.880 --> 00:22:43.760 align:middle line:84% Remember, in a lot of areas, the way 00:22:43.760 --> 00:22:48.040 align:middle line:84% you do early experimentation is massive simulation. 00:22:48.040 --> 00:22:49.498 align:middle line:84% AUDIENCE: You said that the drugs-- 00:22:49.498 --> 00:22:51.082 align:middle line:84% IRVING WLADAWSKY-BERGER: No, in drugs. 00:22:51.082 --> 00:22:53.260 align:middle line:84% But in cars, you might be able to do that. 00:22:53.260 --> 00:22:55.400 align:middle line:84% You might be able to design the car, 00:22:55.400 --> 00:22:59.520 align:middle line:84% start testing its aerodynamic qualities, its manufacture. 00:22:59.520 --> 00:23:05.560 align:middle line:84% And the amount we can simulate keeps going out and out and out. 00:23:05.560 --> 00:23:15.340 align:middle line:84% So out there is not necessarily customers in the usual sense. 00:23:15.340 --> 00:23:18.640 align:middle line:90% 00:23:18.640 --> 00:23:24.320 align:middle line:84% But we'll talk in the next section, often it is. 00:23:24.320 --> 00:23:29.120 align:middle line:84% And of course, there the key is to select the right customers. 00:23:29.120 --> 00:23:29.960 align:middle line:90% Please. 00:23:29.960 --> 00:23:33.860 align:middle line:84% AUDIENCE: Failure, by nature, it is defined by market. 00:23:33.860 --> 00:23:35.680 align:middle line:84% I mean, that's the way you define. 00:23:35.680 --> 00:23:39.667 align:middle line:84% So how do you not go to the market and discover failure? 00:23:39.667 --> 00:23:41.000 align:middle line:90% IRVING WLADAWSKY-BERGER: No, no. 00:23:41.000 --> 00:23:44.960 align:middle line:84% To discover market failure, you have to go to the market. 00:23:44.960 --> 00:23:53.360 align:middle line:84% To discover that this car just has bad crash properties, 00:23:53.360 --> 00:23:56.080 align:middle line:90% that you can do quickly. 00:23:56.080 --> 00:24:01.520 align:middle line:84% Or even, let's say, to show people-- 00:24:01.520 --> 00:24:04.400 align:middle line:84% and that's why Detroit does concept cars. 00:24:04.400 --> 00:24:05.940 align:middle line:90% And actually, not just Detroit. 00:24:05.940 --> 00:24:08.400 align:middle line:90% Everybody does concept cars. 00:24:08.400 --> 00:24:10.800 align:middle line:84% And I think a lot of the reason for the concept 00:24:10.800 --> 00:24:17.960 align:middle line:84% cars is find out if people go bananas when they see the car, 00:24:17.960 --> 00:24:21.760 align:middle line:84% or if they say, this is the ugliest car I've ever seen 00:24:21.760 --> 00:24:26.450 align:middle line:84% and they don't want anything to do with you. 00:24:26.450 --> 00:24:33.250 align:middle line:84% And again, my view is through extensive use 00:24:33.250 --> 00:24:38.690 align:middle line:84% of simulations modeling, you can probably get a concept car 00:24:38.690 --> 00:24:42.610 align:middle line:84% to a show much faster than in the old days where 00:24:42.610 --> 00:24:45.350 align:middle line:90% you had to rely on modelers. 00:24:45.350 --> 00:24:47.050 align:middle line:90% Isn't that correct? 00:24:47.050 --> 00:24:49.970 align:middle line:84% There is a tremendous amount more you can do today. 00:24:49.970 --> 00:24:53.330 align:middle line:84% So there's a whole set of techniques, but all of them 00:24:53.330 --> 00:24:57.353 align:middle line:84% are getting out there as quickly as possible. 00:24:57.353 --> 00:24:58.770 align:middle line:84% AUDIENCE: What kind of experiments 00:24:58.770 --> 00:25:02.370 align:middle line:84% can you do in the market to produce market changes? 00:25:02.370 --> 00:25:05.218 align:middle line:84% That's really frontier, actually. 00:25:05.218 --> 00:25:06.510 align:middle line:90% IRVING WLADAWSKY-BERGER: It is. 00:25:06.510 --> 00:25:08.730 align:middle line:84% And let me wait 'till the next section, 00:25:08.730 --> 00:25:12.730 align:middle line:84% because we'll talk about that very much 00:25:12.730 --> 00:25:14.290 align:middle line:90% in the whole next section. 00:25:14.290 --> 00:25:17.790 align:middle line:84% I agree totally, that is the frontier. 00:25:17.790 --> 00:25:20.870 align:middle line:90% And we'll talk a lot about it. 00:25:20.870 --> 00:25:21.930 align:middle line:90% Yes, please. 00:25:21.930 --> 00:25:23.930 align:middle line:84% AUDIENCE: I mean, in software, which 00:25:23.930 --> 00:25:29.290 align:middle line:84% is where I come from, lots of prototype mockups, 00:25:29.290 --> 00:25:32.170 align:middle line:84% you run it by users and you just keep 00:25:32.170 --> 00:25:34.810 align:middle line:84% doing that very, very quickly in the process 00:25:34.810 --> 00:25:38.290 align:middle line:90% and then throw it out for you. 00:25:38.290 --> 00:25:42.250 align:middle line:84% At the end, you have made all the mistakes very quickly 00:25:42.250 --> 00:25:46.730 align:middle line:84% and validated with at least some of the public from the consumer 00:25:46.730 --> 00:25:50.570 align:middle line:90% that it is not working. 00:25:50.570 --> 00:25:53.770 align:middle line:84% Even if that's a feasible model, you just throw it out. 00:25:53.770 --> 00:25:55.312 align:middle line:84% IRVING WLADAWSKY-BERGER: Well, that's 00:25:55.312 --> 00:25:58.370 align:middle line:84% why alphas and betas are so important. 00:25:58.370 --> 00:26:04.610 align:middle line:84% That's why a-- I think John Patrick mentioned 00:26:04.610 --> 00:26:08.290 align:middle line:84% we developed this notion of alpha works 00:26:08.290 --> 00:26:12.050 align:middle line:84% because we had all these ideas in our research labs 00:26:12.050 --> 00:26:18.390 align:middle line:84% that had no channel to be experimented within the market. 00:26:18.390 --> 00:26:20.250 align:middle line:90% So we said, well, screw it. 00:26:20.250 --> 00:26:21.950 align:middle line:90% Make it downloadable. 00:26:21.950 --> 00:26:24.470 align:middle line:84% And the first time we told that to the lawyers, 00:26:24.470 --> 00:26:27.250 align:middle line:90% they had apoplexy. 00:26:27.250 --> 00:26:31.430 align:middle line:84% And we said, no, no constructors and IP something. 00:26:31.430 --> 00:26:33.530 align:middle line:84% Will you sign here and say you will not 00:26:33.530 --> 00:26:38.450 align:middle line:84% use this for a commercial purpose, only for experimenting? 00:26:38.450 --> 00:26:40.290 align:middle line:90% And we did that. 00:26:40.290 --> 00:26:43.510 align:middle line:84% So that's a vehicle to get it out there. 00:26:43.510 --> 00:26:48.690 align:middle line:84% And software, there is a lot of experimentation of that sold. 00:26:48.690 --> 00:26:52.010 align:middle line:84% I mean, any of you that deal with Yahoo, Google and so on, 00:26:52.010 --> 00:26:54.770 align:middle line:90% you keep getting gadget-- 00:26:54.770 --> 00:26:56.610 align:middle line:90% I mean, there are names for-- 00:26:56.610 --> 00:26:57.470 align:middle line:90% what are the names? 00:26:57.470 --> 00:26:59.650 align:middle line:90% Gadgets and widgets. 00:26:59.650 --> 00:27:00.588 align:middle line:90% Am I correct? 00:27:00.588 --> 00:27:01.630 align:middle line:90% AUDIENCE: Google gadgets. 00:27:01.630 --> 00:27:03.672 align:middle line:84% IRVING WLADAWSKY-BERGER: Google gadgets and Yahoo 00:27:03.672 --> 00:27:04.930 align:middle line:90% has widgets, right? 00:27:04.930 --> 00:27:09.550 align:middle line:84% And I don't know what Microsoft calls whatever they have. 00:27:09.550 --> 00:27:13.190 align:middle line:84% Do we have an IBM widgets and gadgets? 00:27:13.190 --> 00:27:14.730 align:middle line:90% AUDIENCE: [INAUDIBLE] sometimes. 00:27:14.730 --> 00:27:17.090 align:middle line:84% IRVING WLADAWSKY-BERGER: So there are-- 00:27:17.090 --> 00:27:18.390 align:middle line:90% in software, there are. 00:27:18.390 --> 00:27:21.940 align:middle line:84% Now somebody says, oh, can I run the Federal Reserve money 00:27:21.940 --> 00:27:23.420 align:middle line:90% transfer? 00:27:23.420 --> 00:27:24.800 align:middle line:90% We say, please don't. 00:27:24.800 --> 00:27:29.160 align:middle line:84% Or can I run a surgical operating room with that? 00:27:29.160 --> 00:27:31.420 align:middle line:90% No, we don't recommend that. 00:27:31.420 --> 00:27:34.840 align:middle line:84% So you have to be very careful about who you pick. 00:27:34.840 --> 00:27:37.920 align:middle line:84% Otherwise, your whole brand goes up in smoke 00:27:37.920 --> 00:27:39.900 align:middle line:90% if you pick the wrong sections. 00:27:39.900 --> 00:27:42.560 align:middle line:84% But you pick the right sections, and in the next section, 00:27:42.560 --> 00:27:46.180 align:middle line:84% we'll talk a lot about it, it's good. 00:27:46.180 --> 00:27:48.900 align:middle line:84% AUDIENCE: It's interesting you brought up Google. 00:27:48.900 --> 00:27:50.860 align:middle line:90% They use beta. 00:27:50.860 --> 00:27:52.520 align:middle line:90% They put beta on everything. 00:27:52.520 --> 00:27:55.110 align:middle line:84% In fact, I think Google Maps is still beta. 00:27:55.110 --> 00:27:56.860 align:middle line:84% IRVING WLADAWSKY-BERGER: Is it still beta? 00:27:56.860 --> 00:27:59.100 align:middle line:84% AUDIENCE: Google Mail is still is. 00:27:59.100 --> 00:28:00.815 align:middle line:90% Everybody's using Gmail. 00:28:00.815 --> 00:28:02.440 align:middle line:84% IRVING WLADAWSKY-BERGER: Gmail is beta? 00:28:02.440 --> 00:28:03.800 align:middle line:84% AUDIENCE: Yeah, everything's beta. 00:28:03.800 --> 00:28:05.383 align:middle line:84% AUDIENCE: Yeah, they call it all beta. 00:28:05.383 --> 00:28:06.760 align:middle line:90% It's a marketing spiel. 00:28:06.760 --> 00:28:10.380 align:middle line:84% So if something goes wrong, like you don't get too angry at beta, 00:28:10.380 --> 00:28:10.880 align:middle line:90% right. 00:28:10.880 --> 00:28:13.560 align:middle line:84% So they get around to on all of that. 00:28:13.560 --> 00:28:15.460 align:middle line:84% IRVING WLADAWSKY-BERGER: But remember-- 00:28:15.460 --> 00:28:16.900 align:middle line:90% let me ask you a question. 00:28:16.900 --> 00:28:20.220 align:middle line:90% How much do you pay for Gmail? 00:28:20.220 --> 00:28:21.098 align:middle line:90% No, no. 00:28:21.098 --> 00:28:22.640 align:middle line:84% AUDIENCE: But advertisers are paying. 00:28:22.640 --> 00:28:23.700 align:middle line:84% IRVING WLADAWSKY-BERGER: No, no, I know that. 00:28:23.700 --> 00:28:24.120 align:middle line:90% I know that. 00:28:24.120 --> 00:28:24.620 align:middle line:90% No, no. 00:28:24.620 --> 00:28:25.980 align:middle line:90% But consumers-- 00:28:25.980 --> 00:28:28.040 align:middle line:84% AUDIENCE: Yahoo has the same business model. 00:28:28.040 --> 00:28:30.020 align:middle line:84% Yet when Yahoo Mail goes down, they 00:28:30.020 --> 00:28:31.920 align:middle line:90% get all kinds of irate users. 00:28:31.920 --> 00:28:35.580 align:middle line:84% I mean, it's a marketing strategy. 00:28:35.580 --> 00:28:41.140 align:middle line:84% AUDIENCE: Yahoo Mail was made up for two years before anything. 00:28:41.140 --> 00:28:42.140 align:middle line:90% They did the same thing. 00:28:42.140 --> 00:28:45.140 align:middle line:84% AUDIENCE: But Gmail has been around for four years or so. 00:28:45.140 --> 00:28:49.060 align:middle line:84% AUDIENCE: The key to Gmail is that their recovery process is-- 00:28:49.060 --> 00:28:51.700 align:middle line:84% Google can-- they put code out every week. 00:28:51.700 --> 00:28:54.940 align:middle line:84% But if there's an issue, that rollback is within an hour. 00:28:54.940 --> 00:28:56.500 align:middle line:90% That's the key to what they are. 00:28:56.500 --> 00:28:58.160 align:middle line:84% So they're able to do production. 00:28:58.160 --> 00:28:59.577 align:middle line:84% IRVING WLADAWSKY-BERGER: So that's 00:28:59.577 --> 00:29:03.060 align:middle line:84% equivalent of very quick failure and recovery, 00:29:03.060 --> 00:29:08.060 align:middle line:84% because you are expecting that anything goes wrong, 00:29:08.060 --> 00:29:10.040 align:middle line:90% you'll immediately roll it out. 00:29:10.040 --> 00:29:12.500 align:middle line:84% AUDIENCE: Or is that rapid experimentation? 00:29:12.500 --> 00:29:14.208 align:middle line:84% IRVING WLADAWSKY-BERGER: Well, it's both. 00:29:14.208 --> 00:29:20.000 align:middle line:84% I mean, again, remember this gets back to why I want-- 00:29:20.000 --> 00:29:23.660 align:middle line:84% I think it's so critical that the engineers be very involved 00:29:23.660 --> 00:29:26.700 align:middle line:84% in marketing, because we're talking marketing. 00:29:26.700 --> 00:29:29.140 align:middle line:84% But this is the kind of marketing 00:29:29.140 --> 00:29:31.885 align:middle line:84% that can have a huge impact on your architecture. 00:29:31.885 --> 00:29:33.260 align:middle line:84% And if you don't have the right-- 00:29:33.260 --> 00:29:37.820 align:middle line:84% you cannot do the rapid recovery, somebody says, well, 00:29:37.820 --> 00:29:41.300 align:middle line:84% I know you build this architecture where everything is 00:29:41.300 --> 00:29:42.820 align:middle line:90% monolithic. 00:29:42.820 --> 00:29:44.920 align:middle line:90% There are no components. 00:29:44.920 --> 00:29:46.480 align:middle line:90% I want you to do rapid recovery. 00:29:46.480 --> 00:29:48.300 align:middle line:84% I don't think you can do that, right? 00:29:48.300 --> 00:29:52.260 align:middle line:84% It has to be built into the basic architecture, which 00:29:52.260 --> 00:29:55.940 align:middle line:84% is why these feedbacks are so important. 00:29:55.940 --> 00:29:56.920 align:middle line:90% You had a point? 00:29:56.920 --> 00:29:57.545 align:middle line:90% AUDIENCE: Yeah. 00:29:57.545 --> 00:30:00.220 align:middle line:84% I just wanted to make [INAUDIBLE] 00:30:00.220 --> 00:30:03.660 align:middle line:84% was in perpetual beta so you never come out of it. 00:30:03.660 --> 00:30:07.220 align:middle line:84% You take it out to the marketplace like this 00:30:07.220 --> 00:30:09.980 align:middle line:84% and then figure out if-- for example, Gmail 00:30:09.980 --> 00:30:12.330 align:middle line:84% they put out a new thing almost every day. 00:30:12.330 --> 00:30:14.080 align:middle line:84% IRVING WLADAWSKY-BERGER: No, that's right. 00:30:14.080 --> 00:30:15.540 align:middle line:90% AUDIENCE: Are they OK with it? 00:30:15.540 --> 00:30:15.590 align:middle line:90% And keep on-- 00:30:15.590 --> 00:30:17.110 align:middle line:84% IRVING WLADAWSKY-BERGER: Right, but remember, it 00:30:17.110 --> 00:30:18.530 align:middle line:90% depends on the product. 00:30:18.530 --> 00:30:23.430 align:middle line:84% I mean, it's very important to know that a hospital management 00:30:23.430 --> 00:30:26.850 align:middle line:84% system cannot be in perpetual beta. 00:30:26.850 --> 00:30:32.970 align:middle line:84% So you need to be careful about what products you need to use. 00:30:32.970 --> 00:30:37.630 align:middle line:84% Let me call it more classic development cycles. 00:30:37.630 --> 00:30:42.110 align:middle line:84% And even the classics, you can innovate, 00:30:42.110 --> 00:30:45.410 align:middle line:84% and what products are in the perpetual beta category. 00:30:45.410 --> 00:30:49.510 align:middle line:84% And in fact, I think that one of the critical innovations when 00:30:49.510 --> 00:30:55.350 align:middle line:84% doing any product is to look at the overall-- 00:30:55.350 --> 00:30:57.870 align:middle line:84% and when I say "product," I mean "project" 00:30:57.870 --> 00:31:00.470 align:middle line:90% because it can be services. 00:31:00.470 --> 00:31:07.190 align:middle line:84% And almost analyze task by task and do a kind of triage. 00:31:07.190 --> 00:31:09.290 align:middle line:84% Is this part of the mission critical? 00:31:09.290 --> 00:31:11.150 align:middle line:90% Then it goes into this bucket. 00:31:11.150 --> 00:31:16.810 align:middle line:84% But if there are tasks that are not in that category, 00:31:16.810 --> 00:31:22.230 align:middle line:84% then you can, in your design, put them in a different portion 00:31:22.230 --> 00:31:26.110 align:middle line:84% and allow experimentation here and not 00:31:26.110 --> 00:31:28.630 align:middle line:84% in the ones that have to be mission-critical. 00:31:28.630 --> 00:31:30.910 align:middle line:84% And that's a big change in design 00:31:30.910 --> 00:31:36.570 align:middle line:84% to do that kind of careful analysis of the tasks. 00:31:36.570 --> 00:31:37.070 align:middle line:90% Please. 00:31:37.070 --> 00:31:39.730 align:middle line:84% AUDIENCE: [INAUDIBLE] two comments. 00:31:39.730 --> 00:31:44.370 align:middle line:84% One, Flickr was in gamma for [INAUDIBLE] and-- 00:31:44.370 --> 00:31:46.136 align:middle line:84% IRVING WLADAWSKY-BERGER: Was in gamma? 00:31:46.136 --> 00:31:46.803 align:middle line:90% AUDIENCE: Gamma. 00:31:46.803 --> 00:31:49.990 align:middle line:90% Flickr gamma [INAUDIBLE]. 00:31:49.990 --> 00:31:52.750 align:middle line:84% I would just add that Oracle who does the same sort of thing 00:31:52.750 --> 00:31:53.330 align:middle line:90% as Google. 00:31:53.330 --> 00:31:56.870 align:middle line:84% They're offering SaaS-type delivery now. 00:31:56.870 --> 00:31:59.670 align:middle line:84% And they're adopting the same Google model 00:31:59.670 --> 00:32:01.630 align:middle line:90% for mission-critical apps. 00:32:01.630 --> 00:32:04.990 align:middle line:84% They're putting out incremental small patches every single day 00:32:04.990 --> 00:32:07.950 align:middle line:90% with a back [INAUDIBLE]. 00:32:07.950 --> 00:32:12.670 align:middle line:84% So it's not just to please or avoid criticism 00:32:12.670 --> 00:32:16.510 align:middle line:84% from the customer by putting out the beta [INAUDIBLE] 00:32:16.510 --> 00:32:20.232 align:middle line:90% make any difference. 00:32:20.232 --> 00:32:21.690 align:middle line:84% IRVING WLADAWSKY-BERGER: Very good. 00:32:21.690 --> 00:32:24.210 align:middle line:90% 00:32:24.210 --> 00:32:24.710 align:middle line:90% Oops. 00:32:24.710 --> 00:32:30.870 align:middle line:90% 00:32:30.870 --> 00:32:38.290 align:middle line:84% So now, let's switch over to democratizing innovation, 00:32:38.290 --> 00:32:45.670 align:middle line:84% which is available for nothing from Eric's website, which 00:32:45.670 --> 00:32:47.590 align:middle line:84% is a question you can ask him how come 00:32:47.590 --> 00:32:49.910 align:middle line:90% everybody's selling their book? 00:32:49.910 --> 00:32:59.070 align:middle line:84% He has a hard copy of it, which I would encourage at least some 00:32:59.070 --> 00:33:00.330 align:middle line:90% of you to buy. 00:33:00.330 --> 00:33:03.990 align:middle line:90% 00:33:03.990 --> 00:33:06.790 align:middle line:84% It's a very nice book, but if you 00:33:06.790 --> 00:33:09.110 align:middle line:84% decide you don't believe in buying 00:33:09.110 --> 00:33:10.730 align:middle line:90% what you can get for nothing-- 00:33:10.730 --> 00:33:12.440 align:middle line:90% [LAUGHTER] 00:33:12.440 --> 00:33:15.760 align:middle line:84% --you can download it off his site. 00:33:15.760 --> 00:33:23.920 align:middle line:84% And he makes a very big point about the changing nature 00:33:23.920 --> 00:33:28.960 align:middle line:84% of innovation and the distinction 00:33:28.960 --> 00:33:33.940 align:middle line:84% between manufacturing-based innovation 00:33:33.940 --> 00:33:37.280 align:middle line:84% or classic innovation that takes place 00:33:37.280 --> 00:33:42.800 align:middle line:84% by a manufacturer and user-centric innovation, which 00:33:42.800 --> 00:33:46.700 align:middle line:84% I think it's a little bit easier to talk about in the next slide. 00:33:46.700 --> 00:33:52.440 align:middle line:90% 00:33:52.440 --> 00:33:56.640 align:middle line:84% In manufacturing innovation, what 00:33:56.640 --> 00:34:03.960 align:middle line:84% happens is if I'm the developer, I get all the information, 00:34:03.960 --> 00:34:07.980 align:middle line:84% then I go on, develop the product by myself, build it, 00:34:07.980 --> 00:34:11.860 align:middle line:84% and so on, and then I push it into the marketplace. 00:34:11.860 --> 00:34:14.840 align:middle line:84% That's the classic approach to innovation. 00:34:14.840 --> 00:34:19.920 align:middle line:90% And that's been used forever. 00:34:19.920 --> 00:34:24.679 align:middle line:84% In user-based innovation, the people 00:34:24.679 --> 00:34:31.780 align:middle line:84% using the product themselves are participating in the innovation, 00:34:31.780 --> 00:34:35.540 align:middle line:84% sometimes with the support of the manufacturer, 00:34:35.540 --> 00:34:39.639 align:middle line:84% sometimes without the support of the manufacturer, 00:34:39.639 --> 00:34:43.040 align:middle line:84% whether it's hacking or doing whatever they do. 00:34:43.040 --> 00:34:48.960 align:middle line:84% And what Eric has observed is the growth 00:34:48.960 --> 00:34:54.440 align:middle line:84% of user-based innovation, how important 00:34:54.440 --> 00:34:57.600 align:middle line:84% that has taken place in the old days. 00:34:57.600 --> 00:35:01.680 align:middle line:84% In this case, the users are passive-- they 00:35:01.680 --> 00:35:03.840 align:middle line:90% buy whatever you produce. 00:35:03.840 --> 00:35:09.440 align:middle line:84% In this case, the users are very, very active. 00:35:09.440 --> 00:35:21.280 align:middle line:84% And where user-based innovation is most important 00:35:21.280 --> 00:35:28.440 align:middle line:84% is in the very early stages of a potential product, where 00:35:28.440 --> 00:35:29.880 align:middle line:90% you don't even. 00:35:29.880 --> 00:35:33.000 align:middle line:90% It's almost pre-alpha. 00:35:33.000 --> 00:35:37.480 align:middle line:84% You don't know what it is that you are going to be doing. 00:35:37.480 --> 00:35:41.160 align:middle line:84% And so you very much want leading-edge users 00:35:41.160 --> 00:35:46.120 align:middle line:84% that have the interest to work with you to participate and help 00:35:46.120 --> 00:35:46.800 align:middle line:90% you. 00:35:46.800 --> 00:35:50.360 align:middle line:84% And once things become clearer and you 00:35:50.360 --> 00:35:55.840 align:middle line:84% are in the marketplace looking to scale up production, scale up 00:35:55.840 --> 00:36:01.520 align:middle line:84% support, and deal with users who don't care to innovate with you, 00:36:01.520 --> 00:36:04.800 align:middle line:84% then the more classic manufacturing principles 00:36:04.800 --> 00:36:06.250 align:middle line:90% take over. 00:36:06.250 --> 00:36:14.650 align:middle line:84% And this is extremely important, this user-based innovation, 00:36:14.650 --> 00:36:19.770 align:middle line:84% especially when you're dealing with projects 00:36:19.770 --> 00:36:24.650 align:middle line:84% of the kind of complexity that internet-based projects tend 00:36:24.650 --> 00:36:26.530 align:middle line:90% to have. 00:36:26.530 --> 00:36:32.790 align:middle line:84% Let me give you an example from my experience. 00:36:32.790 --> 00:36:36.930 align:middle line:90% 00:36:36.930 --> 00:36:41.450 align:middle line:84% Before I led IBM's internet project, 00:36:41.450 --> 00:36:46.290 align:middle line:84% I was very involved in supercomputing in IBM. 00:36:46.290 --> 00:36:51.250 align:middle line:84% And in particular, we started a whole line of parallel 00:36:51.250 --> 00:36:57.250 align:middle line:84% supercomputing in the early '90s. 00:36:57.250 --> 00:37:01.650 align:middle line:84% And at that point, that was happening 00:37:01.650 --> 00:37:07.190 align:middle line:84% at the same time as the massive bipolar to CMOS transition 00:37:07.190 --> 00:37:08.770 align:middle line:90% we've talked about. 00:37:08.770 --> 00:37:12.650 align:middle line:84% A similar transition was happening in supercomputing. 00:37:12.650 --> 00:37:15.530 align:middle line:84% And before that, Cray Research was 00:37:15.530 --> 00:37:17.810 align:middle line:90% the leader in supercomputing. 00:37:17.810 --> 00:37:22.750 align:middle line:84% And they had really exotic technologies. 00:37:22.750 --> 00:37:26.490 align:middle line:84% But their machines were incredibly expensive. 00:37:26.490 --> 00:37:33.330 align:middle line:84% And parallel architectures based on initially risk technology 00:37:33.330 --> 00:37:38.370 align:middle line:84% started showing up and gave tremendous power 00:37:38.370 --> 00:37:40.270 align:middle line:90% at a fraction of the cost. 00:37:40.270 --> 00:37:44.210 align:middle line:90% 00:37:44.210 --> 00:37:50.450 align:middle line:84% We honestly didn't know what the hell we should be building. 00:37:50.450 --> 00:38:00.050 align:middle line:84% And what we did was pick a couple of key customers 00:38:00.050 --> 00:38:04.810 align:middle line:84% to be our launch customers and do something just 00:38:04.810 --> 00:38:10.050 align:middle line:84% for these two customers as quickly as possible. 00:38:10.050 --> 00:38:14.210 align:middle line:84% Now, the two customers we picked were the Argonne National 00:38:14.210 --> 00:38:19.170 align:middle line:84% Laboratory, which got DOE funding to participate here, 00:38:19.170 --> 00:38:25.890 align:middle line:84% and the Cornell Supercomputing Center, which is an NSF center. 00:38:25.890 --> 00:38:33.730 align:middle line:84% Well, when you're dealing with universities and research labs, 00:38:33.730 --> 00:38:36.330 align:middle line:84% it's been my experience that they consider it 00:38:36.330 --> 00:38:41.730 align:middle line:84% an insult if you don't trust them enough to have them 00:38:41.730 --> 00:38:44.810 align:middle line:90% as lead users. 00:38:44.810 --> 00:38:53.930 align:middle line:84% In fact, before the new IBM era, post-near-death experience, 00:38:53.930 --> 00:38:57.330 align:middle line:84% IBM didn't know how to deal with them because products 00:38:57.330 --> 00:38:58.270 align:middle line:90% were products. 00:38:58.270 --> 00:38:59.910 align:middle line:84% And we say, no, here is the product. 00:38:59.910 --> 00:39:02.780 align:middle line:84% And they would say, well, but it's finished. 00:39:02.780 --> 00:39:04.160 align:middle line:90% Well, of course it's finished. 00:39:04.160 --> 00:39:05.720 align:middle line:90% We only do finished product. 00:39:05.720 --> 00:39:08.080 align:middle line:84% But we don't like finished products. 00:39:08.080 --> 00:39:10.640 align:middle line:90% 00:39:10.640 --> 00:39:13.240 align:middle line:84% That only means you don't trust us. 00:39:13.240 --> 00:39:15.760 align:middle line:84% You don't think we should participate. 00:39:15.760 --> 00:39:17.560 align:middle line:84% And the answer was probably, yeah, 00:39:17.560 --> 00:39:19.540 align:middle line:84% we don't really trust you because you're 00:39:19.540 --> 00:39:20.540 align:middle line:90% a pain in the ass. 00:39:20.540 --> 00:39:25.020 align:middle line:84% And you're so different from our classic commercial customers. 00:39:25.020 --> 00:39:27.820 align:middle line:84% So maybe we won't do business with you. 00:39:27.820 --> 00:39:32.500 align:middle line:84% And by the way, I'd pay the huge prices 00:39:32.500 --> 00:39:37.040 align:middle line:84% of being removed from tremendous innovation, which IBM did, 00:39:37.040 --> 00:39:38.940 align:middle line:90% and others would do that. 00:39:38.940 --> 00:39:45.980 align:middle line:84% But not only did the Argonne Lab and the Cornell Supercomputing 00:39:45.980 --> 00:39:50.300 align:middle line:84% Center were our clients, but in fact, the reason 00:39:50.300 --> 00:39:54.720 align:middle line:84% they got their funding from DOE and the NSF, 00:39:54.720 --> 00:39:59.540 align:middle line:84% respectively, is because they went 00:39:59.540 --> 00:40:04.540 align:middle line:84% to make the case that they would work with IBM to influence 00:40:04.540 --> 00:40:07.620 align:middle line:90% IBM's design. 00:40:07.620 --> 00:40:13.100 align:middle line:84% So that was a feature of what they were doing, because that's 00:40:13.100 --> 00:40:15.660 align:middle line:90% what a National Lab should do-- 00:40:15.660 --> 00:40:18.860 align:middle line:84% have an impact on the private sector. 00:40:18.860 --> 00:40:23.540 align:middle line:84% And that's what an NSF supercomputing center would do. 00:40:23.540 --> 00:40:26.880 align:middle line:84% And the first machines shipped in April. 00:40:26.880 --> 00:40:30.220 align:middle line:90% And essentially. 00:40:30.220 --> 00:40:32.560 align:middle line:84% I mean, they were really not very good. 00:40:32.560 --> 00:40:35.780 align:middle line:84% By that, I mean, they were incomplete. 00:40:35.780 --> 00:40:38.700 align:middle line:84% And you had to do a lot of hand-holding. 00:40:38.700 --> 00:40:40.800 align:middle line:84% People from the lab had to go there. 00:40:40.800 --> 00:40:45.060 align:middle line:84% But you're dealing with national labs. 00:40:45.060 --> 00:40:48.940 align:middle line:84% Hey, these people, what's a little assembly language 00:40:48.940 --> 00:40:53.360 align:middle line:84% programming or doing hand-coding parallel architecture? 00:40:53.360 --> 00:40:57.940 align:middle line:84% In fact, they almost became an integral part of our development 00:40:57.940 --> 00:41:00.710 align:middle line:90% process in doing this stuff. 00:41:00.710 --> 00:41:05.870 align:middle line:84% And the work with them allowed us to then release 00:41:05.870 --> 00:41:09.770 align:middle line:84% the first version of the product in a few months later. 00:41:09.770 --> 00:41:12.850 align:middle line:84% I think it was in October of '93. 00:41:12.850 --> 00:41:15.790 align:middle line:84% We shipped the first machines to them April '93. 00:41:15.790 --> 00:41:18.510 align:middle line:84% Then we shipped the first version of the machine 00:41:18.510 --> 00:41:19.630 align:middle line:90% in October. 00:41:19.630 --> 00:41:22.890 align:middle line:84% And at that point, we had a few more customers, 00:41:22.890 --> 00:41:24.970 align:middle line:84% including some commercial customers. 00:41:24.970 --> 00:41:27.070 align:middle line:84% But in supercomputing, commercial customers 00:41:27.070 --> 00:41:31.390 align:middle line:84% means petroleum companies and engineering companies. 00:41:31.390 --> 00:41:34.910 align:middle line:84% And then the following year, in '94, we really 00:41:34.910 --> 00:41:39.990 align:middle line:90% announced the product. 00:41:39.990 --> 00:41:43.070 align:middle line:90% Became known as the SP2. 00:41:43.070 --> 00:41:49.110 align:middle line:84% And by the product, I mean any kind of classic cycle would say, 00:41:49.110 --> 00:41:52.190 align:middle line:84% you don't announce until you have this. 00:41:52.190 --> 00:41:54.390 align:middle line:84% But by the time we announced the SP2, 00:41:54.390 --> 00:41:57.910 align:middle line:84% we had been in the marketplace for a year. 00:41:57.910 --> 00:42:00.950 align:middle line:84% And it's almost like our people in the lab 00:42:00.950 --> 00:42:06.030 align:middle line:84% And even our people in the field had been in spring training 00:42:06.030 --> 00:42:07.650 align:middle line:84% in, how do you work with-- remember, 00:42:07.650 --> 00:42:10.390 align:middle line:84% there's a whole process to how do 00:42:10.390 --> 00:42:12.910 align:middle line:84% you support customers, hold their hands, 00:42:12.910 --> 00:42:14.290 align:middle line:90% how many people you need? 00:42:14.290 --> 00:42:21.810 align:middle line:84% And if you started practicing that once you ship the product, 00:42:21.810 --> 00:42:25.710 align:middle line:84% you're in deep trouble, because the customers 00:42:25.710 --> 00:42:27.350 align:middle line:90% will want you to do. 00:42:27.350 --> 00:42:35.430 align:middle line:84% But by picking the right leading-edge customers, 00:42:35.430 --> 00:42:40.610 align:middle line:84% all of a sudden, it just plays into your marketing. 00:42:40.610 --> 00:42:44.030 align:middle line:84% It just totally plays into your marketing. 00:42:44.030 --> 00:42:50.270 align:middle line:84% I've talked about an example for a new product 00:42:50.270 --> 00:42:54.350 align:middle line:84% and a new pretty research-oriented product. 00:42:54.350 --> 00:42:57.390 align:middle line:84% Now with commercial products, whether it's 00:42:57.390 --> 00:43:03.350 align:middle line:84% mainframes or other things, the equivalent 00:43:03.350 --> 00:43:08.510 align:middle line:84% is to pick selected clients with whom you have a very, very 00:43:08.510 --> 00:43:10.990 align:middle line:90% close relationship. 00:43:10.990 --> 00:43:14.730 align:middle line:84% And that you typically pick five of them, 00:43:14.730 --> 00:43:20.550 align:middle line:84% would you say, or so, who become your early clients, who've 00:43:20.550 --> 00:43:26.470 align:middle line:84% been dying for the features that this release is taking 00:43:26.470 --> 00:43:31.830 align:middle line:84% to market, and with whom you have such a close relationship 00:43:31.830 --> 00:43:35.570 align:middle line:84% that even if they are really upset at you, 00:43:35.570 --> 00:43:38.870 align:middle line:84% they won't go write blogs about it. 00:43:38.870 --> 00:43:42.610 align:middle line:84% But they may go back to the original release, 00:43:42.610 --> 00:43:45.630 align:middle line:84% but they would not embarrass you. 00:43:45.630 --> 00:43:47.870 align:middle line:84% And so you pick them very carefully. 00:43:47.870 --> 00:43:52.230 align:middle line:84% And you work very carefully with you. 00:43:52.230 --> 00:43:55.800 align:middle line:84% A big part of what you do here is 00:43:55.800 --> 00:44:00.500 align:middle line:84% you are supporting those early adopters directly from the lab. 00:44:00.500 --> 00:44:04.200 align:middle line:84% So I mean, here, you can do that. 00:44:04.200 --> 00:44:09.880 align:middle line:84% Here, you need the whole field support mechanism and everybody 00:44:09.880 --> 00:44:10.640 align:middle line:90% else. 00:44:10.640 --> 00:44:14.280 align:middle line:84% But in fact, I think what usually determines 00:44:14.280 --> 00:44:16.800 align:middle line:84% how many of these you can do is how many can you 00:44:16.800 --> 00:44:20.880 align:middle line:84% afford to support out of the lab, and people flying in 00:44:20.880 --> 00:44:23.720 align:middle line:84% and people doing all kinds of things. 00:44:23.720 --> 00:44:28.520 align:middle line:90% And then you move in this area. 00:44:28.520 --> 00:44:32.905 align:middle line:84% Now, I would expect in the pharmaceutical companies-- 00:44:32.905 --> 00:44:34.280 align:middle line:84% I mean, let's take somebody who's 00:44:34.280 --> 00:44:39.040 align:middle line:84% doing an experimental drug to cure a certain kind of leukemia. 00:44:39.040 --> 00:44:43.360 align:middle line:84% And there are patients that they've tried everything else. 00:44:43.360 --> 00:44:48.280 align:middle line:84% And can you get permission from the-- 00:44:48.280 --> 00:44:52.280 align:middle line:90% not the FAA, the FDA to-- 00:44:52.280 --> 00:44:56.720 align:middle line:84% AUDIENCE: I guess you have accelerated clinical trials. 00:44:56.720 --> 00:44:57.253 align:middle line:90% [INAUDIBLE] 00:44:57.253 --> 00:44:58.920 align:middle line:84% IRVING WLADAWSKY-BERGER: So you might be 00:44:58.920 --> 00:45:00.720 align:middle line:90% able to get special permission. 00:45:00.720 --> 00:45:03.440 align:middle line:84% AUDIENCE: [INAUDIBLE] for phase 1. 00:45:03.440 --> 00:45:04.560 align:middle line:90% It's safe. 00:45:04.560 --> 00:45:07.040 align:middle line:84% And you might be able to do that. 00:45:07.040 --> 00:45:08.880 align:middle line:84% You might be able to combine that phase 1 00:45:08.880 --> 00:45:12.340 align:middle line:84% safety on these patients with these patients, so 00:45:12.340 --> 00:45:15.680 align:middle line:84% that [INAUDIBLE] they are faster than the market 00:45:15.680 --> 00:45:18.960 align:middle line:84% and faster than the [INAUDIBLE] result. 00:45:18.960 --> 00:45:20.520 align:middle line:84% And that's what actually we ended up 00:45:20.520 --> 00:45:25.360 align:middle line:84% doing with a current run [INAUDIBLE] clinical trials. 00:45:25.360 --> 00:45:28.640 align:middle line:84% We tested on a bunch of patients for whom 00:45:28.640 --> 00:45:30.453 align:middle line:84% the traditional therapies did not work. 00:45:30.453 --> 00:45:32.120 align:middle line:84% IRVING WLADAWSKY-BERGER: Yeah, so that's 00:45:32.120 --> 00:45:40.300 align:middle line:84% an example where these people are superb leading-edge users. 00:45:40.300 --> 00:45:42.600 align:middle line:84% And they're the people who take care of them 00:45:42.600 --> 00:45:45.760 align:middle line:90% should be big innovators, right? 00:45:45.760 --> 00:45:49.000 align:middle line:90% I know people have said-- 00:45:49.000 --> 00:45:53.220 align:middle line:84% there's this famous hospital in Houston, the Anderson Hospital. 00:45:53.220 --> 00:45:55.920 align:middle line:90% 00:45:55.920 --> 00:45:58.680 align:middle line:84% I think it's the Anderson Hospital. 00:45:58.680 --> 00:46:02.800 align:middle line:84% If you go to Anderson, that's where they 00:46:02.800 --> 00:46:05.640 align:middle line:90% do the most advanced research. 00:46:05.640 --> 00:46:09.200 align:middle line:84% But it's not good to land in Anderson. 00:46:09.200 --> 00:46:10.880 align:middle line:90% [LAUGHTER] 00:46:10.880 --> 00:46:12.780 align:middle line:84% Because, I mean, it's not a joke, 00:46:12.780 --> 00:46:20.300 align:middle line:84% but It means all the other things have not worked. 00:46:20.300 --> 00:46:21.740 align:middle line:90% So you are now landing-- 00:46:21.740 --> 00:46:26.400 align:middle line:84% I mean, for what it does, it may be one of the world's-- 00:46:26.400 --> 00:46:28.720 align:middle line:84% the first heart transplant by DeBakey 00:46:28.720 --> 00:46:31.480 align:middle line:90% was done at Anderson in Houston. 00:46:31.480 --> 00:46:33.680 align:middle line:90% But you don't want to go there. 00:46:33.680 --> 00:46:35.600 align:middle line:90% You want to stay away from that. 00:46:35.600 --> 00:46:40.820 align:middle line:84% So this is a very interesting paradigm and approach. 00:46:40.820 --> 00:46:42.320 align:middle line:84% And by the way, as we are talking, 00:46:42.320 --> 00:46:47.920 align:middle line:84% keep assembling questions to discuss with Professor von 00:46:47.920 --> 00:46:51.490 align:middle line:90% Hippel when he gets here. 00:46:51.490 --> 00:47:05.650 align:middle line:84% Now, when IBM in the '70s and '80s decided, 00:47:05.650 --> 00:47:11.730 align:middle line:84% for a variety of reasons, to concentrate on just more pure 00:47:11.730 --> 00:47:18.930 align:middle line:84% commercial markets and commercial customers and avoid 00:47:18.930 --> 00:47:21.670 align:middle line:84% universities, avoid research labs, 00:47:21.670 --> 00:47:25.190 align:middle line:84% because we didn't have those products, we could say, well, 00:47:25.190 --> 00:47:27.330 align:middle line:90% there was a segment you had. 00:47:27.330 --> 00:47:35.210 align:middle line:84% But we knew it even then, but in retrospect, it's clear. 00:47:35.210 --> 00:47:41.730 align:middle line:84% What we were doing is cutting ourselves off totally 00:47:41.730 --> 00:47:46.730 align:middle line:84% from this incredible source of innovation. 00:47:46.730 --> 00:47:56.130 align:middle line:84% So it's well accepted that the supercomputing community 00:47:56.130 --> 00:47:59.710 align:middle line:84% is working on the kinds of products that, 00:47:59.710 --> 00:48:02.810 align:middle line:84% a few years later, will gravitate 00:48:02.810 --> 00:48:05.210 align:middle line:90% toward the general marketplace. 00:48:05.210 --> 00:48:07.210 align:middle line:84% And if anybody wants me to give you 00:48:07.210 --> 00:48:11.630 align:middle line:84% exhibit A, how about the internet and world wide web, 00:48:11.630 --> 00:48:15.390 align:middle line:84% which all came from the supercomputing community? 00:48:15.390 --> 00:48:17.650 align:middle line:84% They were the first users of both 00:48:17.650 --> 00:48:21.690 align:middle line:84% of them, let alone parallel architectures, the browser, 00:48:21.690 --> 00:48:23.890 align:middle line:90% and on and on and on and on. 00:48:23.890 --> 00:48:33.850 align:middle line:84% So these lead users are incredibly important. 00:48:33.850 --> 00:48:39.410 align:middle line:84% Because no matter how good your labs are, 00:48:39.410 --> 00:48:44.530 align:middle line:84% your own labs are different than the users'. 00:48:44.530 --> 00:48:47.210 align:middle line:84% They tend to be more developing technologies, 00:48:47.210 --> 00:48:49.530 align:middle line:90% developing products. 00:48:49.530 --> 00:48:52.490 align:middle line:84% And your people in the field are usually, 00:48:52.490 --> 00:48:56.130 align:middle line:84% one, not users, and two, they may not 00:48:56.130 --> 00:48:58.610 align:middle line:90% be as sharp as the users. 00:48:58.610 --> 00:49:01.390 align:middle line:84% They may be brilliantly sharp people. 00:49:01.390 --> 00:49:06.050 align:middle line:84% It's just that these people know their stuff so much better. 00:49:06.050 --> 00:49:15.170 align:middle line:84% So cutting yourself off from dealing with these users 00:49:15.170 --> 00:49:16.590 align:middle line:90% is a sin. 00:49:16.590 --> 00:49:22.170 align:middle line:84% I mean, not a sin-- it's bad business practice, 00:49:22.170 --> 00:49:25.850 align:middle line:84% because you're cutting yourself off from one of the biggest 00:49:25.850 --> 00:49:27.390 align:middle line:90% sources of marketplace. 00:49:27.390 --> 00:49:29.370 align:middle line:84% Let me give you another example-- marketplace 00:49:29.370 --> 00:49:32.570 align:middle line:90% innovation. 00:49:32.570 --> 00:49:38.250 align:middle line:84% A few years ago, I had just become the chairman 00:49:38.250 --> 00:49:40.550 align:middle line:90% of IBM's Academy of Technology. 00:49:40.550 --> 00:49:44.410 align:middle line:84% And some people came to see me to give me 00:49:44.410 --> 00:49:49.980 align:middle line:84% the results of a major study that the Academy had conducted. 00:49:49.980 --> 00:49:54.940 align:middle line:84% And it was the impact of game players and related 00:49:54.940 --> 00:49:56.740 align:middle line:90% technologies. 00:49:56.740 --> 00:49:59.440 align:middle line:84% And they said IBM has to be involved in that. 00:49:59.440 --> 00:50:03.000 align:middle line:84% And I had no idea what they were talking about. 00:50:03.000 --> 00:50:09.020 align:middle line:84% But this is the Academy, so you listen. 00:50:09.020 --> 00:50:13.860 align:middle line:90% 00:50:13.860 --> 00:50:19.220 align:middle line:84% And essentially, the point was that in the same way 00:50:19.220 --> 00:50:24.860 align:middle line:84% that supercomputing is foreshadowing where fields are 00:50:24.860 --> 00:50:31.180 align:middle line:84% going, a lot of what's going on in game players, 00:50:31.180 --> 00:50:35.300 align:middle line:84% a lot of what's going on in massively multiplayer 00:50:35.300 --> 00:50:38.140 align:middle line:84% online games, World of Warcraft, and then 00:50:38.140 --> 00:50:41.380 align:middle line:84% role-playing games like Second Life, 00:50:41.380 --> 00:50:46.080 align:middle line:84% are leading-edge indicators of where things are going to go. 00:50:46.080 --> 00:50:49.480 align:middle line:84% So I took that very, very seriously. 00:50:49.480 --> 00:50:52.360 align:middle line:84% And then we charter a few additional studies, 00:50:52.360 --> 00:50:55.340 align:middle line:84% which eventually led to IBM becoming 00:50:55.340 --> 00:51:00.500 align:middle line:84% very active in virtual worlds, and things like that. 00:51:00.500 --> 00:51:07.620 align:middle line:84% Another manifestation of this area 00:51:07.620 --> 00:51:13.180 align:middle line:84% is what you see happening, let's say, in social networks, 00:51:13.180 --> 00:51:19.380 align:middle line:84% where most of the time in social networks, Myspace and Facebook 00:51:19.380 --> 00:51:23.420 align:middle line:84% and so on, as you know, I think Microsoft 00:51:23.420 --> 00:51:29.940 align:middle line:84% is going to buy 5% or 10% of Facebook for $1.5 billion. 00:51:29.940 --> 00:51:34.080 align:middle line:84% But Facebook was done by a couple of kids from Harvard. 00:51:34.080 --> 00:51:36.820 align:middle line:84% I don't think it's clear which couple of kids. 00:51:36.820 --> 00:51:41.540 align:middle line:84% I think there are some lawsuits going on about that. 00:51:41.540 --> 00:51:46.260 align:middle line:84% And I don't know where Myspace came from. 00:51:46.260 --> 00:51:49.620 align:middle line:90% And Craigslist was done by-- 00:51:49.620 --> 00:51:51.540 align:middle line:90% I forget his first name, Craig. 00:51:51.540 --> 00:51:52.740 align:middle line:90% Craig. 00:51:52.740 --> 00:51:54.470 align:middle line:90% Craig is his first name? 00:51:54.470 --> 00:51:55.220 align:middle line:90% AUDIENCE: Newmark. 00:51:55.220 --> 00:51:56.180 align:middle line:90% IRVING WLADAWSKY-BERGER: Huh? 00:51:56.180 --> 00:51:57.180 align:middle line:90% AUDIENCE: Craig Newmark. 00:51:57.180 --> 00:51:59.540 align:middle line:84% IRVING WLADAWSKY-BERGER: Craig Newmark? 00:51:59.540 --> 00:52:06.540 align:middle line:84% And this is a very good example of user innovation, all 00:52:06.540 --> 00:52:08.940 align:middle line:84% these areas that these people, whether it's 00:52:08.940 --> 00:52:15.880 align:middle line:84% Facebook in the Harvard example or the other ones, 00:52:15.880 --> 00:52:18.760 align:middle line:84% they were just done for themselves. 00:52:18.760 --> 00:52:21.120 align:middle line:84% And they started to do more and more and more. 00:52:21.120 --> 00:52:24.980 align:middle line:84% And then they found there was a market here. 00:52:24.980 --> 00:52:27.980 align:middle line:84% And eventually, the commercial world 00:52:27.980 --> 00:52:30.940 align:middle line:84% discovered there was a market there. 00:52:30.940 --> 00:52:35.340 align:middle line:84% But what's very important here, you are a business, 00:52:35.340 --> 00:52:37.420 align:middle line:84% and you're not watching what's going 00:52:37.420 --> 00:52:39.680 align:middle line:90% on in those people who owe you. 00:52:39.680 --> 00:52:41.000 align:middle line:90% That's all I can say. 00:52:41.000 --> 00:52:47.670 align:middle line:84% I mean, something can happen, and you'll be in deep trouble. 00:52:47.670 --> 00:52:54.950 align:middle line:84% Now we've talked about a number of things 00:52:54.950 --> 00:52:59.710 align:middle line:84% here that lead users, by definition of lead users, 00:52:59.710 --> 00:53:03.430 align:middle line:90% are at the edge of this market. 00:53:03.430 --> 00:53:12.350 align:middle line:84% So in '93, 92/93, our lead users were the people most interested 00:53:12.350 --> 00:53:14.390 align:middle line:90% in parallel architectures. 00:53:14.390 --> 00:53:17.350 align:middle line:84% Our lead users now in supercomputing 00:53:17.350 --> 00:53:21.050 align:middle line:84% are interested in petaflop supercomputers. 00:53:21.050 --> 00:53:24.850 align:middle line:84% Petaflops, just enormous machines 00:53:24.850 --> 00:53:27.110 align:middle line:84% in Los Alamos National Lab, University 00:53:27.110 --> 00:53:31.830 align:middle line:84% of Illinois NSF Center, Argonne National Lab 00:53:31.830 --> 00:53:33.670 align:middle line:90% continues to be there. 00:53:33.670 --> 00:53:40.270 align:middle line:84% So these are the people that are just pushing the envelopes 00:53:40.270 --> 00:53:42.250 align:middle line:90% at every single level. 00:53:42.250 --> 00:53:46.670 align:middle line:84% They just want to keep the latest, 00:53:46.670 --> 00:53:52.750 align:middle line:84% and they will only partner with you 00:53:52.750 --> 00:53:57.670 align:middle line:84% and be your lead users, not your competitors, 00:53:57.670 --> 00:54:01.770 align:middle line:84% if you're willing to help build what they are looking for. 00:54:01.770 --> 00:54:04.250 align:middle line:90% So this is very, very important. 00:54:04.250 --> 00:54:06.650 align:middle line:84% By the way, you may decide you can't afford it. 00:54:06.650 --> 00:54:09.610 align:middle line:84% You may decide it's diverting your energy. 00:54:09.610 --> 00:54:13.490 align:middle line:84% So these are things you have to do very, very carefully. 00:54:13.490 --> 00:54:21.830 align:middle line:90% 00:54:21.830 --> 00:54:28.790 align:middle line:84% Another point that Eric makes in his book 00:54:28.790 --> 00:54:34.750 align:middle line:84% is the difference in knowledge that the more classic 00:54:34.750 --> 00:54:41.550 align:middle line:84% manufacturing people have versus the users. 00:54:41.550 --> 00:54:45.910 align:middle line:84% The users, by definition of users, 00:54:45.910 --> 00:54:49.670 align:middle line:84% are expert on their requirements. 00:54:49.670 --> 00:54:52.930 align:middle line:84% I mean, that shouldn't be a big theme. 00:54:52.930 --> 00:54:56.990 align:middle line:84% But if somebody came here and said, 00:54:56.990 --> 00:55:00.470 align:middle line:84% no, no, the experts on the users requirement 00:55:00.470 --> 00:55:06.030 align:middle line:84% are market analysis people-- so you hire your favorite market 00:55:06.030 --> 00:55:10.990 align:middle line:84% research company, and they will tell you, no, no. 00:55:10.990 --> 00:55:14.990 align:middle line:84% Remember, market analysis, by definition, 00:55:14.990 --> 00:55:19.950 align:middle line:84% means you need information to analyze. 00:55:19.950 --> 00:55:24.030 align:middle line:84% When you are here in the marketplace, 00:55:24.030 --> 00:55:26.350 align:middle line:90% you have information. 00:55:26.350 --> 00:55:30.190 align:middle line:84% So that's when those companies can do their job. 00:55:30.190 --> 00:55:33.950 align:middle line:84% When you are here, there is no information. 00:55:33.950 --> 00:55:38.840 align:middle line:84% The information is in the head of these leading-edge users. 00:55:38.840 --> 00:55:43.760 align:middle line:84% Which is why if you try to apply the processes that 00:55:43.760 --> 00:55:47.360 align:middle line:84% work for more mature products, where 00:55:47.360 --> 00:55:51.200 align:middle line:84% you get a lot of information and you do segmentation, 00:55:51.200 --> 00:55:56.080 align:middle line:84% you just totally miss these end users. 00:55:56.080 --> 00:56:02.260 align:middle line:84% So the users are the ones who have the information. 00:56:02.260 --> 00:56:10.400 align:middle line:84% Now, the manufacturers have a pretty good idea 00:56:10.400 --> 00:56:13.580 align:middle line:90% about what they can build. 00:56:13.580 --> 00:56:18.320 align:middle line:84% They don't know if anybody would buy it. 00:56:18.320 --> 00:56:21.920 align:middle line:84% They have a big space they could build. 00:56:21.920 --> 00:56:24.380 align:middle line:84% I mean, let's take an example, which, 00:56:24.380 --> 00:56:29.880 align:middle line:84% again, is at the leading edge of petaflop supercomputing. 00:56:29.880 --> 00:56:34.000 align:middle line:84% I mean, if somebody here says, I want a petaflop supercomputer 00:56:34.000 --> 00:56:39.800 align:middle line:84% for a certain amount of money, and the manufacturers 00:56:39.800 --> 00:56:43.220 align:middle line:84% have no idea how to do that, that's that. 00:56:43.220 --> 00:56:47.780 align:middle line:84% You want it, but we don't know how to build it. 00:56:47.780 --> 00:56:49.560 align:middle line:90% So you have to wait. 00:56:49.560 --> 00:56:52.480 align:middle line:84% What happens is that eventually, you 00:56:52.480 --> 00:56:57.920 align:middle line:84% have more and more information about what kinds of approaches 00:56:57.920 --> 00:57:00.720 align:middle line:84% you could take to a petaflop supercomputer. 00:57:00.720 --> 00:57:03.480 align:middle line:84% And there will be multiple approaches. 00:57:03.480 --> 00:57:08.040 align:middle line:84% And these people know the kinds of problems they want solved. 00:57:08.040 --> 00:57:11.440 align:middle line:84% And so now you start working with them to see, 00:57:11.440 --> 00:57:18.360 align:middle line:84% is what we can do something that can apply to your problems? 00:57:18.360 --> 00:57:21.760 align:middle line:84% And if you start getting a match, then all of a sudden, 00:57:21.760 --> 00:57:24.680 align:middle line:84% you start making the connection between the manufacturer 00:57:24.680 --> 00:57:25.600 align:middle line:90% and the user. 00:57:25.600 --> 00:57:26.120 align:middle line:90% Yes. 00:57:26.120 --> 00:57:28.440 align:middle line:84% AUDIENCE: In some of the examples that you've mentioned 00:57:28.440 --> 00:57:31.120 align:middle line:84% in the last few minutes, it seems like the manufacturing 00:57:31.120 --> 00:57:33.640 align:middle line:84% and the user, especially the lead users, 00:57:33.640 --> 00:57:35.500 align:middle line:90% are actually one and the same. 00:57:35.500 --> 00:57:38.760 align:middle line:84% So if you think about Mark Zuckerberg and Facebook or Craig 00:57:38.760 --> 00:57:41.960 align:middle line:84% and Craigslist and even in the supercomputing example-- 00:57:41.960 --> 00:57:44.127 align:middle line:84% IRVING WLADAWSKY-BERGER: They were the face of the-- 00:57:44.127 --> 00:57:48.640 align:middle line:84% no, no, let's say Facebook, they were, for instance, 00:57:48.640 --> 00:57:49.620 align:middle line:90% the beginning. 00:57:49.620 --> 00:57:55.380 align:middle line:84% But then weren't there other kids at Harvard using Facebook? 00:57:55.380 --> 00:57:57.720 align:middle line:84% AUDIENCE: There definitely were, but the manufacturers 00:57:57.720 --> 00:57:59.300 align:middle line:90% were users themselves. 00:57:59.300 --> 00:58:01.200 align:middle line:84% So Mark Zuckerberg was a student who 00:58:01.200 --> 00:58:03.180 align:middle line:90% wanted to be able to interact. 00:58:03.180 --> 00:58:04.720 align:middle line:90% And Craig wanted to sell. 00:58:04.720 --> 00:58:08.260 align:middle line:84% With the supercomputing example, you didn't have that. 00:58:08.260 --> 00:58:12.280 align:middle line:84% But you ended up opening up supercomputing source code 00:58:12.280 --> 00:58:15.440 align:middle line:84% and the infrastructure so that the users became manufacturers 00:58:15.440 --> 00:58:16.440 align:middle line:90% to some degree. 00:58:16.440 --> 00:58:16.980 align:middle line:84% IRVING WLADAWSKY-BERGER: Exactly. 00:58:16.980 --> 00:58:17.880 align:middle line:84% AUDIENCE: So in every example, it 00:58:17.880 --> 00:58:20.220 align:middle line:84% seems like instead of having the distinction-- 00:58:20.220 --> 00:58:21.762 align:middle line:84% IRVING WLADAWSKY-BERGER: No, exactly. 00:58:21.762 --> 00:58:24.920 align:middle line:84% In fact, that is the most critical distinction 00:58:24.920 --> 00:58:29.120 align:middle line:84% that whereas traditional you have 00:58:29.120 --> 00:58:35.650 align:middle line:84% a gap between the manufacturers and the users, 00:58:35.650 --> 00:58:37.910 align:middle line:84% in the world we're talking about, 00:58:37.910 --> 00:58:42.170 align:middle line:84% and we will discuss how big is that world, 00:58:42.170 --> 00:58:46.410 align:middle line:84% those two are much closer together than ever. 00:58:46.410 --> 00:58:49.195 align:middle line:90% And that's your point, correct? 00:58:49.195 --> 00:58:50.570 align:middle line:84% AUDIENCE: Can it happen in a case 00:58:50.570 --> 00:58:52.410 align:middle line:84% where the manufacturer isn't the user 00:58:52.410 --> 00:58:55.675 align:middle line:84% or the user isn't able to manufacture? 00:58:55.675 --> 00:58:57.050 align:middle line:84% IRVING WLADAWSKY-BERGER: Well, it 00:58:57.050 --> 00:58:59.150 align:middle line:84% depends what you mean by manufacture. 00:58:59.150 --> 00:59:07.530 align:middle line:84% If we include software, and if you develop a layer of open 00:59:07.530 --> 00:59:11.650 align:middle line:84% source and you encourage your users to innovate in those 00:59:11.650 --> 00:59:14.770 align:middle line:84% layers, then your users are, quote, unquote, 00:59:14.770 --> 00:59:17.210 align:middle line:90% "manufacturing"-- 00:59:17.210 --> 00:59:20.990 align:middle line:84% I mean, we can now argue, are they designing or manufacturing? 00:59:20.990 --> 00:59:25.930 align:middle line:84% But in the case of software, the difference is not very large. 00:59:25.930 --> 00:59:30.610 align:middle line:84% But no, I think the users, by this definition, 00:59:30.610 --> 00:59:35.010 align:middle line:84% have to be actively involved in helping you design and build. 00:59:35.010 --> 00:59:37.390 align:middle line:84% And each case would be a little different, 00:59:37.390 --> 00:59:39.530 align:middle line:90% how active they would be. 00:59:39.530 --> 00:59:40.933 align:middle line:90% Yes, please. 00:59:40.933 --> 00:59:42.350 align:middle line:84% AUDIENCE: Come back to this point. 00:59:42.350 --> 00:59:45.750 align:middle line:84% I think in the case of software and the simple software, 00:59:45.750 --> 00:59:49.050 align:middle line:84% it's easy for the user to be the manufacturer. 00:59:49.050 --> 00:59:50.550 align:middle line:90% But in other cases, it may not. 00:59:50.550 --> 00:59:54.470 align:middle line:84% I might want to claim that it's very fuel-efficient, 00:59:54.470 --> 00:59:55.930 align:middle line:90% but I can't go and make it. 00:59:55.930 --> 00:59:56.350 align:middle line:90% IRVING WLADAWSKY-BERGER: No. 00:59:56.350 --> 00:59:58.650 align:middle line:84% AUDIENCE: The least I can do is ask somebody to. 00:59:58.650 --> 01:00:02.570 align:middle line:84% IRVING WLADAWSKY-BERGER: But even in Facebook, after a while, 01:00:02.570 --> 01:00:09.330 align:middle line:84% the users of Facebook could require the people 01:00:09.330 --> 01:00:11.130 align:middle line:90% who had Facebook, right? 01:00:11.130 --> 01:00:15.170 align:middle line:84% So after a while, there was more of a distinction 01:00:15.170 --> 01:00:18.010 align:middle line:84% between the manufacturer and the user. 01:00:18.010 --> 01:00:21.590 align:middle line:84% Or in the case of Google, there's a lot you can do, 01:00:21.590 --> 01:00:24.270 align:middle line:84% but there is the two are distinct. 01:00:24.270 --> 01:00:26.450 align:middle line:84% It's just that you're collaborating. 01:00:26.450 --> 01:00:26.993 align:middle line:90% Yes, please. 01:00:26.993 --> 01:00:28.410 align:middle line:84% AUDIENCE: Yeah, I mean, von Hippel 01:00:28.410 --> 01:00:30.972 align:middle line:84% gives a lot of examples that are software-based. 01:00:30.972 --> 01:00:32.430 align:middle line:84% But he also gives us the hard ones, 01:00:32.430 --> 01:00:35.770 align:middle line:84% like the surfer who modifies the board so 01:00:35.770 --> 01:00:36.990 align:middle line:90% that he straps his feet in. 01:00:36.990 --> 01:00:40.470 align:middle line:84% So actually the manufacturer had nothing to do with that. 01:00:40.470 --> 01:00:42.190 align:middle line:84% It wasn't until later that they learned. 01:00:42.190 --> 01:00:45.870 align:middle line:84% And then the whole plant thing where he took the wires. 01:00:45.870 --> 01:00:49.530 align:middle line:84% And the user was a manufacturer, but he wasn't 01:00:49.530 --> 01:00:51.490 align:middle line:90% a manufacturer of that piece. 01:00:51.490 --> 01:00:54.230 align:middle line:84% But he had a better way to use it. 01:00:54.230 --> 01:00:56.150 align:middle line:84% And I think that that happens quite a bit. 01:00:56.150 --> 01:00:58.490 align:middle line:84% At least where I come from, we get something and say, 01:00:58.490 --> 01:00:59.073 align:middle line:90% you know what? 01:00:59.073 --> 01:01:00.290 align:middle line:90% It wasn't what it is. 01:01:00.290 --> 01:01:02.070 align:middle line:84% IRVING WLADAWSKY-BERGER: Remind me, which industry? 01:01:02.070 --> 01:01:04.330 align:middle line:84% AUDIENCE: Well, I'm in the Coast Guard, but we get something 01:01:04.330 --> 01:01:05.413 align:middle line:90% and we say, you know what? 01:01:05.413 --> 01:01:08.190 align:middle line:84% This is OK for this, but it's really good for this. 01:01:08.190 --> 01:01:10.370 align:middle line:84% And we're using it for something totally different. 01:01:10.370 --> 01:01:11.210 align:middle line:90% And-- 01:01:11.210 --> 01:01:12.570 align:middle line:84% IRVING WLADAWSKY-BERGER: And you give the feedback back 01:01:12.570 --> 01:01:13.797 align:middle line:90% to the manufacturer? 01:01:13.797 --> 01:01:14.630 align:middle line:90% AUDIENCE: Sometimes. 01:01:14.630 --> 01:01:15.515 align:middle line:90% [LAUGHS] But-- 01:01:15.515 --> 01:01:16.890 align:middle line:84% IRVING WLADAWSKY-BERGER: Are they 01:01:16.890 --> 01:01:19.450 align:middle line:84% searching for the feedback from you? 01:01:19.450 --> 01:01:21.450 align:middle line:90% AUDIENCE: Not really. 01:01:21.450 --> 01:01:24.870 align:middle line:84% But sometimes if we need them to modify it on a global scale, 01:01:24.870 --> 01:01:31.460 align:middle line:84% we'll say, like at Cisco, we modified the Cisco routers to do 01:01:31.460 --> 01:01:33.700 align:middle line:90% something through RS-232 ports. 01:01:33.700 --> 01:01:35.840 align:middle line:84% And we asked them if they could do it. 01:01:35.840 --> 01:01:36.520 align:middle line:90% They said no. 01:01:36.520 --> 01:01:37.853 align:middle line:90% And we figured out how to do it. 01:01:37.853 --> 01:01:39.145 align:middle line:90% And then we went and told them. 01:01:39.145 --> 01:01:41.280 align:middle line:84% And now they sell it to me to be able to do that. 01:01:41.280 --> 01:01:42.980 align:middle line:84% IRVING WLADAWSKY-BERGER: Did you give them the code? 01:01:42.980 --> 01:01:43.420 align:middle line:90% AUDIENCE: Yeah. 01:01:43.420 --> 01:01:44.040 align:middle line:90% Yeah, absolutely. 01:01:44.040 --> 01:01:45.400 align:middle line:84% Because we were like, hey, if you do this, 01:01:45.400 --> 01:01:46.440 align:middle line:90% we'll buy a whole bunch. 01:01:46.440 --> 01:01:47.000 align:middle line:84% IRVING WLADAWSKY-BERGER: No, of course. 01:01:47.000 --> 01:01:49.260 align:middle line:84% And then you don't have to support it yourself. 01:01:49.260 --> 01:01:50.105 align:middle line:90% No, no, precisely. 01:01:50.105 --> 01:01:51.480 align:middle line:84% AUDIENCE: They modified it, and-- 01:01:51.480 --> 01:01:52.938 align:middle line:84% IRVING WLADAWSKY-BERGER: But that's 01:01:52.938 --> 01:01:54.980 align:middle line:84% a wonderful example of what we're talking about. 01:01:54.980 --> 01:01:56.700 align:middle line:90% Sung-yu, you had a point? 01:01:56.700 --> 01:01:59.360 align:middle line:90% Sung-yu is Eric's student. 01:01:59.360 --> 01:02:01.380 align:middle line:84% So he knows a lot about this area. 01:02:01.380 --> 01:02:04.880 align:middle line:84% AUDIENCE: So I'm a little biased as to my experience. 01:02:04.880 --> 01:02:09.820 align:middle line:84% But I think the definition between the user innovation 01:02:09.820 --> 01:02:14.400 align:middle line:84% and manufacturing innovation, I have [INAUDIBLE] idea. 01:02:14.400 --> 01:02:19.140 align:middle line:84% I think it's based on how you benefit. 01:02:19.140 --> 01:02:22.420 align:middle line:84% If you benefit from innovating just 01:02:22.420 --> 01:02:24.440 align:middle line:84% by solving the problems for yourself, 01:02:24.440 --> 01:02:26.180 align:middle line:90% then there's user innovation. 01:02:26.180 --> 01:02:30.098 align:middle line:84% If you benefit by selling the innovation that you did, 01:02:30.098 --> 01:02:31.640 align:middle line:84% then it's a manufacturing innovation. 01:02:31.640 --> 01:02:35.140 align:middle line:84% I think the point is a lot of users 01:02:35.140 --> 01:02:40.380 align:middle line:84% innovate because that benefits themselves, not by selling it. 01:02:40.380 --> 01:02:43.340 align:middle line:84% And these happen a lot in the marketplace. 01:02:43.340 --> 01:02:46.540 align:middle line:84% And because users have specific problems situation 01:02:46.540 --> 01:02:51.260 align:middle line:84% and they need to solve it and no manufacturers is supporting 01:02:51.260 --> 01:02:55.180 align:middle line:84% that problem-solving, so they have to do that for themselves. 01:02:55.180 --> 01:02:56.480 align:middle line:90% And this happens a lot. 01:02:56.480 --> 01:03:00.380 align:middle line:84% And you can see that in open source and other things as well. 01:03:00.380 --> 01:03:02.920 align:middle line:84% So I think that's the key definition. 01:03:02.920 --> 01:03:05.750 align:middle line:84% And I think that affects a lot of conceptualization. 01:03:05.750 --> 01:03:07.000 align:middle line:90% IRVING WLADAWSKY-BERGER: Yeah. 01:03:07.000 --> 01:03:11.080 align:middle line:84% I think there's a lot more to talk about, but I want to make-- 01:03:11.080 --> 01:03:15.460 align:middle line:84% there's a point that I think is very, very important why-- 01:03:15.460 --> 01:03:19.740 align:middle line:84% I mean, we can ask ourselves, so why is this happening now? 01:03:19.740 --> 01:03:25.130 align:middle line:84% I mean, we can say, what's going on here? 01:03:25.130 --> 01:03:29.930 align:middle line:84% And I think it's a combination of factors. 01:03:29.930 --> 01:03:36.750 align:middle line:84% First of all, the, quote, unquote, "industrial age" 01:03:36.750 --> 01:03:40.830 align:middle line:84% in the Industrial economy where it's primarily about physical 01:03:40.830 --> 01:03:47.830 align:middle line:84% things, it's just harder for the users to be involved with 01:03:47.830 --> 01:03:50.790 align:middle line:90% physical things. 01:03:50.790 --> 01:03:53.990 align:middle line:90% You can hot-rod cars. 01:03:53.990 --> 01:03:55.650 align:middle line:90% Hot rod cars are example. 01:03:55.650 --> 01:03:56.830 align:middle line:90% You take the car. 01:03:56.830 --> 01:03:58.530 align:middle line:90% You do different things. 01:03:58.530 --> 01:04:00.110 align:middle line:90% So it's not impossible. 01:04:00.110 --> 01:04:02.990 align:middle line:84% It's just much harder with physical things 01:04:02.990 --> 01:04:05.630 align:middle line:90% in an industrial economy. 01:04:05.630 --> 01:04:13.110 align:middle line:84% In a knowledge-based economy, where so much 01:04:13.110 --> 01:04:18.710 align:middle line:84% of the way you embody the function can be captured 01:04:18.710 --> 01:04:21.650 align:middle line:84% in software, which you would all agree is happening, 01:04:21.650 --> 01:04:25.410 align:middle line:84% and not just for classic software, 01:04:25.410 --> 01:04:27.750 align:middle line:84% but there's probably no discipline where that's not 01:04:27.750 --> 01:04:33.310 align:middle line:84% happening, the combination of it being software 01:04:33.310 --> 01:04:36.710 align:middle line:84% and then having the internet and collaborative 01:04:36.710 --> 01:04:41.550 align:middle line:84% technologies and the ability to distribute the software, 01:04:41.550 --> 01:04:46.790 align:middle line:84% makes it so much easier to do this distributed 01:04:46.790 --> 01:04:51.310 align:middle line:84% collaboration that would have been impossible before. 01:04:51.310 --> 01:04:59.630 align:middle line:84% Do you all agree that that's a huge change that is industrial 01:04:59.630 --> 01:05:01.530 align:middle line:90% versus knowledge economy? 01:05:01.530 --> 01:05:04.550 align:middle line:90% 01:05:04.550 --> 01:05:10.830 align:middle line:84% And then even within, let me say, IT industry-- 01:05:10.830 --> 01:05:12.390 align:middle line:84% well, it's not just the IT industry, 01:05:12.390 --> 01:05:35.050 align:middle line:84% but in classic engineering, you tend to apply technology. 01:05:35.050 --> 01:05:40.370 align:middle line:84% You do automation with machines, pretty much, 01:05:40.370 --> 01:05:43.950 align:middle line:90% or even on back-office things. 01:05:43.950 --> 01:05:51.990 align:middle line:84% And here are the users, and you are changing things back here. 01:05:51.990 --> 01:05:54.670 align:middle line:84% If you're a bank, you're not going 01:05:54.670 --> 01:05:58.110 align:middle line:84% to have the users of the bank help you do the transaction 01:05:58.110 --> 01:06:00.050 align:middle line:90% processing system for the bank. 01:06:00.050 --> 01:06:02.070 align:middle line:90% That would be really bizarre. 01:06:02.070 --> 01:06:04.790 align:middle line:84% So the transaction processing system 01:06:04.790 --> 01:06:08.910 align:middle line:84% is a very good example of a back-office system that supports 01:06:08.910 --> 01:06:10.490 align:middle line:90% the operations of the bank. 01:06:10.490 --> 01:06:14.070 align:middle line:84% And obviously, that's where a lot of the initial innovations 01:06:14.070 --> 01:06:16.110 align:middle line:90% happen in IT. 01:06:16.110 --> 01:06:20.640 align:middle line:84% Or medical record management-- you're not 01:06:20.640 --> 01:06:25.800 align:middle line:84% going to have your users work on the security of medical record 01:06:25.800 --> 01:06:28.040 align:middle line:90% systems. 01:06:28.040 --> 01:06:31.640 align:middle line:90% That's not what you would do. 01:06:31.640 --> 01:06:39.280 align:middle line:84% But as we talked before, we are increasingly 01:06:39.280 --> 01:06:42.600 align:middle line:90% innovating of the stack. 01:06:42.600 --> 01:06:48.080 align:middle line:84% We are getting to the layers of problems 01:06:48.080 --> 01:06:54.800 align:middle line:84% that are services that are not industrial but are services 01:06:54.800 --> 01:06:57.440 align:middle line:90% in nature. 01:06:57.440 --> 01:07:03.920 align:middle line:84% Services, by definition, mean a person is doing 01:07:03.920 --> 01:07:05.740 align:middle line:90% something for another person. 01:07:05.740 --> 01:07:07.920 align:middle line:84% I mean, that's what a service is. 01:07:07.920 --> 01:07:10.800 align:middle line:84% And those may be employees helping each other, 01:07:10.800 --> 01:07:15.000 align:middle line:84% and maybe a doctor treating a patient. 01:07:15.000 --> 01:07:19.680 align:middle line:84% It may be somebody selling something to somebody else 01:07:19.680 --> 01:07:22.200 align:middle line:90% or giving them information. 01:07:22.200 --> 01:07:26.840 align:middle line:84% So the services portion of the economy 01:07:26.840 --> 01:07:31.640 align:middle line:84% involves people dealing with each other. 01:07:31.640 --> 01:07:35.080 align:middle line:84% The part one hasn't focused much in engineering 01:07:35.080 --> 01:07:39.000 align:middle line:84% because we haven't had any goddamn tools to do anything 01:07:39.000 --> 01:07:39.760 align:middle line:90% with. 01:07:39.760 --> 01:07:41.540 align:middle line:84% Say, well, we'll automate this service. 01:07:41.540 --> 01:07:43.720 align:middle line:84% Well, I don't want you to automate this service. 01:07:43.720 --> 01:07:48.620 align:middle line:84% If I'm the patient, I don't want to interact with an AI doctor. 01:07:48.620 --> 01:07:50.520 align:middle line:90% I want a doctor. 01:07:50.520 --> 01:07:52.620 align:middle line:84% So he said, well, then screw you. 01:07:52.620 --> 01:07:55.440 align:middle line:90% Well, then I'll go away. 01:07:55.440 --> 01:07:59.440 align:middle line:84% But because of the new technologies 01:07:59.440 --> 01:08:03.880 align:middle line:84% we have now, enter information analysis supercomputing, 01:08:03.880 --> 01:08:11.240 align:middle line:84% we can start bringing a whole plethora of new tools 01:08:11.240 --> 01:08:17.560 align:middle line:84% to enable the people performing the service and the people 01:08:17.560 --> 01:08:22.240 align:middle line:84% receiving this service to better work with each other. 01:08:22.240 --> 01:08:25.520 align:middle line:84% You're not automating the doctor. 01:08:25.520 --> 01:08:29.640 align:middle line:84% You're providing tools for the doctor 01:08:29.640 --> 01:08:32.279 align:middle line:90% to better serve the patient. 01:08:32.279 --> 01:08:37.479 align:middle line:84% Like, maybe tools that let them know the genomics makeup 01:08:37.479 --> 01:08:38.500 align:middle line:90% of the patient. 01:08:38.500 --> 01:08:41.479 align:middle line:84% So you can do prescriptions much better. 01:08:41.479 --> 01:08:51.720 align:middle line:84% Or maybe also information about what treatments have worked 01:08:51.720 --> 01:08:55.880 align:middle line:90% for a community of patients. 01:08:55.880 --> 01:08:59.240 align:middle line:84% And maybe in that case, the innovation 01:08:59.240 --> 01:09:04.920 align:middle line:84% may happen between the doctor and a community of patients that 01:09:04.920 --> 01:09:10.540 align:middle line:84% had the disease and were treated who are providing information, 01:09:10.540 --> 01:09:13.279 align:middle line:84% how well they felt, how well they didn't feel. 01:09:13.279 --> 01:09:15.500 align:middle line:90% So you get this user innovation. 01:09:15.500 --> 01:09:19.710 align:middle line:84% So everything that Eric is talking about, 01:09:19.710 --> 01:09:24.370 align:middle line:84% the reason it is so profound is because I 01:09:24.370 --> 01:09:27.350 align:middle line:84% don't know which is the chicken and which is the egg. 01:09:27.350 --> 01:09:32.130 align:middle line:84% That is, I don't know if because of tools like this 01:09:32.130 --> 01:09:34.770 align:middle line:84% and techniques like this, that's what's 01:09:34.770 --> 01:09:38.450 align:middle line:84% pushing us into the knowledge economy 01:09:38.450 --> 01:09:40.609 align:middle line:84% and allows us to bring technology 01:09:40.609 --> 01:09:42.790 align:middle line:90% and engineering to services. 01:09:42.790 --> 01:09:47.490 align:middle line:84% Or because we're in a knowledge economy and in services, 01:09:47.490 --> 01:09:49.270 align:middle line:90% we can do stuff like that. 01:09:49.270 --> 01:09:51.569 align:middle line:84% But there is something that feels 01:09:51.569 --> 01:09:56.810 align:middle line:84% very different from classic users of technology 01:09:56.810 --> 01:09:57.750 align:middle line:90% and engineering. 01:09:57.750 --> 01:10:00.250 align:middle line:84% Do you all agree with that observation. 01:10:00.250 --> 01:10:01.330 align:middle line:90% Please. 01:10:01.330 --> 01:10:05.450 align:middle line:84% AUDIENCE: Well, one thing is user innovation may not always 01:10:05.450 --> 01:10:07.090 align:middle line:90% be what the manufacturer wants. 01:10:07.090 --> 01:10:09.890 align:middle line:84% Because, for example, if you take 01:10:09.890 --> 01:10:13.770 align:middle line:84% the example of Apple iPhone, unlocking the iPhone 01:10:13.770 --> 01:10:15.390 align:middle line:84% might be a user design innovation, 01:10:15.390 --> 01:10:17.710 align:middle line:84% but the manufacturer wouldn't want to do that. 01:10:17.710 --> 01:10:22.070 align:middle line:84% So I think user innovation is a good thing, 01:10:22.070 --> 01:10:25.180 align:middle line:84% but it might actually violate the-- 01:10:25.180 --> 01:10:26.930 align:middle line:84% IRVING WLADAWSKY-BERGER: No, I understand. 01:10:26.930 --> 01:10:28.350 align:middle line:90% I agree totally. 01:10:28.350 --> 01:10:33.850 align:middle line:84% And we don't know yet whether unlocking the phone 01:10:33.850 --> 01:10:37.610 align:middle line:84% is good for Apple or not, because you need a little time 01:10:37.610 --> 01:10:42.890 align:middle line:84% to know whether Apple refusing to-- 01:10:42.890 --> 01:10:47.250 align:middle line:90% ah, you came just in time. 01:10:47.250 --> 01:10:51.590 align:middle line:84% We'll ask Professor von Hippel this very good question 01:10:51.590 --> 01:10:52.430 align:middle line:90% you brought up. 01:10:52.430 --> 01:10:57.850 align:middle line:90% 01:10:57.850 --> 01:11:00.370 align:middle line:84% Let me first introduce you to Eric von 01:11:00.370 --> 01:11:08.650 align:middle line:84% Hippel, who wrote the book Democratizing Innovation. 01:11:08.650 --> 01:11:12.210 align:middle line:84% And Eric, before you got here, I had made a comment 01:11:12.210 --> 01:11:22.450 align:middle line:84% that a critical part why user innovation is so important 01:11:22.450 --> 01:11:28.710 align:middle line:84% now is, in my opinion, two critical things are changing. 01:11:28.710 --> 01:11:34.330 align:middle line:84% One is we're moving-- please-- from an industrial economy 01:11:34.330 --> 01:11:38.330 align:middle line:84% to a more knowledge-based economy. 01:11:38.330 --> 01:11:42.690 align:middle line:84% And in a knowledge-based economy, a lot of the innovation 01:11:42.690 --> 01:11:47.490 align:middle line:84% is being captured in software, which makes it easier 01:11:47.490 --> 01:11:50.730 align:middle line:84% to do this innovation as opposed to physical things 01:11:50.730 --> 01:11:54.610 align:middle line:84% is harder, not impossible, but harder. 01:11:54.610 --> 01:12:02.370 align:middle line:84% And the second one is that in the past, 01:12:02.370 --> 01:12:04.730 align:middle line:84% we didn't have the technologies, the tools 01:12:04.730 --> 01:12:09.290 align:middle line:84% to innovate at the services layers of businesses 01:12:09.290 --> 01:12:10.930 align:middle line:90% and economies. 01:12:10.930 --> 01:12:13.820 align:middle line:84% And by definition, services involve people 01:12:13.820 --> 01:12:15.360 align:middle line:90% doing things for each other. 01:12:15.360 --> 01:12:20.140 align:middle line:84% We were innovating in the more back-office part. 01:12:20.140 --> 01:12:22.160 align:middle line:90% You can see why John is so good. 01:12:22.160 --> 01:12:23.700 align:middle line:90% Of course, he was out there. 01:12:23.700 --> 01:12:27.060 align:middle line:84% And we had VP of marketing, and he was out there. 01:12:27.060 --> 01:12:29.940 align:middle line:84% Also, it depends on the-- the thing 01:12:29.940 --> 01:12:35.780 align:middle line:84% we are most careful with is financial analysts. 01:12:35.780 --> 01:12:40.060 align:middle line:84% Because if you say the wrong thing to financial analysts, 01:12:40.060 --> 01:12:43.060 align:middle line:84% there are even legal implications to. 01:12:43.060 --> 01:12:45.920 align:middle line:90% So you have to be careful. 01:12:45.920 --> 01:12:50.660 align:middle line:84% IT analysts and press, the reason you have to be careful 01:12:50.660 --> 01:12:56.100 align:middle line:84% is because if you are not savvy about how your words could get 01:12:56.100 --> 01:12:59.220 align:middle line:84% interpreted, you may find reading something 01:12:59.220 --> 01:13:03.500 align:middle line:84% and [? Sadwick ?] said, I never said that. 01:13:03.500 --> 01:13:06.642 align:middle line:84% Well, but [? Sadwick, ?] maybe you 01:13:06.642 --> 01:13:08.100 align:middle line:84% said something that could have been 01:13:08.100 --> 01:13:10.540 align:middle line:90% interpreted as you said that. 01:13:10.540 --> 01:13:13.740 align:middle line:84% Also, once you do this, you become 01:13:13.740 --> 01:13:17.140 align:middle line:84% good at reading what the reporter is 01:13:17.140 --> 01:13:18.788 align:middle line:90% trying to take you to. 01:13:18.788 --> 01:13:20.580 align:middle line:84% And if you think they're trying to take you 01:13:20.580 --> 01:13:24.240 align:middle line:84% along a certain path, you start taking them along. 01:13:24.240 --> 01:13:26.520 align:middle line:90% You need experience. 01:13:26.520 --> 01:13:30.780 align:middle line:84% And I believe that experience is very, very important. 01:13:30.780 --> 01:13:35.300 align:middle line:84% I think in the first lecture we had in class, I think one of you 01:13:35.300 --> 01:13:38.380 align:middle line:84% asked me, well, isn't that the job of marketing people-- 01:13:38.380 --> 01:13:42.060 align:middle line:90% to communicate that? 01:13:42.060 --> 01:13:46.560 align:middle line:84% Remember, we're talking about disruptive innovations, 01:13:46.560 --> 01:13:49.920 align:middle line:84% complex technology, complex systems. 01:13:49.920 --> 01:13:53.100 align:middle line:90% 01:13:53.100 --> 01:13:58.420 align:middle line:84% The people that are often the best with press and analysts 01:13:58.420 --> 01:14:00.500 align:middle line:84% are the technical people who have 01:14:00.500 --> 01:14:05.100 align:middle line:84% a really good in-depth understanding 01:14:05.100 --> 01:14:07.320 align:middle line:90% and who can best represent that. 01:14:07.320 --> 01:14:10.440 align:middle line:84% That is, yes, you have these messages, 01:14:10.440 --> 01:14:14.700 align:middle line:84% but the really good technical people 01:14:14.700 --> 01:14:19.460 align:middle line:84% know where these messages came from, what's behind them. 01:14:19.460 --> 01:14:24.860 align:middle line:84% And they will be very, very good at reacting. 01:14:24.860 --> 01:14:30.420 align:middle line:84% And if you don't have that, you come 01:14:30.420 --> 01:14:32.140 align:middle line:90% across a little more shallow. 01:14:32.140 --> 01:14:34.500 align:middle line:90% I don't mean you're shallow. 01:14:34.500 --> 01:14:37.980 align:middle line:90% You just don't have the depth. 01:14:37.980 --> 01:14:45.380 align:middle line:84% And analysts and press, they can smell that. 01:14:45.380 --> 01:14:52.400 align:middle line:84% They can smell that, oh, so this is where I have to go. 01:14:52.400 --> 01:14:58.740 align:middle line:84% Whereas they know that no matter where they want to go, 01:14:58.740 --> 01:15:02.080 align:middle line:84% oh, this person is really well prepared. 01:15:02.080 --> 01:15:07.060 align:middle line:84% And this person is also market savvy and articulate, 01:15:07.060 --> 01:15:10.270 align:middle line:84% particularly because he or she has thought about it. 01:15:10.270 --> 01:15:12.650 align:middle line:90% It will just be easier. 01:15:12.650 --> 01:15:17.470 align:middle line:84% So it's a very big part of the job. 01:15:17.470 --> 01:15:22.590 align:middle line:84% And the alternative is that you do wonderful work, 01:15:22.590 --> 01:15:26.590 align:middle line:84% and then you're reading the paper that your competitor is 01:15:26.590 --> 01:15:27.930 align:middle line:90% getting all the glory. 01:15:27.930 --> 01:15:31.830 align:middle line:84% And you'll say, I've been out marketing. 01:15:31.830 --> 01:15:34.430 align:middle line:90% And that is humiliating. 01:15:34.430 --> 01:15:37.390 align:middle line:84% I mean, if somebody's going to outmarket you, 01:15:37.390 --> 01:15:40.270 align:middle line:90% let them have a better product. 01:15:40.270 --> 01:15:42.030 align:middle line:84% But there's nothing worse than you 01:15:42.030 --> 01:15:45.350 align:middle line:84% have a superb technology, a superb product, 01:15:45.350 --> 01:15:48.630 align:middle line:84% you didn't pay attention to these things. 01:15:48.630 --> 01:15:52.970 align:middle line:84% And you competitor walked away with all the glory. 01:15:52.970 --> 01:15:57.670 align:middle line:84% And you have a better technology and product, 01:15:57.670 --> 01:16:01.870 align:middle line:84% and you can write a book whining about it. 01:16:01.870 --> 01:16:05.310 align:middle line:84% But I cannot tell you, the marketplace could care less 01:16:05.310 --> 01:16:07.260 align:middle line:90% about you.