Craig Mundie: Microsoft Research Faculty Summit 2007

Remarks by:
Craig Mundie, Chief Research and Strategy Officer, Microsoft Corporation
Rick Rashid, Senior Vice President, Microsoft Research
Sailesh Chutani, Director, External Research, Microsoft Corporation
Daniel Reed, Director, Renaissance Computing Institute
Jeanette Wing, Assistant Director, Computer and Information Science and Engineering, National Science Foundation
Ed Lazowska, Professor of Computer Science and Engineering, University of Washington
Microsoft Research Faculty Summit 2007
Redmond, Washington
July 16, 2007

ANNOUNCER: Ladies and gentlemen, please welcome Harold Javid, Faculty Summit Chair, Microsoft Research. (Applause.)

HAROLD JAVID: Welcome, everyone. It’s so wonderful to see you all here today on another beautiful day in Seattle. Once again we’re defying the odds of this great area, and we’re so happy to see you. I hope you all had good trips here to Redmond, and we’re looking forward to a great summit. You know the theme of the summit this year is Optimism for the Future of Computing, and there are a lot of great reasons to feel that way. So, I’m looking forward to the program, as you are, and it should be very exciting.

What I would like to do now is to introduce Sailesh Chutani, who is the director of External Research and Programs, and we’ll begin today’s activities with him providing a brief overview of our activities.

SAILESH CHUTANI: Hello, good morning, and welcome to the eighth Summit. I hope you are recovering fine from yesterday’s party. We have another great party later in the evening, so save up some energy there.

It’s a real privilege and honor to host you again. It’s my fourth summit, and I’m really excited about this particular one, and I’ll tell you why. I’m pretty sure that you share the same excitement as I do about computing as a discipline. I think it’s a relatively unique discipline, because the pace of innovation continues to accelerate. It’s not slowing down. And not only is the discipline innovating at a very dramatic pace, it’s beginning to impact pretty much every other scientific endeavor you can imagine. And, in fact, it’s impacting every endeavor of our lives.

To sustain this innovation requires quite a few things. It requires investments. It requires partnerships. It requires partnerships between academia, between industry, and the government agencies, and it requires talent. In fact, to the extent that we are able to attract the best and the brightest to the discipline, that will determine the success of the discipline long-term. And it is to these ends that Microsoft Research created External Research and Programs to essentially work with you, to invest in very high-end innovative research, and to ensure that the ecosystem thrives and continues to attract the best and the brightest.

Of course, in the last 3.5 years since ER&P was created, we have collaborated with you in many, many areas of cutting edge research. These collaborations have taken the form of requests for proposals, projects, and joint institutes that we have set up with you across the world. We’ve also capitalized on many conversations about issues that are both important to you, and to us. So issues such as gender equity, the crisis in the pipeline in R&D funding, these are issues we have spoken to and lend our voice. And we continue to hold workshops and conferences to help build the research community, and Faculty Summit is an example of such a summit. And we have done our share through programs like the New Faculty Fellowship to ensure that the next generation of academic leaders have the resources to do cutting edge research, to take the risks that they want to take without worrying about funding.

Of course, in all of these programs we have been guided by our conversations with you. You have been very generous with your insights, with your time, and quite forthright with your feedback and critiques as well. So, for that, I would like to thank you.

Thanks to your collaboration, for your participation in our programs, at this point we have set up a network of 11 institutes across the world that explore areas that range from computational biology to collaboration techniques in the classroom, and basic computing. Our RFP program, at this point, I believe, is the largest such program in the IT industry. So, for all of these, I would like to thank you again, and it is exciting, all these programs have enabled us to work with thousands of graduate students and faculty across the world in some very exciting research.

One innovative thing we have tried recently is in Latin America. We set up a virtual research institute, and the goal of the institute is to really help address the problem of lack of critical mass that exists in each of the countries. There’s a lot of talent, but not enough of it in any given country. So, what we’re trying to do with the virtual institute is to connect the people in different countries together to provide them seed resources to work on problems that are pertinent to the region, and at the same time help them improve their visibility.

Now, this is a pretty unique public/private partnership, and not only are the government agencies involved, but we are also working with organizations such as the Organization of American States and the Inter-American Development Bank to get this off the ground. And if it succeeds, I think the impact on the economic development of the region will be quite significant, so this is quite exciting.

Of course, we are not stopping there, we will continue to work on very interesting things next year.

There are a couple I’d like to call out. I think all of you are aware that climate change is rising to the top of everyone’s agenda, it’s an important issue, and we think it’s important for the computing community also to look for ways to reduce both the energy and the material footprint of computing devices.

So, to that end we are announcing an RFP on sustainable computing, which will look for ways to encourage development of tools, benchmarks, software and hardware architectures, to essentially reduce the energy consumption while maintaining the performance and the functionality. And, of course, this is our starting point. You will see us more active in this space throughout the year and further on.

We continue to look at healthcare. I think it is still very important to improve access to healthcare. Currently a very large proportion of the population in the world, including in countries such as the United States, do not have access to healthcare.

And we think mobile devices, cell phone as a platform is a quite interesting way to approach the problem of healthcare. So you’ll see us announce an RFP to explore that space.

Of course, the quality of healthcare is important as well. And to that end we’ll be announcing an RFP to explore personalized medicine, and in particular we will be looking at genome-wide innovation studies, and work with the organizations in that space to do some cutting-edge research.

Of course, computing is still at the core of Microsoft, so we’ll continue to explore areas such as multi-core, human-robot interaction, and next generation of search, because I think that game is just starting out. There’s still a lot of very interesting problems to solve in doing domain specific search using semantic approaches, and such.

I’m really pleased to announce the creation of a Center for Collaborative Technologies at the University of Washington. This center will be led by Professor Richard Anderson, who is seated in the audience, and I can just call him out. There he is. I think most of you are probably familiar with him and his work on collaboration technologies. This really represents a culmination of five years of collaboration between us and the researchers who are looking at competency and other tools. This is a three-year commitment. It represents an investment of about US$750,000 from our side, but more than just the financial investment, it is about  it is the code base that the center will be working on has been developed in collaboration over the last five years, so we’re essentially turning that over to the community to evolve, and use it for all kinds of scenarios, be it scientific collaborations, be it in classroom collaboration, or for distance learning scenarios as well.

I’m also very pleased to announce a special award that we have created, and it’s named after Richard Newton. I think some of you who were present here last year saw Richard Newton chair the panel. He has been quite a towering figure in the computing community, and he has been a very good friend and advisor to Microsoft Research. So this is our little way to honor his memory and his legacy. And what we are doing here with these awards is to target interdisciplinary cutting edge research, the kind of thing that Richard felt very strongly about and passionate about. And by design, we have not picked the subject area, but we have defined the attributes of the kind of research we would like to support through this program.

There are a couple of other things that change as well. This is the eighth year of the Faculty Summit, so we have decided to revisit and rethink the format, and the structure of the summit. The next summit we’ll be organizing will now be two years from now, in the year 2009. However, we are creating a new event that will take place next year, and also that will be on a two-year cycle. And this event will focus more on Ph.D. students, graduate students, and early career faculty. So essentially we’ll have now two different distinct events in two-year cycles. The Faculty Summit as you know today, the next one will take place in 2009; but next year we will have an event focused on graduate students, and that will also be on a two-year cycle.

Finally, I’m really delighted to announce the winners of the latest new faculty fellowship awards. I think what’s interesting about this cohort and the cohorts that have preceded is that they represent a very broad range of interests, backgrounds, and universities. However, they all have something in common. I think they’re all very  not only are they very keen intellects, but they are united in the passion to change the world. And it is my hope, I would say it is my firm belief, that man of these young faculty members will go on to make breakthroughs, to win Touring Awards, and Nobel prizes. We would be very, very optimistic about these cohorts. And at this point, I would like to invite Rick to give the plaques to the winners of the new Faculty Fellowship who are here today. So I’m going to call out the names and the plaques, and if you could just step on the stage, that would be great.

The first award is to Magdalena Balazinska from the University of Washington. (Applause.)

The next is to Xixin Chen from Washington University in St. Louis. (Applause.)

Finally, Luis von Ahn, Carnegie Mellon University. (Applause.)

Thank you. And now, of course, we have a real treat for you. We have assembled a very impressive panel of luminaries who will discuss the future of computer science with you. The panel will be moderated by Ed Lazowska, who will be here in a moment, and I think all of you know Ed, he’s the Bill and Melinda Chair for Computer Science and Engineering at the University of Washington. And of course, I don’t think I should spend any time introducing him, because it will take most of the panel. Ed will, in turn, introduce the rest of the panel. Here are all the panel members, they should be here in a moment, because I think I took less time than foreseen.

Please welcome the panel. (Applause.)

ED LAZOWSKA: Thanks, it’s great to be here. I’m Ed Lazowska, and it’s a privilege to be able to moderate this panel today. Let me do a few preliminaries. First, I want to introduce the panelists. In order, Craig Mundie, who is the Chief Research and Strategy Officer for Microsoft. Rick Rashid, who is the Senior Vice President for Microsoft Research. Jeannette Wing, who until very recently was the Chair of Computer Science at Carnegie Mellon University, and on July 1, I guess, began as the new assistant director for Computer and Information Science and Engineering at the National Science Foundation. So assistant director of the person who runs the computer science part of NSF. And Dan Reed, who is the director of the Renaissance Computing Institute at the Research Triangle Universities, and also the Chair of the Computing Research Association Board.

So please welcome these folks, if you would. (Applause.)

Now let me tell you what we’re going to do. Our goal here is to have a discussion with each other and with you about the exciting future of computer science over the next 10 years or so. From my point of view, and I’ve just taken over the leadership of this badly named outfit called the Computing Community Consortium. The goal is to try and encourage the field to articulate a more audacious research vision, look a little further out and do a better job of presenting the excitement of the field, so we attract more students, and more exciting folks to the field, more research funding for the field, better interactions with our partners in industry, and our partners around the world.

When I look over the field of computer science it seems to inarguable that, first of all, we have absolutely changed the world in the past 10 or 20, or 50 years, just every aspect of the way we live, the way we work, the way we play, the way we learn is influenced by computing. Secondly, the research that all of us do has played an instrumental role in all of this. In 1994 or ’95, I was part of the Brook Sutherland Panel that did that tire track diagram of research innovations in computing. And the particular thing that we looked at was billion dollar industries that have come out of the research enterprise. And that’s, of course, only one narrow measure of impact, but still it’s a pretty impressive one.

And as we were doing that diagram in ’94, we were crying in our beer in the evening, because we couldn’t see which research that was going on at that time was going to lead to the next generation of billion dollar industries, that sort of impact. And eight years later in 2003 we redid the diagram, and there were six or eight new billion dollar industries in computing sub-sectors that we had totally unanticipated in 1994. We hadn’t assessed properly the impact of the Web. We didn’t understand the impact of wireless technology. We didn’t understand entertainment technology, whether videogames, or Pixar style animation. All sorts of things like this we just missed. So the innovations keep happening, we’re not very good at predicting which are going to be the big ones, but somehow we keep achieving them.

And finally, as I look towards the future, I see a dozen different areas in which we can have huge impact, whether it’s reducing the cost of automobile accidents, which are something like half a trillion dollars a year in the United States now. And our technology is going to change that. Transforming education with adaptive tutoring systems, impacting the third world through the design of appropriate technology and the deployment of that technology. There are dozens of areas in which we’re going to have enormous impact going forward.

So we want to talk about that sort of stuff with you now. The format is going to be, first of all, an interaction that I’ll have with the panelists, asking them a few questions, and getting their reactions. Then we’ll have lots of time for you to ask questions. There are folks with microphones in the audience. So when we get to that stage, just put up your hand, a microphone will head your direction. And if I can see the microphone through the lights shining in my eyes I’ll call on them, and we’ll get your question. The panelists will respond.

The way we want to begin, though, is asking you a set of questions on a response system we’ve got set up. So in front of you you should have this little Turning Point clicker deal. So pick this guy up, if you would, and you’ll see that there are keys numbered from one to nine, then in the center middle there’s one called zero. That’s not actually zero, that’s 10. So we’ll ask you questions that have responses numbered from sometimes one to four, sometimes 1 to 10. There’s a button on the lower left called Go, and a button on the lower right called Question Mark. If I were in your position I would immediately push those buttons. But, I have been told that that’s not the right thing to do.

So let me beg you not to do as I would do in your situation, and lay off those buttons on the lower left and upper right, and focus on the 10 buttons that are the number keys, so we can actually get some answers out of you.

Let’s pop up the first question, if we could. Hang on a sec, we’ll be there in a minute. Hopefully I don’t have to do something to make this happen. Technology marches forward, there we are.

Okay. The question is, will Harry Potter die? This is just to see if this thing is working. Let’s have some responses and we’ll see what happens. You have to click  if enough of you don’t click  

JEANNETTE WING: Do you have to hit go?

ED LAZOWSKA: No, you should be able to just hit one or two, I believe. Someone will tell us if that’s wrong. And once enough of us  we should have the Don’t Care button. Once enough of us have responded  actually there was supposed to be a line on this slide that said, who’s Harry Potter? There we are. All right. Boy, it’s overwhelmingly no. Okay. Great. Let’s  Rick will take us all to the movie this weekend and find out.

All right. So let’s move onto the first real question. Okay. So just demographics, what type of organization do you work in, university or college, government, or corporate? Most of you are university folks, I think, but we do have some number of government or corporate folks. So please answer one, two, or three, and we’ll use this to calibrate some of the later responses we get.

DAN REED: So don’t know isn’t an option here?

ED LAZOWSKA: No is not an option here, right, one, two, or three. There’s supposed to be a little light, mine stays on perfectly long. I’m sorry. Come on, answer, guys. There we are. That’s what we expected. Okay. Next question. Let’s move these along here. What’s your position, dean, department chair, professor, associate professor, assistant professor, researcher, not on the list above, I’m not with a university, so one through eight. Again, we’ll use this to sort some of the subsequent results.

Okay. Click that button, as soon as we get enough we’ll see a result. Okay. We’re an ancient group, as I expected, 28 percent over the hill. Next question.

Sorry. Many more over the hill, 50 percent over the hill.

Okay. So here’s one that I’m interested in responses to, are you glad you chose computer science in academia as your career? This is for the 92 percent of you in academia. So the answers are: Yes; I’m glad about computer science, but not sure about academia; I’m glad about academia, but not sure about computer science; or I’m not so sure about either of those things. So let’s try and have honest responses. There will be psychiatrists outside at the break to talk to everybody.

So that’s good, 60 percent are fundamentally content with the choice they made, 20 percent yes for computer science not so sure for academia, 13 percent yes for academia not so sure for computer science, 9 percent really should have a beer at lunch. All right. Next question. Which of the following best describes your principle technical specialty? So take a minute to read the list, we didn’t have room for an other. So pick one of these respond, and we’ll see where people are. And, of course, this is not the demographics of the field, it’s the demographics people Microsoft were good enough to invite to this event. But, it will show us who we are. Just do the best you can to stick yourself in one of these bins. If you truly don’t fit into one of these bins, then I guess just don’t respond. These are roughly the areas into which Microsoft Research divides its activities.

Okay. And there are maybe 300 of us here, so, for example, 4 percent in algorithms and theory means a dozen folks. And 20 percent in systems and in search retrieval and knowledge management, interesting, so that’s about 60 folks in each of those. And we are light in cyber security, as is the whole field. Okay.

CRAIG MUNDIE: And robotics.

ED LAZOWSKA: And robotics.

CRAIG MUNDIE: And machine learning.

ED LAZOWSKA: Thank you, Craig. Okay. Next question. So here’s a harder one, which sub-field of computing research do you think will have the greatest impact in the next decade, and impact measured however you like. So one version of this could be, where do you wish you were. That’s, in fact, how I had originally phrased the question. But, rather, which field do you think will have the greatest impact over the next decade? It should have been there. Computational biology should have been there, but, again, we only have 10 keys. So something to bring up later is, what fields do you think should have been on here. Try and pick one of these, and we can have that discussion later.

Okay. Systems and mobility is high, search retrieval and knowledge management, social computing is huge, machine learning is huge. So this  I’m surprised at graphics, vision, and animation actually, because it’s so cool. That’s very interesting.

Fair enough. Okay. So here are a couple of questions about outreach. To what extent is your academic unit involved in K to 12 outreach to broaden participation in computing, significantly, some, not a lot, none? Again, we maybe should have a don’t know, but we’ll come to almost that in the next slide. So which of your units are doing how much K to 12 outreach? Okay. So 40 percent essentially nothing.

All right. Next question. To what extent would you personally like to be involved in this, significantly, some, not much, or none? So this is just for you, if you had the opportunity would you be doing this, or is that for someone else to do? Okay. That’s pretty impressive, actually. Great. Okay. That’s very impressive. That says that about three-quarters of the people here, more than three-quarters of the people here would actually like to have some or significant involvement in K to 12 outreach to broaden participation. Fantastic. As a department chair, for those of you here, part of what that means is we’ve got to provide ways for faculty to hook into this, because they’re willing if vehicles are available. And I think the barriers to doing it are pretty high.

To what extent is your academic unit changing its intro curriculum, introductory curriculum to meet the needs of a broader range of students, significant extent, some, not much. So the intro curriculum, is it changing in an effort to make it more appealing, and more useful to a broader range of students? Push those buttons, and the man behind the curtain will let us see the results. That’s impressive, very impressive, assuming people are telling the truth and actually responding. Okay. That’s great. We don’t have the data on how many people are actually responding. It could be 12 of you. All right. Next question.

The next question is, to what extent is your unit changing its majors curriculum to meet the needs of a broader range of students? Okay. That is, are you cookie cutter cranking out one flavor of computer scientist these days, or are you accommodating a broader rang of, say, career options. Obviously we’re all familiar with the Georgia Tech Threads and Roles initiatives, for example. So if your department doing something like that, significantly, some, not a lot, or none that I know of? Okay. So still 70 percent are doing something, but it looks like we’re doing more at the introductory level than we are at the majors level, and there’s perhaps some work to be done there. Interesting.

Next, we’re getting towards the end. In which fields are you personally most actively involved in interdisciplinary activity? And notice that there is an other at the bottom, and there’s also a none. All right. So are you personally engaged in interdisciplinary activity with some other field, and is it science, engineering, social sciences, humanities, art, law, other, or none? This is a question about you personally, and I guess your students. Obviously you have to pick the one that most resonates with you. I heard a yawn.

Okay. So that’s really interesting, because only 15 percent is none. So the fact is that 85 percent of the people in this room claim to have some interdisciplinary involvement. And I dare say that that wasn’t going to be the case five years ago, or something like that, so very impressive.

Next, 10 years from now what proportion of the students taking upper division undergraduate, or beginning grad courses in computer science will be pursuing a computer science degree? All right. So if you say less than 25 percent, you’re saying that three-quarters of the people in those classes will be doing something else, greater than 75 percent means three-quarters of the people will be pursuing a computer science degree, but 25 percent will be doing something else, but taking our courses. So what percent do you think will be pursuing a computer science degree? I’ll say from my unit now, honestly, the way our courses are set up, 95 percent of the people in our upper division undergraduate and early graduate courses are pursuing a computer science degree. That’s where we are today. So what’s it going to be 10 years from now?

Okay. So that’s interesting. All right. So that says that half of us believe that less than half of the people 10 years from now in our upper division courses will be our own majors. That’s really interesting, and it says something about what those courses have to convey, and what they have to assume in terms of background. Next.

Well, this is a little wishy-washy, but the question is, do you think we’ve got to do something to enable longer-term research visions for the field, or are you satisfied with the sense of vision of the field today? I probably got the negatives wrong here. Do you believe we need to do more to enable longer-range research in the field? What we should do is a whole different question, of course. Yes, maybe, or no?

So now, I think this is very interesting. These are  in many ways the responses are more forward looking than I would have guessed from us as a community. And on the other hand it’s clear that we all agree we’ve got a lot of work to do. So what we’d love is for everyone to leave this session and this Microsoft Faculty Summit committed to work with each other, and with our units, and with Microsoft and our other friends, to actually enable this new future for the field, which we all sense we need.

Now, we’ll have the opportunity to use these guys later on if you have a question you want asked we can probably get our friends behind the curtain here to gin up a question. So that’s a possibility for later. But, let me now talk to our panel a bit. And then I’ll open this for your questions.

So let me start with Craig, you’re the Chief Research and Strategy Officer, you travel the whole world, you talk to heads of state, you talk to heads of research, and I know from my discussions with you that they’re often asking you what they should be engaged in if they want to be competitive in research 10 and 15 years from now, and at the same time they’re telling you what directions they’re thinking. So I’m interested in what you hear from them, and what you tell them about the future of science, the future of engineering, the future of computer science, what our impact it going to be?

CRAIG MUNDIE: I think one of the things that I tell them at lot, this was reflected in one of the questions, is that we have to approach many of the problems in a more multidisciplinary way. I think that breaking down the barriers that have naturally grown up between the field of computer science, and sort of the rest of the sciences environment, and engineering environment, I think has to happen now. So many of the people that I talk to are actually not in places where computer science is particularly strong, particularly in the emerging economy countries, and they say, so what kind of institutions should we build. And the one I encourage them to build is one where you start with the idea that you have to assemble teams of people to solve problems. There’s no problem of interest that you can probably attack without computing. Yet, we haven’t really found a very effective way to bring them together. But, I think that is the principle problem.

ED LAZOWSKA: So I’m going to ask Rick and Jeanette about this in a minute, because I think the question for universities and for corporate research organizations like Microsoft, is how do you build an agile environment, where you can assemble multidisciplinary teams that respond to an opportunity or a challenge. And for universities there’s a problem, we’ve got this administrative structure, which is schools and colleges, and departments, and at one level you’ve got to have those administrative organizations just to make the management tree have a reasonable breadth, and on the other hand when those are disciplinary administrative boundaries you wind up isolating people from one another. So to me the university in the future that’s going to be successful is the one that really is agile. And I don’t think any of us are set up to do that.

CRAIG MUNDIE: Certainly some of the institutions, like MIT and others, are giving the institution challenges, like energy, saying, hey, this is a problem, society cares about these problems, let’s form up in some way to attack this problem. And I think that those type of macroscopic objectives to the organization may be a way that people can get some of this collaboration going at a higher level.

ED LAZOWSKA: Right. I think the hard part is that schools and colleges don’t come and go with very great  at least they don’t go with great frequencies. Departments don’t go with great frequency. And the truth is, labs don’t go with great frequency. I think if you look at Stanford or MIT, which are lab-based organizations, you know, the computer systems lab has been at Stanford forever, and the AI lab, and the systems lab has been at MIT forever, although they’ve become joined at the hip in the past few years, but the fundamental structure is there. So I think the question still is, within those structures how you create and respond to these opportunities.

So let’s talk about technology assisting scientific discovery, and engineering discovery. Obviously, again, you talk about this a lot, and think about this a lot. We’re all computer scientists, so we probably have the view that we’re all at the center of the discovery universe, but say a little more about it.

CRAIG MUNDIE: I think that one of the challenges that many people in the sciences have, when we talk to them, is that they have way too many of the people doing their computing work who aren’t really contemporary in computing. That you go to the big physics experiments, I met with some people running one of the big ones and I said  they said, look, we can’t do this any more without more and more computational capability, we have to have a lot more software involved.

I said, well, how many of the people who are writing your software are people who are really professionals in software, as opposed to people who are physicists, who basically just have become programmers. They said 75 percent of the people were just physicists, who have sort of worked towards solving the software problem. And it was clear that there just was not a lot of interaction between the people who were driving forward in the field of computing, and the people who were driving the projects forward in that field. And clearly there is some coupling, but not at the level that I think is going to be required to solve society’s biggest problems.

So some of the work that we’ve been doing with the MSR team, and Tony Hey, who has been working with me, and now is going to work with Rick and the research people, is to look for areas where our research assets can play some kind of role in changing the game in the way that people look for some of these big solutions. Whether it’s in biology, or medicine, or health, or education, I think that there are opportunities for breakthroughs there, but I think we really have to look aggressively for how to apply these techniques in those areas. And without that, I think we’re not going to get the desired results.

PARTICIPANT: I think the lesson, though, is that the coupling between computer science and the other disciplinary fields has to be really close, and that requires a really significant investment on our part. There’s a real startup overhead to doing useful, impactful work in those areas.

ED LAZOWSKA: And you can’t not know the discipline.

CRAIG MUNDIE: One of the things I was struck by, and we had a conversation about this was some of the advanced computer science topics with a group in India at one of the IITs. And as we were walking across the campus after this, the department had said, you know, your comments make me think that the people who should be able to take these risks of either collaboration or extension are the most senior faculty. They’re tenured, you know, that’s supposed to be what gives us the ability to do that. But most of them 

ED LAZOWSKA: That’s all it affords us, right.

CRAIG MUNDIE: We just don’t seem to be able to or want to go off and do that. And so we’re kind of stuck. We push aggressively forward, but then somehow when the really big ones are put in front of us, we sort of lack the courage to go out and walk into the space that we don’t understand. I thought that was a very interesting comment from a department head, tenured faculty member, and to the extent that that comment is true, then I think when we face these problems we really are going to have to come back to why we give people this status and say, look, you have to go and do them.

ED LAZOWSKA: Here’s my personal theory about this, which may be, of course, totally bogus, as are most of my theories, but it seems to me that the question is, who is going to take a risk in moving to a new area, a new field, blaze new territory? And it seems to me the people who are likely to do that are either the folks who are in the top couple percent of the field, because they’re visionary, and they’re used to being successful, and they believe they can do it. So it’s not surprising that (Dick Carp ?) moved into computational biology, you know, that’s the sort of thing you expect of him. And then the bottom third of the field is going to do it because they’ve got nothing to lose. They may not get much done, but, what the hell, it’s not going so great, I might as well try something else. And the vast majority of us who are between the 30th percentile, and the 97th percentile are not so confident we can make the transition, and we have a fair amount to lose because we’ve managed to claw our way to a modicum of success in what we’re doing. So, my belief is that that big middle band, which goes a long way towards the top of the field, is very reluctant to take a risk, because there’s something to lose, and it’s not assured they’re going to succeed. And that’s human nature.

CRAIG MUNDIE: I think it’s interesting to ask whether the infamous innovator’s dilemma really needs to be applied to some of the fields of computer science. If we know how businesses die because they fail to adapt, then the question is, do we need a level of adaptation here, risk taking, or some mechanism to institutionally force us into the unknown spaces. If you look at the list, and the question showed on the monitor a minute ago, where we need to go in some of these new spaces, some of these things should or will make a big difference, but they’re not the area where we’ve been investing in now. What’s going to institutionally force us to go put the significant part of the resource into those spaces. That’s an interesting question.

ED LAZOWSKA: So, let me talk to Jeanette for a second. You’ve just taken over at NSF now. I’m not quite sure how to phrase this question because I realize that  

JEANETTE M. WING: It doesn’t matter how you phrase the question, I’ll just answer what I want to answer anyway.

ED LAZOWSKA: At some level, you’ve got a bully pulpit, right. But something I learned when I became department chair is that I can’t really have any ideas of my own, because I’m now management. Okay, so someone else has to have the ideas, and then I can facilitate. So what are the directions for the field that you would like to be facilitating over the next five years, maybe that’s the way to put the question?

JEANETTE M. WING: Well, thanks, Ed. I really do have two parts to my response. And, first, I want to speak more generally about my own vision for the field of computing, and many of you in the audience will not be surprised to hear what I have to say. And, second, I would like to speak more specifically about an initiative that NSF will be starting, and I think that will be of great interest to the academics in the community, and in the audience here.

First, as far as my grand vision for the field, I really do believe that computational thinking will be a fundamental skill used by everyone in the world by the middle of the 21st Century. And by that I mean, computational thinking as a fundamental skill, just like reading, writing and arithmetic. So imagine every child learning and knowing how to think like a computer scientist. This is a pretty grand vision. And I think it’s not just as Craig was saying, it’s not just other scientists and engineers using our computational concepts, and principles, and methods, and models, but it’s every other discipline. And we saw that in the responses, the arts, the humanities, law, medicine, entertainment. And, moreover, I see it’s not just the undergraduates, and the graduate students who we work with, but I think it is all age groups. So we do have to focus at K through 12. In some sense, when we get them as undergraduates, it’s already too late. And so these minds should be thinking, how is it that we can be teaching computational thinking at the K through 12 level. And I like to think that we’re actually as a field, given the theme of optimism for computing, I actually like to think that we’re undergoing a paradigm shift, if you’ll let me use that tired phrase, where it is not just our metal tools, but our mental tools that’s going to transform society for tomorrow. And that’s the difference between just using our technology, the wires and the transistors, but our fundamental abstractions, our concepts, again, the computational thinking that is really going to change the way in which we operate in our daily lives. So that’s the first point I wanted to make.

ED LAZOWSKA: The one comment, which is, I think it’s important to realize what Jeanette has just emphasized, and that is that computational thinking can take place without any computing going on. And so on the one hand, biology uses computation, and on the other hand, biology has become an information science. And if you’re in biology, and you’re not thinking of it as an information science, then you’re collecting frogs in some swamp. You’re miles and miles off the forefront of the field. And that’s what computational thinking is.

JEANETTE M. WING: That’s right. And you actually do see that, of course, in biology, when we’re now viewing things like the cell cycle and poaching/folding as processes, dynamic processes that we might want to model in terms of state machines or process algebras. But we’re seeing every other science, in physics, in geosciences, my colleagues now at NSF, they have in the geosciences have a really, really grand vision. They’d like to model everything from the earth’s surface to the sun, and from the earth’s surface to the inner core. And it’s  of course, we’re nowhere near the ability to model all the systems that are interacting with each other, and they talk about API, and they talk about emergent behavior, just in the way we talk about it. And so they’re already talking in terms of computational thinking ideas. It’s really gratifying to me.

So now let me turn to the second part, which is this new initiative by NSF, National Science Foundation, it’s called Cyber-enabled Discovery and Innovation, CDI, $52 million for Fiscal Year 2008 will be invested by NSF in CDI. It’s across the entire agency, so not just CISE, but math and physical sciences and engineering, also for cyber infrastructure and so on will be playing a role. It’s anticipated that we will have continued funding for the next few years. CISE, of the $52 million in FY ’08, as an innate, CISE will be putting in $20 million. So that’s a significant amount for computer science.

What is CDI? In a nutshell, CDI is computational thinking for scientists and engineers, and it’s exactly what Craig was saying, getting teams of people, multidisciplinary teams of people working together. We’ve got a computer scientist working with a biologist, and they’re working in a partnership. It’s more like the computer sciences might have possibly an incomplete solution to a problem coming from biology, and together they create new computer science and new biology. And so CDI is meant to promote and support that kind of research. And it’s not just biology, it’s really any other science and engineering, and that’s what NSF is about.

ED LAZOWSKA: It’s also not just traditional computational science.

JEANETTE M. WING: Right. And let me actually now, if you’ll let me say a little more specifically about CDI, there are two major intellectual themes that really we’re trying to promote. One is the recognition that we’re drowning in data, data, data, data. In fact, we’re drowning in data, but what we really want is to extract the knowledge that this data represents. The scientists and engineers are collecting massive amounts of data. In fact, they can’t even store all the data they could collect. They’re throwing data away. The problem is, you have all this data and you actually want to analyze it and discover new science. That’s the knowledge. And so there’s a theme of really extracting knowledge from data. That’s one of the main themes of CDI.

The other is the fact that the systems that the scientists and engineers are building, either by manmade systems, or studying in nature, are very complex. Again, if you think of the earth, or you think of the universe as a big state machine, how do you model that? How do you model it, and then predict the behavior of the earth day using the kind of simulation models that scientists and engineers use. And so the second theme is understanding complexity, understanding system complexity, be they manmade systems, or systems found in nature. And there’s a nice feedback loop, because the more data you analyze and extract knowledge from, the finer your predictive models can be. And then you start generating more data, and so on, and so forth.

The final third dimension of that that cross-cuts the other two is the notion of using our technology in virtual organization, and so this is cyber infrastructure along with the people who are at the end of the use of this technology.

So now we can imagine collaboratories across the globe, scientists working together across cultural, institutional, national boundaries, through our technology, through the cyber infrastructure. And so that will also include, of course, interest in social networking, and things like Second Life, and so on. So those are the three major dimensions of CDI. I think it’s very exciting, it speaks directly to what Craig was saying about multidisciplinary teams of collaborators. So I wanted to makes sure that the audience knew that this was coming.

ED LAZOWSKA: Great. We will come back to this. The one thing I want to add is, CDI, while these programs change, morph in various ways as they move forward, the genesis of this in many ways was work that (Kristos Papadimitrious ?), Dick Carp and others did in the theory community. Kristos and Dick’s view of this was algorithms as a lens on the sciences. And they coupled with Mike Foster at NSF, and the result through many turns of the crank, and passages through the meat grinder became CDI, which is $52 million this year, but it’s projected to grow to $250 million annually over five years. So that’s an example of what folks, granted exceptional folks, in our field, the sort of ball they can get rolling by saying, hey, you know, you algorithms folks, it would be great to be coupling more to other disciplines, and thinking about the problems that we can solve that are going to have high impact by this sort of coupling.

JEANETTE M. WING: I want to pick up on what you said, Ed. I want people here in the audience to realize that one person, you, as an individual, can make a big difference. So it was the big act crowd, Dick Carp and Kristos, who just through their own efforts, along with NSF, Mike Foster in particular, were able to push along this idea that grew bigger than algorithms as a lens, but grew bigger into CDI. And so this is really a small group of people that made such a big difference. So each of you has that potential.

ED LAZOWSKA: Right. Dan.

So, Dan and I served together for a couple of years on the President’s Information Technology Advisory Committee. The President put that committee out of business a couple of years ago, and sent all of us off to the packing house, except for Dan, who was appointed to the President’s Council of Advisors on Science and Technology, where he’s been working for the past nine months or a year on a report that looks at the National Information Technology Research and Development Program. That report is about to become public, but I’m hoping you can tell us a little bit about what you’ve found looking at these programs. This is the cross-agency programs in computing research.

DAN REED: Right, so the NITRD program is actually is about 14 agencies across the government. It’s more than just DARPA and NSF that we normally think of. It’s about $3.1 billion a year. The last time the program was reviewed was in 1999. The report that (Kent Kennedy and Bill Joy ?) shared. Well, 1999 is dinosaurs roamed the earth in the rate that computing changes. So the high level notion that we took initially were to look at computing as an enabler and how it fits in a broad ecosystem for the competitive positions of the country and the world. And so we looked at the industry issues, we looked at the academic issues, we looked at the research portfolio, and we looked a whole set of issues related to education. So I’ll give you a little bit of sound bite. The report that will go to the press probably in about a week or 10 days, and should be out a little bit later this summer.

But here are some of the takeaways. We touched on some of them already. On the education front, a lot of sense from talking to a lot of you, and a lot of others, that our curricula need to evolve to be more multidisciplinary and engaging to deal with the issue, and our perception that there ain’t no jobs in computer science. And we know better than that. There’s a huge shortfall of qualified workers to fill the jobs. But we have to get the message out that it’s really about computing, and how it engages in a variety of disciplines to create innovation. And so that’s one thing that affects what we do as academics.

On the related issues to that, a big push to simplify these processes for our international graduate students, so that those who receive degrees from U.S. universities will be able to stay in the country, and get permanent residency, recommendations for entry scholarship for graduate students, so the curricula issue is both an undergraduate and a graduate issue.

On the research front, a strong sense from all kinds of folks that our research emphases are too short-term and incremental, that’s exactly the question that we were debating earlier. That we have to look longer-term, more audacious, and more multidisciplinary, and at larger scale. So there’s a whole ecosystem of projects from the single investigator and his or her graduate student to large multidisciplinary teams that attack complex problems.

We also looked at the set of issues and priorities, what areas have emerged as new areas for investment, and what came through very clearly with overriding emphasis, the single top priority that emerged from this session across a wide range of groups was, computing systems that interact with the physical world. You can read that at one level as cyber-physical systems, you can also read it as inter-networked embedded systems, all of the closed-loop, adaptive kind of systems that involve the pieces of computing. Software are a perpetual problem, was number two. How do we build better, more robust, efficient software systems, and do that more inexpensively. Networking, and as Jeanette said, large scale data management, those were the top four priorities that emerged, with systems that interact with the physical world at the top of the list.

I guess the last thing I would say is, related to this issue that Craig touched on about multidisciplinary engagement is, it was also very clear as we talked to people that we tend to adopt the blind man and the elephant approach in thinking about computing and its applications. So the notion, we do computing and biology, or we do computing and civil engineering, or we do computing and something else, if you think about the kind of complex problems that we face in this century, I believe, and I think the committee believes, and certainly the folks we talk to believe, that they require multidisciplinary teams, not just computing and one thing, but teams that span a whole variety of areas.

So if you think about how we deal with environmental issues, for example, or healthcare for an aging population, those are not just monitoring or data management issues, they’re social issues, they’re healthcare issues, they’re biology issues, they’re physical science issues. Bringing together people across those disciplines to work on the problems is part of what we’re encouraging the government to foster more support for. But, that means we need to put together the change to make that happen.

So think audaciously, and stimulate the government to try to support that is really the take away message.

ED LAZOWSKA: Great. Thanks.

So finally, Rick, based on what we’ve heard so far, one thing I’m particularly interested in learning from you, as someone who has grown Microsoft research from kind of one person in 1991 to, what, 800 people today.

RICK RASHID: Pretty close, yes.

ED LAZOWSKA: How do you organize Microsoft research, first of all, to achieve both fundamental research impact, and impact on the company’s success? And secondly, how do you maintain agility, how do you reform structures inside Microsoft research? You, of course, spent a lot of time in universities, as well, I believe you came out here in ’91, because the alterative was becoming department chair at CMU, probably.

RICK RASHID: It was dean.

ED LAZOWSKA: Dean, right.

RICK RASHID: I think it was a bad idea at the time.

ED LAZOWSKA: Right. So what do you think about sort of organizational agility and organization structures, and forming teams?

RICK RASHID: It’s interesting, I think the approach I’ve always taken is one which says that you’re  first off, you’re fundamentally hiring people, not areas, right. You’re not  you have to start with the idea that you’re open to hiring the best people, sometimes in areas that weren’t the ones you were necessarily thinking about at the time. You take advantage of opportunities when they occur.

My personal philosophy is to build critical mass groups. I don’t believe in just having the one smart person in each area. I think you have to have enough people with enough diversity of expertise and abilities to really be able to play off against each other, and be able to get some synergy.

At the same time I also believe in having many different areas represented in different kinds of expertise. So as we built Microsoft Research, and we’ve expanded to a pretty large number of different areas, historically many of them didn’t have anything to do with Microsoft. We started a 3D graphics group before anybody in the company could sell 3D. We had a computer vision group when it was arguably difficult to figure out why we would use one.

I think that’s a statement about the fact that those are exciting areas of the field, and 

ED LAZOWSKA: I remember personally I was very surprised when you hired decision theory folks.


ED LAZOWSKA: It just hadn’t occurred to me that this was a smart thing to do 10 years ago, or whenever you did it.

RICK RASHID: It’s almost 15 now.

ED LAZOWSKA: I think it’s 

RICK RASHID: We’ve had medical doctors, we’ve gone into areas, we now, for example, have a group that’s doing a lot of work in AIDS research, and malaria. If you go to our labs in Europe, for example, and Cambridge, they have a very active effort in computational biology, looking at how some of the underlying theory of programming languages, and program structure apply to biological systems, and vice versa, and taking advantage of that.

So there’s a lot of that that goes on, and I think part of it is we put very little structure on what people do. We hire great people, we try to get them in an environment where we don’t feel bad about having more than one person in one area. We don’t set boundaries on the individual labs we have around the world. So there’s no  there’s nothing that says that the lab in Cambridge can’t work on exactly the same things that the lab in Redmond does, just whatever they work on. That fosters collaboration, it fosters interaction, it allows people to explore areas that wouldn’t necessarily be exactly down the beaten path, and it also allows them to change easily.

So I think it’s  if you don’t put a lot of structure in place, you put an environment in place where there’s a lot of opportunity for synergy, for cross-group collaboration. You remove barriers to collaboration. I think one of the big barriers I see in many environments is, frankly, the whole proposal bundling process, where you’re basically  if you force a lot of different groups to have individual budgets, and do individual proposals, that almost creates barriers between them. And there’s tendency for somebody to say, well, if I help some other group, if I collaborate with them, then well maybe they’re more successful and then next year they have a bigger budget than I do. Right.

There’s a tendency for people to naturally think of it as a zero sum game. And my approach is, well, no, you do your work, and if the organization as a whole is more successful, then all boats rise, and that’s just the way it works. And people are encouraged to interact, they’re rewarded individually for helping others, and working with others, and that’s sort of part of the system, and part of what makes it work.

ED LAZOWSKA: Right, so the reward system angle seems really important actually. I often think that even in universities if you say that that impact is what you’re assessing, you still try and come up with metrics for assessing that impact, and those metrics probably encourage people to keep doing what they’re doing in the short term at least, which means they don’t change in the long term.

RICK RASHID: Right. Well, I think there’s a big danger of that. You know, it’s so easy — I mean, you mentioned this earlier when you were talking about it. If you become the expert in an area, your overhead for writing the next paper and getting it published in a great place is very low, and there’s a natural tendency to just do that, because it’s so easy.

I think back when I was a professor, I think my record was four hours to get a paper written, submitted, and then actually won a best paper award, right? (Laughter.) You know, it just got to be so easy. In some sense that tells you it’s time to move on. (Laughter.) If it’s that easy, right, someone else should be doing it. Your skills are being wasted, and you should move on.

And that’s one of the things we try to encourage our people to do, which is don’t ride your field down into the ground or your particular area down into the ground, or your idea. Keep moving on to other things.

And so we set up a lot of mechanisms for people, are asked to interact with each other. We do these things called mind swaps. We get people together from different areas, and also from the product groups, and we bring in people from academia, and we try to get that level of interaction going. Just individually the managers I think can try to encourage that type of interaction. And we reward it and we don’t disincent it in the reward process, and I think that’s an important part of it.

CRAIG MUNDIE: I just want to add one thing to that. I think there’s really a fundamental issue for this whole community in this question of how we fund proposals. When I think about this, I always keeping coming back and saying what is the equivalent of venture capital in this community. And the peer reviewed research proposal is not the equivalent of that. In fact, if you think of business as opposed to research, today you could say, well, you’ve got the big businesses, and they kind of drive things forward. You’ve got venture capital that will help new things emerge. And you recognize there’s a special breed of people both on the funding side and the receiving side that tend to take those risks. And at the other end you could say you’ve got the private equity people now that say, well, I have no idea about how that business got to be what it was, but if there’s some way I can change the rules and let it be something better, we’ll take some of the shackles off that the system has created.

I think it’s one of the reasons private equity is such a boom right now in the business environment, because we kind of over-constrained the problem for how do you solve this thing.

When I think of research, to me at a macro level it’s sort of sitting there in the middle like big businesses are. There’s no private equity equivalent that’s emerging to take and change the rule set for the people who are working in this space in a fundamental way.

And there’s really no equivalent of venture capital. You know a couple of years ago, a few years ago, my wife and I decided to make a gift to the Fred Hutch Cancer Research Center. And I knew Lee Hartwell well; he has just won the Nobel Prize. We decided we were just going to give Lee the money with unilateral signature authority. And there were two things that were very interesting that happened when we did that. For the first time ever, the management of the institution itself took their own money and used it to match our gift. And I said, why the hell would they do that? And they said, now you don’t understand. It’s so extraordinary that people give us the ability to have an unconstrained gift where a single person can decide the direction that we will go as opposed to a group, that that’s so rare we’re going to put our own money behind it. And I view that as sort of them making a venture investment around the idea that Hartwell himself had discretion. The first thing he did was he backed a bunch of projects that had been rejected by the peer review process. One of them, for example, was could you put people in suspended animation and bring them back for medical purposes. And people thought this was sort of ridiculous, and had been completely turned down. In three years they’ve done it, one guy.

It’s that kind of thing that I think is missing in our field and broadly in the research community now. We haven’t figured out when or how we’re going to give some people discretion. That’s the thing that makes business go, the Bill Gates of the world or the Nathans or the Ricks or others, and the thing I worry about the most when we sit here and talk about the long term requirement to drive change, to move into new areas is this anchoring effect of this sort of peer review process as the thing that produces the venture funding.

ED LAZOWSKA: If I could pick up on that, because I absolutely agree with that, and that’s one of the things that came out of the PCAST discussions was that in some ways the feedback loop that we have as proposal writers and that government has by managing the review process is inherently very conservative, and when funding is scarce, as we feel it has been the last few years, that drives us even more to conservatism and incrementalism. And there isn’t a good way to make high risk bets on ideas. And if you think about your retirement portfolio, to sort of take that financial analogy a little longer, if you put all your money in T-bills you’re not going to lose your shirt, but neither are you going to be able to retire to Tahiti either. You’ve got to have a balance.

And we spent a long time in PCAST trying to think of ways we can structurally change the funding mechanisms and the reward structures inside government to try to incent more of that creativity, because it is a hole right now.

DAN REED: I mean, the Macarthurs are an example. (Louis Spanon ?) is here. He’s one person who has an active one in computer science that I can think of; there may be others. They’re very rare.

RICK RASHID: Well, a lot of times people ask me, you know, how do you decide what to do research on, and I say, well, I don’t. I decide who we hire and who we fire, and at the end of the day that’s the measure of — I mean, that’s the way you manage the organization.

It’s wrong for me to pick topics. I mean, I want to encourage people to take risks, and they need to feel that they’re going to be supported in that. Obviously if someone just is a bad researcher, we’ll figure it out, and we’ll find another person to fill that position.

But I think the notion that you have some high level committee that says you do this or you do that, that’s always going to limit choice. I mean, the way I usually say it is you’re biasing the front-end of the innovation pipeline if you make too many choices like that. You want to harvest the back-end; you don’t want to bias the front-end.

CRAIG MUNDIE: But the luxury here is that it’s been only Rick and Bill and a tiny handful of people who do have entrepreneurial instincts, who make the funding decisions. It’s not an unknown committee of people or someone else who’s trying to assess whether that’s going to be valuable or important, and I think that’s one of the differences, of course, between business funded research and government or university research, but somehow these things, the rest of the world has got to start to look a bit more like Rick’s world I think.

RICK RASHID: Well, and I think if you go back historically I think there was a time at DARPA when it did look a little bit more like that, right, there was a lot more risk-taking, there was a little bit less committee work, a lot more risk-taking.

DAN REED: But there really was a lot more betting on great —

JEANETTE M. WING: That’s right.

RICK RASHID: On people.

CRAIG MUNDIE: The program managers.

RICK RASHID: Right, but that’s what I’m saying is you bet on people, you don’t bet on the proposal, right, because you bet on people and you assume that they’ll be successful, and over time if someone is not successful, you go somewhere else. But you have to make your bet on the people and their ideas.

JEANETTE M. WING: I’d like to say a couple of comments on this particular topic. One is, of course, in the new AD in CISE this is something before I joined NSF I recognized from the community is one of the problems with the incrementalism, the conservatism, that review panels and so on tend to make.

So, what I’ve charged the program directors in CISE is really to take that risk, bet on the good people and the good ideas, and to fund the kind of long term, impactful research that we would like to see for our field. Now, so that’s one message I’ve sent very clearly.

ED LAZOWSKA: So, how do they do this in the context of a panel review process?

JEANETTE M. WING: Well, they actually — you know, they are the ones that make the decision, the panel reviews, you know, make the recommendations, and so it really is the program directors who have to be willing to take the risk.

Of course, the other thing I want to remind you is in particular with NSF and its peer review process is that it’s the only government agency that does follow the very careful peer review process, and it is this process that gives NSF the freedom that it has unlike the other agencies from, say, the political pressure. So, because we have this merit review process, we are not told by the — we don’t have the political pressures to say we need to find this area or that area. We can go back to the peer review or merit review process and say it’s really the community.

So, I think it’s really up to the panel reviews, the reviewers, you who also have to be bolder when you’re doing your reviews, and look for the long term potentially audacious kind of research that we would like to at the field be funding.

CRAIG MUNDIE: So, we need a new bumper sticker, right, like “Just Say No”.

JEANETTE M. WING: Well, actually my first slide is say no to incrementalism, and I’d like to also say that it actually has to be a collective effort, it can’t just be the government funding agencies, it’s got to be the universities as well. When they’re doing evaluation for promotion and reappointment of cases, it’s not just count those publications, it’s count, if you will, the impact. That’s what you’re really trying to measure. Now, that’s very hard to measure, but that’s what we’re looking for. Or that’s what we’d like to think we’re looking for.

ED LAZOWSKA: There are a lot of things we haven’t touched on. I want to talk at some point about Microsoft’s different models of interaction with universities and things like that. But let me try and turn this over to you folks. We’ve got till 10:45, despite what this timer says here, so we’ve got time for questions from you for any of these folks; we’d love to hear it.

QUESTION: I would like to go back to one of the first comments that Craig made about who does innovative research. In some sense the archetype is you get tenure, then you do something funky and weird because you have tenure and you’re guaranteed a job, but that doesn’t seem to work.

I think partly — and then to what Jeanette said at the end, it’s not just the funding agencies, it’s the evaluators. That would be the deans and chairs in this audience.

There’s a seven-year gap between when you graduate and when they decide whether or not you get to keep this job that we all say we love. Given that the letters are going out after five and a half years in general, that doesn’t give you time to do innovative research, because if it’s really innovative, that means a lot of the times you’re going to fail. That should be a measure of success in some sense, but you can’t get letters saying, “Oh, he had a great idea and he failed.”

So, I think one of the things that is great about Carnegie Mellon, even though I complained about it in the last year of the process, was that the tenure process is 10 years. And that small change now, and it’s not such a small change to the 20 percent in the audience here who don’t have tenure, but gives you the freedom to do something that you didn’t do when you were a graduate student, and I think that in and of itself would really change who’s doing innovation in the field.

So, to the deans and chairs, I know this is never going to happen, but give people a little bit more time, and then they can actually think hard about what they want to do, and they can have a year or two when they don’t even have any publications and they can do something interesting.

So, if it’s not just the funding agencies, but it’s also the universities, then I would say that that’s something that people ought to think about.

CRAIG MUNDIE: One of the big differences, of course, in business is Bill Gates only — when I came in to start things in non-PC computing, you know, he said, look, I only ask one thing. He said don’t make the same mistake twice. Other than that, there’s sort of like no ground rules.

And I think the culture here has been one which we keep trying to keep, but it is harder as it gets bigger and bigger, to encourage people to take risks and not penalize them for not having 100 percent success rate. And I think unless we can get more people to have that orientation outside the business environment — so I agree with you — it’s going to be a real challenge. I mean, your own statement to me is a bit of a dichotomy in that the people who should have no compunction about taking the risk for some reason don’t seem to be willing to do it.

QUESTION: They’re over the hill.

CRAIG MUNDIE: Fine, but right now it’s interesting. I’m 58. I don’t actually feel like I’m over the hill, and my job at Microsoft is actually starting things.

So, I think there are — and one of the things I argue about at the company, and I guess I’ll say that perhaps this community needs to think about, even in academia, is I do believe that there are people who are genetically predisposed to be more oriented to take risks, that it is sort of a genetic thing almost, and that what you need to do is figure out who they are.

So, what I think is important and we try to do at Microsoft is try to figure out who are the people who like starting stuff. And I don’t care whether you’re 58 or 28, you can start it. People who start companies tend to be younger because in their macroscopic sense they haven’t got families yet or they haven’t had to start paying the kids’ college education, they take some of those risks at a practical level that they wouldn’t take when they got older.

QUESTION: I guess my point is that the risk-takers are penalized in academia because — you know, what they say about investment, someone starting a company, you have to have failed first before I’m really going to give you any money. You don’t have time to fail first. In other words, the risk-takers, the people who naturally want to take risks really don’t have permission to do that in some sense if they only have five years to take a risk, fail, take another risk, and succeed, and then get some letters saying that they should get tenure at your wonderful institution.

CRAIG MUNDIE: Yeah, no, I agree with you.

QUESTION: And so if you had 10 years, it’s true, some people who don’t naturally take risks are not going to be rewarded, but if what we’re really worried about is the amount of innovation in our field, we need to restructure it fundamentally to allow those people who naturally take risks to stay in academia and take those risks.

CRAIG MUNDIE: But the trick there is you have to have the free replay.

QUESTION: That’s why you need the 10 years. If you can’t succeed at least once during that 10 years, then you probably aren’t going to be successful.

ED LAZOWSKA: Right, you don’t have time for failure in seven years is I guess the point.

DAN REED: So, let me pick up on the restructuring and just pose something for you to think about. I mean, we tend to believe that the structure of the research university was brought down from the mountain on stone tablets. In fact, the way that the American research university works with federal funding and the like is a creature of post second World War II, and if you look at the history and evolution of universities, that’s not a particularly long time. There’s no reason to believe in 30 years it has to look like it looks now. It could change.

ED LAZOWSKA: Yeah, the question is how to reorg.

QUESTION: I want to combine the question or the answer that we just heard with a theme that all of you talked about, which is very exciting, and that is that all the easy problems have been asked and answered, and the ones we have to tackle now are the multi-disciplinary, complex ones going forward.

It’s not a question in my mind of time. Time is actually a big differentiator between industry and academia, is that folks that go into academia are trained from early on to be solo contributors. As a matter of fact, anybody and everybody in this room who’s going through masters, PhD and everything else, it’s always about specialized, super specialized, and you are the only expert in the world in your particular discipline, whereas in industry actually people are taught to work in teams. And the biggest thing that universities would have to do in order to be more like business is to begin to organize themselves so that experts in a particular discipline in the university — forget the rewards, forget everything else, I don’t want to get into those — are at least in a position and have learned how to leverage the expertise of other experts in their organization, not at the lowest common denominator of their common conversation, but at the leading edge of the arm of the expertise of their colleagues.

There is nothing — talk about time — in the experiences of a 26, 28, or 30-year old person who for the last 25 years of their life were told, work by yourself, it’s your PhD, it’s only what you do that matters, that suddenly as creatures of habit are going to throw all that away in order to become a multi-disciplinary team worker, particularly — and with all the respect I don’t think six years or seven years or 10 years makes any difference, particularly if the folks that are going to be looking at that dossier are going to say, oh, this person had 10 years to do what another person had six years to do, so therefore we’ll divide by six, multiply by 10, and therefore they should have 40 percent more than anybody else. And, oh, by the way, whose papers are these? Is it the first author, the sixth author or the 75th author on the article?

ED LAZOWSKA: Well, part of what you pointed out though is there can be an institutional or organizational climate that actually either encourages or discourages this. I do agree that your PhD tends to be a solo activity, and on the other hand the tone that’s set in your department, school, or college can make a difference in whether this happens or not.

JEANETTE M. WING: Absolutely.

ED LAZOWSKA: So, I think there’s sort of nothing worse than a bad department chair.

QUESTION: I have two parts to this question. One is for Jeanette to talk about the (CPAS ?) program, because I think that would be relevant and interesting to a lot of people.

JEANETTE M. WING: Absolutely.

QUESTION: I know it’s a little early to —

JEANETTE M. WING: Where are you? I can’t even see —

QUESTION: I’m back here.

Let me mention the second part of my question as well. I really appreciate it, and I think it’s crucial that we have all of these career awards and fellowship programs and so forth for young people, but we’ve already heard that there are some of us older folks who have a history of doing risk and have had success in doing innovative kinds of things. Canada and some other places have senior programs for career activities. Why not have Microsoft, NSF, and others do five-year awards only for senior faculty to do something high risk, collaborative with other fields, whatever kind of criteria we want to put in, so we have a chance for those of us who want to try another venture in different areas to do that?

JEANETTE M. WING: Okay, let me speak about the CPAS program, because it is a program that I’m actually inheriting from Peter Freeman, but it’s an excellent program. The point of CPAS is to revisit undergraduate education in computer science, look at the curriculum again. So, this speaks directly to one of the questions that you were responding to. We just announced the awards for the first CPAS go-around on Friday, and we’re very excited about a lot of the particular projects that we’re funding. They definitely speak to multidisciplinary and project based, really hands-on kind of computing, working in teams, and sort of showing relevance of computing to the real world. And we look forward to actually continuing this program.

So, it really again in some sense speaks to my own vision or mission of looking at how to inspire the best and brightest to go into computing, or even if they don’t go into computing, to be taking our courses and using computing in their own professions and their own careers afterwards. And so CPAS is one of those very important educational programs that CISE is investing in.

Another one related to that is a program called Broadening Participation in Computing, BPC, and that really is focusing on underrepresented minorities, women, Native Americans, the disabled and so on. So, that is again broadening participation at all levels for all kinds.

Your suggestion about the career award counterpart for the senior people is a good one. I don’t know whether we have such a thing. I can certainly look into that.

I do know we have a program called CISE SGER grants. I don’t know if the community is aware of those. But those are meant to be really true risk-taking. You propose a wild and crazy idea, and without a lot of heavy peer reviewing, the program director can actually grant the award. We cannot give a lot of a percentage of our money to that, because of the different peer review process, but it is meant to be specifically for the bold, wild and crazy ideas, and to just give people some freedom to be innovative. But I’ll look into the other suggestion. Thank you.

ED LAZOWSKA: So, I want to mention a Microsoft analogy of that is the Technical Computing Initiative, which Craig began with Tony Hey in charge, and it’s recently moved from Craig’s organization over to Rick’s, and TCI — tell me if I get this too wrong — is essentially long term, deep relationships with multidisciplinary teams, where you’re betting on an investigator, and there’s a coupling with someone in Microsoft Research. But I know that a lot of your biological initiatives are of that nature, so they’re interdisciplinary, it’s people using computing and working with computer scientists. It tends to be somewhat more senior folks, and it’s big bets that are going to yield deep relationships and hopefully big payoffs.

CRAIG MUNDIE: Yeah, that’s right. We decided that we can’t be a funding agency, but we wanted to be able to take a few big things that were important to the society on a global basis and see whether we could help. Rick mentioned the work that his group has done in AIDS vaccine research. That was sort of partly my inspiration. That one happened organically. But when I looked at the effect of that, the emotional effect inside the company of saying, hey, look, yeah, we’re doing good business every day, but this is something that’s important to the society, and we’re making a difference in that. So, we institutionalized that where we’ve very carefully chosen —

ED LAZOWSKA: And let me say, by the way, if I’ve go the right one, this is David Heckerman presumably and others from MSR up to your computing researchers working with Jim Mullins and others at the University of Washington who are again top tier epidemiologists, and it’s a real partnership that’s been tremendously successful.

CRAIG MUNDIE: So, we’ve identified a number of these things, and, in fact, the money that we’ve put behind those very pointed collaborations is large relative to sort of the RFP style things that we do. But we’ll only do six around the world in a year, so they’re very carefully chosen, mostly by what people tell us they think are society’s biggest challenges.

ED LAZOWSKA: By the way, one interesting twist on this program, I’m sure this isn’t uniform, but something that Tony did when he started this up was to realize that in order to get the interactions going that always happens through students or post docs. So, part of what TCI tends to do is to fund a post doc whose job is to serve as the core interdisciplinary link and also the core link between the Microsoft researchers and the academic institution researchers. And that business of funding a post doc who’s charged to be the link both between disciplines and between organizations has been really successful.

CRAIG MUNDIE: Yeah, that’s worked very well.

ED LAZOWSKA: It’s a great idea.

RICK RASHID: Right, and we’ve had a history of doing that, too, in other areas. The Worldwide Telescope work, originally all that work was started by Jim Gray, it’s been carried forward. Curtis Wong has been central to that most recently. We’ve got the Centre for Computational and Systems Biology in Trento that we’ve opened up about a year and a half ago; the Joint Research Center now with INRIA, which is focused on a lot of interdisciplinary areas.

So, I think this whole notion of how can we sort of reach out cross-disciplines where I think one of the unique values that we can bring is often to be a catalyst to do some things that the funding agencies may not start doing on their own, although what we’ve found is inevitably every time we’ve made some investments, that’s been followed up many times over by money from traditional funding agencies, because they’ll say, well, if Microsoft is making an investment here, then maybe there’s something to being a catalyst.

ED LAZOWSKA: Yeah, my favorite example of that is the digital inclusion program where you’ve got 165 responses to this RFP, which meant there was an enormous pent-up demand of people who wanted to do computing for developing regions, and we just created this opportunity.

RICK RASHID: Yeah, and one of the measures when we do funding for research out around the world, one of our measures is actually to say how much money not are we investing, how much money are we driving, and it’s many, many times more than what we invest. And to us that makes a lot of sense because it says we can be a catalyst, we can make some things happen, because we can make decisions much more rapidly in many cases than governments or other kinds of funding agencies, and it works really well.

ED LAZOWSKA: Great. Let’s get some more questions. Door number one.

QUESTION: While I share the sense of many in the audience and the panel that we need to avoid focusing on short term incremental work, and be more long range in our thinking, I think there are also some pitfalls that worry me: the possibility of being seduced by hype as opposed to betting on good people. I think of DARPA BAs from a few years ago that said, “Don’t even bother talking to us unless you can promise three decimal orders of magnitude improvement in some important measure over a five-year period,” none of which anybody ever realistically hoped to achieve but they felt that they had to pretend in order to write the proposal. Or panel reviews that I’ve gotten back or seen peers get back from NSF in recent years that say, “Interesting ideas, fantastic team, incredible track record, clear they will do good work with this, but, you know, it just doesn’t sound audacious enough, unfunded.” So, I worry about some of the pitfalls.

CRAIG MUNDIE: Certainly one of the comments I would make is when we talk — Rick made the comment about DARPA. I mean, my whole view is that DARPA lost its first A, right. I mean, it’s just not doing advanced anymore. It has become a short term contracting agency for solving tactical problems in the Defense Department. The people who used to run many of these programs, they just weren’t charged with that 10, 20 years ago, when at least many of us knew them and were involved with them.

And so I do think that these organizations, the name stay the same, but their mandate is changed by environments in ways that have a profound effect on those longer term outcomes. So, I do think that if DARPA-like things aren’t going to go back to the model they had before, then we’re going to have to look elsewhere, whether it’s to the private sector in conjunction with the public sector and academia, or new programs that Jeanette will put in place. I don’t know what it’s going to be, but we’re clearly in a mode right now where we’re not hitting the right balance.

ED LAZOWSKA: So, here’s the one thing I’d say, and this is going to sound awful, but the fact is I’ve never volunteered to be a DARPA program manager, I’ve never volunteered to be an NSF program manager, I say no to lots of invitations to be on panels because I’m too busy, and then like Michael, I sit around whining about the fate that is dealt to me.

So, I’m sure that Jeanette would be happy to sign any of you up. She’ll be having a little booth outside later on, and if you want to come and change things, she’s got billets for me.

JEANETTE M. WING: That’s absolutely true. I mean, we’re looking for high quality people to make a difference.

First of all, let me just speak a couple of things to your point, which is well taken. One is to remind you that one of the things that I have directed the program directors at NSF inside is good people, good ideas. I don’t know if they understand or they will internalize that, but that’s my directive.

And the second is looking back to your tire track diagram, for instance, it would be interesting to look at what the proposals were, say, at NSF, and what they looked like. Maybe they were just quite mundane.

So, it’s hard, these program directors would have to make those decisions, have hard decisions to make, because it’s hard for them to know, based on a proposal that could be very mundane looking, that, in fact, that’s going to revolutionize the world in 20 years. So, give them a little break there.

But I want to pick up on what Ed said. I encourage all of you to volunteer when you’re asked to, to be on that panel review. It’s the high quality people we need in this peer review, this merit review process. And also you’re most welcome to come to NSF and join me in this crusade.

ED LAZOWSKA: Okay, number two in the back.

QUESTION: Hi. I’m a triple outlier here. I’m a sociologist, possibly the oldest one in this room, and still creative. (Laughter.)

ED LAZOWSKA: That makes you a real outlier.

QUESTION: And I am a Canadian.

But more importantly, I’ve done research exactly on the kind of cross-disciplinary work that people here have been thinking about. We just did a major study of the Canadian water network, which is what’s euphemistically called in Canada the National Centre of Excellence. It basically links scholars and corporate folks and some government folks, oh, about 200 percent.

One of the sad findings — and these are people from public administration, zoology, biology, toxicology, a number of fields interested in water quality. The saddest thing we found is that junior faculty, despite what is being said here, felt frozen out, that they saw it not being a career move to get involved in interdisciplinary work, and only the senior folks who are basically coasting and had very well built networks were prepared to take that risk partially to rejuvenate themselves.

Hence my conclusion is that a proposal to make the tenure process even longer will not help junior faculty to get involved in interdisciplinary —

ED LAZOWSKA: What do you think will help? Let’s have solutions rather than laments.

QUESTION: Oh, God. I can just give you a lament. I think what —

ED LAZOWSKA: No laments; solutions. No laments, no time for laments. If you’ve got a solution, I want to hear it. Otherwise, it’s number three. (Laughter.)

QUESTION: It would be nice if I had a chance to finish the sentence.

First off, I think it’s nice to talk from the basis of data and not from it. Secondly, we are proposing that they have segregated funding for junior faculty led projects, which goes along with the risk-taking styles here. Other people can come up with other solutions. Thank you.

ED LAZOWSKA: Great, thanks.

QUESTION: I don’t have any qualm about everything that I heard about the importance of risk-taking and pushing interdisciplinary research. These are things that I’ve done my best to encourage in my environment. However, I would like us not to forget that the main job of university is not to start $1 billion businesses but to educate, and to educate young researchers. I’m sure that Microsoft needs even more how education that it needs the ideas coming out of research and academia.

The question that we have to think when we think of new models, interdisciplinary research, large projects, adventurous projects that may fail and then you look for something else, to what extent there are measures, to what encourage good education of young PhDs? It’s very bad for a young PhD when you start research and after a year you decide, gee, that is not going to pan, and then you start another project, and after a year you decide, no, something else is coming. It may be very good for productivity or for risk-taking, but very bad for the guy that is doing a thesis. Large projects are not always the best environment for PhDs, doesn’t always give them the ability of choosing where they do the research and in which direction we work.

So, on the other hand, I think it’s extremely important to maintain the connection between education and research in an academic environment.

So, when we speak of how to produce the best possible research, we should temper it with the idea that research is only one of the components, one of our outputs but maybe not the most important one.

ED LAZOWSKA: That’s a great point. Clearly we’re number one in the people business.

QUESTION: I actually want to — I’m one of those genetic oriented risk-takers, so I also would be the first guinea pig for a person who is trained in electrical engineering and the computer science to get into biomedical engineering.

So, I actually want to bring three comments also relating to the solution as a person from bottom up. So, the first one is related to computational thinking. If you’re taking about computational thinking, also if you talk about this promotion of P and T evaluation, I think one of the issues that was mentioned earlier is we need to push computational thinking in every discipline of engineering and science.

So, that’s why if you look at this audience we have lots of deans and chairs, but how many of those are now CS engineers and science chairs and deans? If we want the multidisciplinary research, we want to bring all the disciplines together, they’re also playing an important role in actually promoting computational thinking.

So, that’s why my first solution, suggestion, next time we have our Faculty Summit maybe we should also invite some of the deans and chairs of non-CS departments. So, that’s one of my comments.

The second comment is related to also again talking about what are the solutions to support the risk-taking multidisciplinary research. I have to say that this is really a two-way street. There are people like myself who are taking risks, who go from computer science and electrical engineering to jump into biomedical engineering. I have to tell you the truth: It was not easy at all. I actually had three odds where I got into biomedical engineering department. First of all, I am the first one and the only one in the whole department who came from industry to join a traditional post-doc oriented department, which means that all my teamwork experience doesn’t work in this scenario.

And second of all, I’m the one who is as the female faculty joining a biomedical department who used to have lots of female faculty, however, a white lab based research. I’m the one who’s viewed as a computer guru; however, I’m wearing a round face, I’m not a man, I’m not a traditional computer hacker type.

And third of all, I am making a transition from electrical engineering and computer science into biomedical.

So, you can see through this scenario if I don’t have this genetic orientation of risk-taking, it would be very hard to survive. So, in this process, as I said, we do need some support, which I call some solution. For example, I personally believe Microsoft is a great place in supporting this, because while I was still young, struggling in the new departments, I got support from (Deng Zei ?) and later on by (Yeng Xi ?) from Microsoft University Relationship, really to support their trust on what they call the best excellence, put their trust on a good person.

So, I think this is only one-way street, but have we thought about the second, the other way street, which is was brought out by Rick earlier in the talk. You mentioned about David Heckerman. He is one good example of someone who is joining Microsoft Research, but how well established is such fellowship support, to support a non-CS and electric engineer oriented student to get trained in computing, so we can make computing a computational thinking?

Why am I mentioning that? I actually strongly believe we should start thinking about this, because last year when we thought Microsoft Research fellowships for PhD students for electrical engineer, CS, and other engineers, but this year it’s dropped, only computer science and electrical engineer. So, I think like as a solution I would say that’s another solution to actually think about it the other way around, to bring more David Heckerman like people into our training at the early age.

ED LAZOWSKA: Great, thanks.

QUESTION: So, I have some solutions to talk about. One thing that we have done is we are predominantly an information technology institute, so we have IT for X, so IT for civil engineering, IT for biology, IT for building sciences, IT in computational and natural sciences. So, that brings in labs which are closely connected with the core computer science so that you can do a lot of interdisciplinary work.

The second thing that I want to talk about is that current PhD students that you have, or anybody else has, are retraining them to write research papers or are we trying to train them to solve problems? And are we training them to think interdisciplinary ways? And if we change that, I think we will be able to make an impact.

ED LAZOWSKA: Great, thanks.

We are almost out of time. Let me ask the folks on the panel here if they’ve got one closing sentence or thought apiece, just going down the row here. Craig?

CRAIG MUNDIE: My last comment is don’t forget about computer science itself. So much of this discussion, appropriately, and so much focus I think recently is on the application of computing as we know it, and its refinement in all these multidisciplinary areas. My big concern is we don’t have enough people who are really just doing the hard computer science, thinking about what computing will become at a foundational level. I actually think we’re going to end up having to have more of a two track world. Maybe you’re going to have to have a two-track faculty summit; I don’t know. The people who want to go multidisciplinary help people solve big problems, or at least thinking how are those multidisciplinary people going to come the other way and help us redefine computing, because it’s going to require redefinition, too.

ED LAZOWSKA: Great. Rick?

RICK RASHID: Well, I think the way I said it earlier, I think we really need to bet on people. I mean, in some sense that’s the most important thing. And in that regard I’m betting on Harry Potter. I’m hoping he’s going to live.

ED LAZOWSKA: Jeanette?

JEANETTE M. WING: I just want to say again help me spread the word on computation thinking, and I really do think it’s an age where it’s not just our technology, but it’s our science that’s going to be the long-lasting contributions of our field.


DAN REED: We’re at an incredible crossroads. Computing has had such an impact on our world. The problems in computer science, computer engineering are enormously exciting. But we also have this opportunity to make a societal difference. And think about that PhD, the end of that is philosophy. And as we think about the origins of science and natural philosophy, remember it was about the intersection of knowledge in all its forms, and computing is the enabler for that discovery and its applications. Don’t go out of here and not remember that; that’s critical.

ED LAZOWSKA: Great. So, don’t forget the computer science, bet on people, computational thinking, change the world; it’s not a bad way to wrap up. Please join me in thanking the panelists again. (Applause.)


Related Posts