[This Transcript is Unedited]
Hubert H. Humphrey Building
Room 705A
200 Independence
Avenue, S.W.
Washington, D.C. 20201
Workgroup Members:
Staff:
DR. LUMPKIN: Good morning. My name is John Lumpkin and I am chairman of the committee, the Workgroup on the National Health Information Infrastructure, and I would like to welcome you all to our hearing on a couple of aspects of the National Health Information Infrastructure.
As you are all aware, the report of the committee has now taken on a life of its own. There are a lot of people who are interested in discussing the National Health Information Infrastructure and moving forward to looking at how we can expand that vision.
Today and tomorrow we are going to be talking about two aspects, one related to the personal health record, the second, which we will start off with, is talking about the issue related to the Internet, where it is and where it is going to be going and where it needs to go, despite the fact that over the weekend it wasn't going very well. But that is how we learn is by challenging the system and then responding to those challenges.
We are going to start off with some introductions and I think we will start off with Mary Jo. Before we get started with the introductions, let me just first say that we are going out live over the Internet as we do on this committee. Please, when you speak, identify yourself and also those of us at the table, please put the microphone close to you so it does, in fact, pick up your voice so people can hear you.
Mary Jo.
DR. DEERING: Mary Jo Deering in the Office of Disease Prevention and Health Promotion. I am the lead staff to the NHII Workgroup.
DR. SHORTLIFFE: I am Ted Shortliffe. I am a member of the committee and the subcommittee. I am from Columbia University, Department of Biomedical Informatics.
MR. HUNGATE: I am Bob Hungate. I am a member of the committee and the subcommittee and work under the label of Physician Patient Partnerships for Health.
MS. WILLIAMSON: Michelle Williamson, National Center for Health Statistics, CDC and staff to the NHII Workgroup.
DR. HAYWOOD: Trent Haywood, Region 5 Chief Medical Officer, Centers for Medicare and Medicaid Services and staff to the Workgroup on National Health Information Infrastructure.
DR. YASNOFF: Bill Yasnoff, Office of the Assistant Secretary for Planning and Evaluation at HHS and staff to the workgroup.
MR. VAN HOUWELING: I am Doug Van Houweling from Internet 2, also on the faculty at the University of Michigan.
MR. BUSH: I am Aubrey Bush. I am from the National Science Foundation. I am the director of the Advanced Networking Infrastructure and Research Division.
MR. HEPETER: -- Hepeter(?), medical officer at Centers for Medicare and Medicaid Services, also staff to a number of work groups.
DR. STEINDEL: Steve Steindel, senior advisor for data standards and vocabularies, Centers for Disease Control and Prevention, liaison to the National Committee and staff to the workgroup.
DR. ORTIZ: Good morning. Eduardo Ortiz. I work at the Agency for Healthcare Research and Quality and I am also staff to the working group.
DR. FITZMAURICE: Michael Fitzmaurice, senior science advisor for information technology to that Agency for Healthcare Research and Quality, lead staff on the Secretary's Council on Private Sector Initiatives to Improve Security, Safety, Quality of Health Care, liaison to the National Committee and staff to the Subcommittee on Standards and Security.
MR. BLAIR: Jeff Blair, vice president of the Medical Records Institute, member of the NCVHS and member of the NHII Workgroup.
DR. ZUBELDIA: Kepa Zubeldia with Claredi Corporation, member of the committee and subcommittee.
MR. STRAWN: I am George Strawn from the National Science Foundation, also co-chair of the Interagency Large Scale Networking Working Group.
MR. HITCHCOCK: I am Dan Hitchcock from the Department of Energy and I am also co-chair of the Large Scale Networking and Coordination Group.
MR. FIELDS: Larry Fields, senior executive advisor to the ASH(?).
MS. JACKSON: Debbie Jackson, National Center for Health Statistics, staff to the National Committee.
MR. ESLONGA: Don Eslonga(?), American Health Information Management Association.
MS. WADE: Terry Wade on detail from CDC to Advanced Objectives for Bill Yasnoff.
MS. HARVELL: Ginny Harvell, senior policy analyst in the Office of the Assistant Secretary for Planning and Evaluation.
DR. EDINGER: Stan Edinger, Agency for Healthcare Research and Quality and staff to the Quality Workgroup.
MS. HARRIS: Linda Harris, the National Cancer Institute.
DR. LUMPKIN: Thank you.
We are going to change the order, if that is okay, and start off with Doug.
MR. VAN HOUWELING: I am happy to be here this morning. It was a nice brisk walk across from over at 6th and Pennsylvania, where I was overnight.
I noticed your remark about the Internet over the weekend. We, of course, were following that rather closely. All of you who has some background in epidemiology, which I am sure a number of you do, would probably recognize that we saw an interesting phenomenon on the network over the weekend. That is that the intensity of the infection was -- the virus was so strong that it essentially killed itself off.
When it infected a host, it caused so much damage that the host was shortly unable to continue to replicate. So, it was an interesting kind of phenomenon. We, of course, track these kinds of things very carefully at Internet 2 and as I continue with my discussion, I am sure you will understand why.
Let me, first of all, say a few words about Internet 2 and why the effort that we call Internet 2 exists. The Internet as you all know has been around for quite a long while now and when the people did the original design work on the Internet 2 -- and I wish Vint were here
-- ah, Vint, right on schedule. Well, right on my schedule. In just mentioned your name.
As I was saying, when Vint and his colleagues did the original design work on the Internet. I don't think they thought that it was going to have terabytes of data, hundreds of millions of users and that it would be used for real time interaction in multimedia and it would turn out to be so critical to our infrastructure that security would be a dominant theme.
But we have also learned something, that they did such an excellent job of designing it that we have a capability here that we have discovered can scale, that it can accommodate explosive growth and because it can serve as a common bearer service for so many different media, it can enable the convergence of information work, mass media and human collaboration, all issues that are very important to research in the health care arena.
Internet 2 as an organization is focused on the Internet's potential for our future. We develop and deploy advance network technologies and apply them in research and higher education. Along the way, we work hard to accelerate the creation of tomorrow's Internet. As you all are aware, the Internet, as we know it today, went from a research and development activity all the way to the commercial Internet that we know today.
Internet 2 is yet another spiral on this effort. Working together with our colleagues in government and industry, what we are trying to do is move from research into the capability to privatize and commercialize the new capabilities of the Internet. For instance, we at the technical level support advanced services, multicast, which facilitates conferencing and large scale delivery of multimedia, IPV-6, a net set of protocols that support many of the challenges that the network is dealing with and, of course, measurements so we can understand what the network is doing and security activities.
As a non-profit membership organization, there are more than 200 universities across the United States, 60 corporations and 40 non-profits and government affiliates. In fact, 130 of our member universities have medical colleges, members of the AAMC, which provides us with a capability to put together strong leadership teams with health care and life science foci.
We have a number of government affiliate members. The NIH and the Food and Drug Administration are both affiliate members of Internet 2. People from time to time ask what is the relationship between Internet 2 on the one hand and the federal programs on the other hand. Well, the most important thing for all of to understand is they actually work very closely together.
The government programs are, of course, led by federal agencies and they are driven by the missions of those agencies, but they also develop research test beds and advance our general capabilities in Internet technology. Furthermore, at the National Science Foundation, the Federal Government has a long tradition of being a pioneer in supporting research and development efforts in Internet technologies in the universities through direct funding of university activities.
But, of course, our colleagues at NASA, DOE and NIH and so forth have also been involved in that. That funding goes to the universities that are members of Internet 2. Internet 2 is supported by its members with a small amount of federal grant funding for particular research projects. The university led Internet 2 is, of course, focused on the university objectives. At the base, though, we have common technical meetings in which we make sure that the Internet 2 activities interface and interact in a mutually beneficial way with the federal activities.
You will hear more, of course, about those activities as the day proceeds. Now, what does Internet 2 mean to this National Health Information Infrastructure? The first thing, of course, is that we are very much about a network that provides very high band width and low latency, which means that it is a network that can be used for real time control and interaction in situations where there is a large amount of information that needs to be transmitted.
It can support augmented virtual reality and enable human interaction through telepresence. So, we are now seeing the Internet 2 environment being used for a whole series of applications, ranging from teaching to actually doing research on remote manipulation of equipment and instruments.
Finally, we are very much focused on doing middle ware work that allows for there to be easy to use but secure access to information and computational resources. In that area, we are working with our colleagues in the number of efforts around the world known as the grid. When you think about the challenge we have in front of us, scaling all the way from the atom to the organism, the issues we confront with regard to database and modeling are extraordinary.
We often hear from the high energy physicists about the extraordinary amount of data that they are generating and developing with their large scale experimental apparatus. But frankly, when compared to the amount of data that we are talking about here, the potential in the health sciences and research area is much greater. Today, of course, we are just beginning on the journey to exploring that field of data, but it will become critical that we manage that and allow it to be shared across the world.
Now, our road map for much of our work has been the NRC report that, of course, now is getting to be old and so we are very happy to continue to interact with all of you. The consensus we are hearing from the health science community is that more new information will be graded over the next two years than throughout our entire history in digital form in the health sciences and that medical science is increasingly dependent on advanced computing solutions.
We believe that along with many of our colleagues in health science that instantaneous global communication of information and in support of human interaction is the next killer application of the Internet and that research and development in the health sciences will become more and more reliant on academic and industry partnerships.
Now, we have a number of activities that we are engaged in partnerships to support in Internet 2. Let me just run through these briefly. First of all, there is a Medical Middleware Working Group. This is working on the software that sits between the network and the many applications and supports the kind of security that are required by HIPAA and also by the various kinds of authorization and credentialing activities that have to take place in this arena.
We are exploring with our colleagues in the veterinary area the whole business of allowing students at a remote location to participate in grand rounds and we are beginning to see those same capabilities now explored in human medicine as well. We are very much involved in the visible human project laboratory, which I am sure you all know about. The Biomedical Informatics Research Network or BIRN, which is an NIH-sponsored effort to actually pull together the neural scientists in their research work. Just recently at our member meeting in Los Angeles, we showed an early instance of orthopedic surgeons actually providing on line real time education regarding orthopedic surgery over the network.
Also, we are working very closely to try to streamline the whole area of clinical trials through the Virtual Tumor Board. All of these activities are activities we provide infrastructure for but they are really led by the federal and university efforts in health science. We have been working very hard together with our colleagues on security and privacy guidelines. Here is a map of the BIRN sites. BIRN is anchored at San Diego and, as you can see, has national coverage.
The Virtual Tumor Board is an activity that is undertaken in Seattle where the doctors are exploring how they can more rapidly understand and get broad consultation on tumors and radiographs and micrographs. The fact is that we are engaged in a multidisciplinary activity that requires the collaboration in all of the areas that are outlined here on this slide.
In those multidisciplinary areas, we have a number of partnerships to exploit and develop. I have listed some here. I am not going to obviously go through and list them all here today, but the reason I gave you this relatively long list is because in each one of these areas there is specific work going on that is network facilitated through Internet 2 and we now understand how critical the infrastructure is to the future of the health sciences.
Well, we will have a chance to talk a little bit more later, but this also provides you with a source for information. If you want to know more about Internet 2, you can go to our Internet 2 web page on our activities in health care. You can also e-mail me or Mary Kratz, who is our Internet 2 health sciences program manager. Both of us would be happy to provide you with any further information you would like.
Thank you very much.
DR. LUMPKIN: Thank you.
Vint, welcome.
MR. CERF: Thank you. Sorry to be late. I was actually here on time except that they wouldn't let me park here. So, I wound up in some place like the Capital Gallery and walked about four blocks in the cold. So, I assumed that was your way of improving my health plan.
DR. SHORTLIFFE: I wanted to ask you to explain something to the group here because it is a point of some confusion until you hear it explained and then it is simple.
MR. VAN HOUWELING: Okay.
DR. SHORTLIFFE: An individual at an academic institution that is a member of Internet 2, how does that individual know if they are using Internet 2 rather than the commodity Internet at any given time. How do the local routers, et cetera, know? This confuses a lot of people because they get the feeling these are two separate networks and you have to choose which one to use.
MR. VAN HOUWELING: As a matter of fact, most people who use Internet 2 don't know they are using Internet 2. When they send e-mail, log onto a web site, do a video conference or something like that with a colleague at one or the other institutions that support Internet 2, it just automatically routes that traffic through the Internet 2 infrastructure.
But at the user level, the protocols are very similar and as a consequence, unless you have got a very high performance requirement, which won't work without Internet 2 infrastructure, you are not aware of which one you are using.
DR. LUMPKIN: We apologize -- since I am at a couple of days a year, I don't have to apologize.
MR. CERF: I don't think any apologies are needed, but I am sorry that I wasn't here promptly. It just happens. Are we ready to roll? Okay.
My job this morning is to give you some sense for where I think Internet is headed. You got some pretty good ideas from Doug Van Houweling's presentation about where at least one aspect of the Internet is going, particularly on the high band width side.
MR. VAN HOUWELING: But I counted on you setting the scene for that. So, I am glad you are going to do that now.
MR. CERF: So, let's do a little arithmetic, first of all. This is kind of a five year glimpse of Internet's growth from the middle of 1997 to about the middle or end of 2002. The 22 1/2 million servers that were on the 'Net in 1997 have grown to 162 million servers. That is not counting laptops and other, you know, PDAs and things of that sort.
The number of countries on the 'Net is increasing. In fact, I was pleased to see that the last country in Africa, I think it was Sierra Leone, finally got some access to Internet. So, every country in Africa has a little bit and that is a major milestone.
The number of users has gone up by a factor of 12 in that five year period, but to put it all in perspective, keep in mind that there is over 1.3 billion telephone terminations. So, Internet, in spite of its rapid growth, is still small compared to the scale of the telephone network. However, it continues to grow at a fairly rapid clip.
It is interesting to see where the users are. If I were giving this talk five years ago, the users would have been about 80 percent in North America. That is no longer true. In fact, North America represents less than a third of the users on the network. Europe is now the largest population, but it will soon be overtaken by Asia Pacific, particularly as the Chinese come on line as they are doing very rapidly.
Latin America has 33 million users. That was only 5 million about three years ago. Africa has continued to be a difficult place to install Internet services because the telecommunications infrastructure there is quite weak. In many cases, the only way to get there is by satellite. But it still continues to grow, particularly in South Africa and in the northern part of Africa and in Egypt, for example, and Tunisia. We hope, of course, to see more growth as time goes on, as the economies there improve.
So, that is the distribution. In terms of the growth in the total number of users, my estimates now are that there will be about 2.2 billion users on line by the end of the decade. You will notice that this is not an exponential curve anymore and the reason that I don't believe it will be exponential is that as the network continues to penetrate in places where the economies are weaker, the disposable incomes determine whether or not people can afford to be on the 'Net and can afford the equipment that is required.
Fortunately, the cost of the equipment is coming down. So, that helps as we go farther into places where the economies are weaker. But, nonetheless, it takes time to provide the infrastructure. So, perhaps I have a pessimistic estimate here, but 2.2 billion users is still a fairly good size number. It is more than a third of today's population.
MR. BLAIR: Excuse me. Would you be able to either speak more into the microphone or raise your voice.
MR. CERF: Oh, I am sorry. Why don't I just pull this microphone up here. How is that?
MR. BLAIR: That is great.
MR. CERF: Okay.
In terms of the number of devices that are on the 'Net, I had actually made some estimates in the chart and the yellow bars show that my estimate was about 900 million devices by 2006 would be on the 'Net. I hadn't counted on Internet-enabled mobile telephones and the side effect of that is to add another 1 1/2 billion devices to the 900 million I thought would be there so that by 2006, there will be about 2.4 billion devices on the 'Net. Now, that is in the presence of about a billion users.
So, right away, we have more devices on the 'Net than there are people and as we progress in this talk, you will see that I believe that that trend will continue and that there will be an increasingly large number of devices per person on the network as time goes on.
It is pretty clear, particularly once you heard from Doug Van Houweling and what you know perhaps of the industry, that the backbones of the Internet are increasing in speed. Today, they typically run at 10 gigabytes per second. Some of them are approaching 40 gigabytes per second, although there are some physics problems that make the 40 gigabyte speed not as attractive as the 10 gigabyte speed.
The simple story there is that you need repeaters at more frequent intervals to handle the higher speed transmissions and repeaters take up space and power. So, until you can avoid that, we typically wind up using 10 gigabyte per second per channel on a fiber. But there can be up to 160 different callers being transmitted on a single fiber, optical fiber.
So, even at 10 gigabytes a second, a single fiber can carry 1.6 terabytes per second of traffic in parallel and at 40 gigs, you are talk about 6.4 terabytes per second. So, a fiber cut can have quite a big effect on the traffic going across the system.
Having a high speed backbone is not really much help unless you have high speed access to the 'Net. You heard the term "gigapop" in Doug's presentation. There are interesting new high speed technologies for connecting to the 'Net, Gigabyte Ethernet being one of them. It has two flavors, 1 gigabyte and 10 gigabyte. It typically requires an optical fiber to carry traffic at that data rate.
In the longer term, I am anticipating that we will reach the point where it no longer is possible to look at every single packet that is flowing through the core of the network to decide what to do with it. There will be too many of them flowing too fast. You may have a nanosecond or two to examine each packet in the aggregate and that is not enough time.
So, the side effect of increasing band width is probably pushing us towards optical switching. All that means is that the packets at the edge of the 'Net will be groomed to flow onto a particular color of light and the optical switches will simply switch photons. At this point, you don't need to know what the packet boundaries are or anything. You just need to know what color is the light. Where do I send it now and that speeds things up.
So, in the core of the 'Net, we may not have to do packet switching anymore.
Finally, one of the most visible trends on the 'Net is wireless access. I am sure many of you have heard the term "Y Phi" or "802-11." It is a very popular 2.4 gigahertz and 5 gigahertz technology.
802.16 is a newer technique for sharing radio spectrum, using the code division multiple access method. Ultra wide band is another area of examination where very wide ranges of frequencies are used to transmit data at high speeds. Up until now the FCC has been a little reluctant to allow this kind of use because they have been careful to allocate frequencies to given functions.
Ultra wide band covers frequencies that are dedicated in the past to specific functions, but the idea now is to see if we can share the spectrum more efficiently. There are third generation and fourth generation cell phones that have data rates that are up in the megabytes per second in terms of carrying information.
So, those trends are all feeding Internet's growth. One of the most important commercial transformations that is happening is to use the Internet to carry voice or other audio. I am sure you have all heard streaming audio and maybe even seen streaming video --
DR. LUMPKIN: You are actually on streaming audio right now.
MR. CERF: I am on streaming audio now. I am streaming my head off, right?
What is interesting about the technology, the telephony on the 'Net, is that its economics are quite different from traditional circuit switched technology. SIP means session initiation protocol. It just happens to be the technique that is used for setting up and tearing down calls on the Internet.
It is actually an extremely general protocol and the reason it is so interesting is that it is the kind of protocol that you use to allow devices to negotiate with each other exactly what kind of service the one device is expecting the other to supply.
It is sufficiently general that it will almost certainly be used not simply for setting up and tearing down Internet telephone calls, but for controlling devices. I can easily see in the medical field a variety of systems that are interacting with each other using this very general and expansible protocol.
However, the point I wanted to make here is that the telephone world or telecommunications world is being stood on its head. The economics of Internet telephony lets you bill at fixed prices, rather than price per minute. That is a huge change in the industry. So, there is going to be a period of turmoil between now and surely the middle of the decade, if not the end of the decade when companies that supply telecommunications will have to learn a new paradigm for providing service and figuring out how to reduce their cost.
There is another technique, which has been developed by the Internet engineering task force called ENUM(?) and what it basically lets you do is to associate a telephone number with a domain name. The side effect of that is that you can -- if all you know is someone's telephone number, if that person is registered in the ENUM portion of the domain name system, it will be possible if that person wishes to find not only his telephone number, but also his e-mail address or his fax number or his cell phone number, whatever it is that you want, so that this way it will be possible to communicate with people, other than simply by calling them on the phone.
What is interesting is that you could cause -- you, yourself, could decide how you want your calls to be handled. If somebody discovers they want to call you on the phone, you could divert the call and have it sent to your pager. You could have it sent to a fax machine to leave a message for you, a written message. There are a variety of things that this kind of technique will allow you to accomplish.
More generally I am expecting to see very large numbers of Internet-enabled devices on the 'Net and you have already seen many of them, web TVs, palm pilots, mobile telephones. The video games now are becoming Internet-enabled so that kids can play these games in groups scattered across the network as if they were participating in the same room.
There are even Internet-enabled picture frames. They are very simple little devices with enough memory in them to keep maybe 20 or 30 images and they periodically dial up through the telephone network to the 'Net and go to a specific web site that is associated with -- each picture frame has its own specific web site or web page. It asks should I be downloading any other images.
I have one of these in my office. I have it at home. I have scattered them around my family around the United States. We sort of keep each other up to date about what is happening with the kids by taking digital pictures, uploading them to the appropriate web site and those are picked up and relayed to the picture frames. But you can imagine all kinds of other techniques that would allow you to put any kind of information.
So, now they put news. They put weather reports, you just pop it out onto the picture frame. What is interesting is the picture frame has only two buttons. It has one button to control contrast and the other button says please go look on the 'Net and see if there is anything new. If you don't push any buttons, it will periodically wake up and say I guess I should go check, maybe every 24 hours.
There are Internet-enabled refrigerators for sale now, too. They have liquid crystal displays and touch sensitive liquid crystal displays. Some of them have bar code scanners. So, the refrigerator now knows what it has inside of it. So, while you are off working, it is surfing the 'Net, looking for things it could make with what it knows it has inside. When you get home, you see recipes on the display.
The Japanese have developed an Internet-enabled bathroom scale so when you step on the scale, your weight gets recorded and it becomes part of your medical record. Of course, if the refrigerator gets the same information, you might come home and find diet recipes coming up on the display. I can imagine going shopping and having the refrigerator page me saying, you know, don't forget the marinara sauce. I have everything else I need for spaghetti dinner tonight.
The Japanese and the Dutch are Internet-enabling cars with the global positioning satellite receivers. The side effect of Internet-enabling devices is perhaps captured well in a simple scenario. You are driving down the street. You have your Internet-enabled mobile phone with you. The car has a global positioning satellite receiver and it has a wireless local area network.
There is a navigational display, which has its own Internet address. There is an on-board computer and you make a phone call through the ordinary voice network to a computer that also happens to be on the Internet. And it understands speech. This is perfectly normal because today, in fact, this is a common practice to have machines that understand speech.
So, you say where is the nearest Thai restaurant and the computer you are talking to, once it understands what you want, has found out where you are because your cell phone got that information from the global positioning satellite receiver and relayed that information as data along with the voice. So, the computer that understood the voice then goes out on the 'Net to a geographical index database server and says here is where my customer is. Where is the nearest Thai restaurant? Gets the information back and says in your ear it is two blocks up and to the right. But it sends a map to the navigational display because it knows what the Internet address of that is.
Then it says, well, do you want to see the menu or would you like to place an order or do you want to have a reservation. The thing I want you to take away from that scenario is not the specifics, but rather that a collection of devices with Internet addresses on them were capable of interacting with each other, being assembled briefly to perform a complex function for you and then go off to serve other people.
So, this notion of drawing computers together to do something and then go on to other things is a very, very powerful notion and I believe it will be increasingly common. At the bottom of the slide you see Internet-enabled wine corks and Internet-enabled socks. Since I am short on time, let me just mention the sock thing because it is relevant to the medical field.
I am sure you are aware that clothing is being developed that can sense vital signs, whether it is pulse rate or oxygen levels or, you know, breathing rates or things of that sort. So, I got to thinking what it would be like if you had Internet-enabled clothes in general. So, thought, well, let's see if I had Internet-enabled socks, the first thing I could do would be interrogate my sock drawer and I would get back a report saying that there are 17 matched pairs of socks in the drawer and one sock is missing. Sock No. 144L is not in the drawer.
So, you would send a multicast around the house and you would get back a report from the sock that is missing saying, hi, this is Sock 144L. I am underneath the sofa in the living room. So, I have solved the problem of the missing sock, which I think is a tremendous contribution.
Now, like all clever ideas, though, every engineer has to remember that there is a downside. So, I got thinking of a different scenario. Some guy is -- well, he is wearing his Internet-enabled and he calls his wife and says, hi, honey, I am going to be working late in the lab tonight. There is a brief pause and then she says, well, that is nice sweetheart. It is odd, though, because your shirt is down at 19th Street at the bar. So, maybe this Internet-enabled clothing is a dumb idea and we shouldn't pursue it.
Let me outline some major issues that I think we are facing in the Internet world. One of them is clearly reliability and availability. If you are going to apply this network to the kinds of things that you do or your colleagues do, it has to be reliable and it has to be available at all times. I am sure you are all aware of the major worm attack that took place over the weekend. It was a fairly intense attack. On the other hand, people were able to respond pretty quickly to it.
But I am still not satisfied that we have an infrastructure that is sufficiently reliable and available for you yet. Security, Doug mentioned, and I will, too, it is a huge task ahead of us to secure all of the Internet infrastructure, whether it is the routers or the domaining system or the servers and clients that are at the edge. There are many, many places where reliability can be breached -- or I am sorry -- where security can be breached.
I don't have to tell you how important personal privacy is in connection with the health care industry and once again technology and practices will be critical. Another big challenge that Doug mentioned, too, is the deployment of IP Version 6. We need it because the current address space at IP Version 4, which is what you are running now runs out at 4.3 billion addresses.
Since I think we will be at 2.4 billion by 2006, I am concerned that we move to the larger address base, which handles 10 to the 38th addresses, which I think should be enough until after I am dead and then after that it is somebody else's problem.
Finally, just another example of the major issues getting people or things authenticated so we know who is transacting what with whom and plainly for access control to information that is important for remote control devices, for remote surgery if there ever is such a thing. One would clearly want to make sure that the n points that are understood and known to each other so that we don't have a 15 year old taking over the surgical theater. That would not be a good thing.
There are some huge policy problems ahead. The most obvious ones that I am sure you have all seen have to do with intellectual property protection, whether it is music or books or movies and the like. I think we have beginning to see the same transformation that occurred when the Xerox machine came along. People are struggling to understand how this digital world changes our understanding of intellectual property. Fair use is going to be redefined almost certainly as a consequence.
There are rabid fights over trademarks and domain names. Those two notions are in direct conflict with each other. The trademark is not unique. More than one party is allowed to register a trademark and use it, as long as they are not conflicting in business or don't create confusion. Unfortunately, domain names have the unique property that they have to be unique and only one can have a particular domain name, not two people.
So, those two things are in fundamental conflict with each other. There are huge issues associated with competition among the telecommunications carriers and Internet service providers. The FCC is in the throes of trying to reexamine the Telecommunications Act of 1996. It believes that it is moving down a path that will increase the likelihood of high speed access to the Internet. Some of us don't necessarily agree with their thinking about this.
What they are basically doing is recreating an unregulated monopoly by permitting the local exchange carriers to withhold access from any new infrastructure that they build so that any competing Internet service provider won't not have access to that facility, which unfortunately is counter to the original intent of the Telecommunications Act of 1996.
You can see evidence of concern over information on the network. Some countries, including our own, attempted to censor the content of the 'Net. It doesn't work very well. We are starting to see some very peculiar national efforts. In Panama, for example, my company was instructed to block the ability of the network in Panama to carry Internet voice because it turned out that the Internet were not generating any revenue for the country; whereas, the conventional international calls had a settlement system associated with them.
So, the Internet voice was actually essentially taking money away from the Panamanian Government and, oh, by the way, Cable Mileist(?), which is the party that was running the PTT there. So, we did as we were told, but anyone who understands how that stuff works can work their way around it. So, eventually this is not -- this notion of blocking and censoring things on the 'Net is not going to work very well.
Finally, just as another policy issue to put on the table, as Doug was talking about the collaborative activity in the medical profession, I wonder about the licensing problem. Am I practicing medicine out of state as I consult in these collaborative environments? I don't know how that is going to get resolved but I suppose that is going to be something you have to worry about.
Some of the challenges we face, scaling to billions of devices, making the system more resilient than it is now, making sure that the services that are on line in real time are safe and, finally, achieving cost savings in the health care transaction processing business, which I believe the government will enlarge proportion in our total health care bill.
I thought I would spend just a couple more minutes, if it is all right giving you a slightly longer view of what is going to happen to the 'Net. What I am about to tell is not a Worldcom activity. This is something I have been doing with the Jet Propulsion Laboratory for the last four years.
You know, that we have been exploring the solar system for a long time and Mars has been one of our major targets. You remember the little rover that went to Mars in 1997, the Pathfinder Mission. We now have a new mission, which is about to be launched in May and June of this year. The rovers are 150 kilograms each. They are going to be landing sometime in early 2004 on Mars.
But the reason I bring this up is that they will be Internet-enabled. We are going to be -- we have been developing an interplanetary extension of the Internet with the intention of supporting robotic exploration of the solar system in a more efficient way than we have in the past. So, this little rover is actually quite a bit more sophisticated than the one that we saw in 1997.
There is more here, but -- well, actually I can't resist showing you the reentry method. Here is the -- the rover is entering the Mars atmosphere. There is an ablative fuel and you can see it starting to burn. The parachute will deploy. There is a thin atmosphere in Mars, but it is possible to use it to slow down the landing craft.
The shell of the reentry vehicle blows away and inside of it is a device that blows up into a huge number of balloons that surround the device. Now, I don't recommend this for human landings, but the little rovers seemed to survive okay in 1997. So, we are going to use the same technique for this one.
So, what we are expecting to do -- we have now developed a new set of protocols that can work over interplanetary distances. TCPIP does not work when the round trip times are 40 minutes to 40 hours. So, you can imagine clicking your mouse and waiting for a rather long time for the first bit to come back on the web.
So, we have had to develop a new set of protocols that we now are on the fourth iteration of. We now have prototype software. We are starting to deploy a little bit of it, at the littlest link level, in the new rovers that are going. As the decade wears on, we hope to have more and more of the robotic systems fully enabled for interplanetary operations.
We are going to use standard Internet on the surface of the planets and on the spacecraft because that works just fine in the below delay environments. So, that is where we are. Of course, in the long term, we hope to build up an interplanetary backbone as new missions are launched, each of them carrying a little bit of the interplanetary protocols on board.
These web sites are sources of information that you might find of use. My search up web site has lots of pointers off to interesting places. If you are interested in following the interplanetary work, the next to the last web site is www.interplanetarynetworkspecialinterestgroup or IPNSIG.ORG, which is a special group of the Internet Society.
I certainly invite you to have a look at those sites and I am looking forward to the rest of the day's discussions.
DR. LUMPKIN: Thank you. Thank you very much.
Aubrey.
MR. BUSH: I am going to violate one of the basic tenets of the Internet, which is never follow Vint Cerf on a program. I can't resist one anecdote. Vint was talking about the GPS capability in the automobiles. There was an article in a business newspaper in Atlanta a week or so ago. A fellow there has just given his daughter her -- a teenager -- a new BMW GPS-equipped, wirelessly connected so the car signals dad when she leaves home and signals him again when she gets to school, can potentially if he queries it, tell him where the car is anyway at any time and also signals him, I think, if she ever exceeds 80 miles an hour and that kind of thing.
She seems very pleased to have the new BMW, though. That is, I guess, sort of extreme. From having raised a couple of daughters myself, a better strategy may be simply to not ask any questions you don't want to know the answers to.
Let me tell you some of the activities that the National Science Foundation is undertaking now. In the Networking Division at the Foundation, it historically goes back to the eighties when ARPANET, which Vint was involved in, sort of walked into NSFNET and began to spread into a much broader group of users, actually attract enough commercial attention that the commercial Internet basically took over in the early nineties and the NSFNET was phased out around 1995. That was followed by high performance network connecting supercomputer centers and some other sites that, again, became a much wider spread high performance network, which Vint in fact is PI on.
The program is in a three year no cost extension phase, which ends in April and for the last several years Vint has been the PI on that. The community has through the Internet 2 consortium, which Doug Van Houweling was telling you about earlier, has created an a network, the Abilene Network, which I will say just a bit about further downstream. So, the Networking Division has been involved basically in the expansion of networking activities from a research curiosity into what has become a necessity in life, I think. They have comments about how widespread it is and what the utility of it is.
One place where I use the Internet voice, for example, was things weren't broken here in Washington on 9-11, but the telephone system and the cell systems were all saturated. My wife lives in Atlanta. I wanted to tell her I was okay. I managed to make an Internet call to a gateway in Tennessee and then make a regular call from there to Atlanta to let her know I was okay.
A number of people found that they were able to notify people either with e-mail or with voice over the Internet on 9-11 to convey information, which was convenient for their families anyway.
Let me tell you about the infrastructure programs that are underway at the National Science Foundation in the Networking Division today. I have a list of them here. There is a middleware -- I will say a little bit about each one of these -- middleware program. Still a small high performance network connections program, strategic technologies for the Internet and then two programs, which are just getting off the ground, which will be announced very shortly, although the announcements for these programs are still in the approval process in the depths of the Foundation, the experimental infrastructure networks and research test bed networks.
The network middleware, this is software, which is common to multiple applications. What we found is that even though we have available high performance network backbones, that doesn't really mean that it is an easy one to develop applications that use those backbones and it doesn't mean that things are always going to go smoothly. It doesn't mean you are going to get the performance that the backbone can provide unless you have tuned the, for example, TCPIP parameters of the end stations to take advantage of the network that is available to it.
As many of the scientific disciplines are coming to use the power of computation as an essential tool -- computational biology, for example, is becoming very important. In almost any field, the power of computation is becoming an essential feature of research. As we develop applications for that computation and for access to the large databases that the various disciplines want to use, there is a lot of middleware, which can go between the network and the application that is common to the various applications and doesn't have to be reinvented in a stovepipe sense for every application.
So, what we have is a program, which is attempting to design , develop and deploy and support a set of reusable and expandable middleware functions and services that benefit those applications. The hope is that this will enable things to progress more rapidly and more transparently and that we could have an effective collaboration. Grid technologies is one of the things that is moving ahead. The idea of a grid is that they are computational resources distributed at various sites in a grid. Sometimes the analogy is made with a power grid. Users are certainly distributed geographically and so for distributed users to have access to distributed resources and to interact with each other with advanced services certainly has a potential to expedite research and education.
I think this is very important. Education and health services are two of the key things that we need to improve the quality of life and I think the network and in particular in my opinion the middleware type things that can make that possible are things that have to be in place in order for that to become widespread, as well as broad band access in the last mile.
So, middleware then is the software that makes things transparently used, makes this idea of transparently using things happen and we have a program now, which we are investing about $10 million a year on, which is now entering its third year and we are -- we hope that it will have some successes there. Grid computing, public key interface infrastructure for authentication and so forth are two of the first leading edge components of that middleware that are being developed.
We still have a small connections program. Early on, when the vBNS was just getting off the ground, we established a connections program to enable a large number of research universities to connect the vBNS and to Abilene. As things moved along, the vBNS evolved from a research network into a commercially successful venture, vBNS Plus, and the R&D community has developed on its own as a grass roots bottom up development, the Internet 2 consortium, which Doug heads, and the Abilene Network, which it operates as a very high performance network for that community.
There are still about a $2 million a year program in high performance network connections to enable primarily to this point smaller schools and institutions, who have applications that require at least 45 megabyte connections that can take advantage of the high performance backbone, Abilene and still in a few cases, the vBNS to do so. So, we are still managing to add maybe a dozen new participants a year generally in partnership with some of the research university participants in the Abilene Network.
The strategic technologists of the Internet, again, this is about a $10 million a year program. This year from budget constraints, we will limit it to about $6 million. It focuses on -- it is a research program, but it focuses on things that have a potential to impact the Internet infrastructure in a three to five year time frame.
A couple of projects that I will just mention, there is an HPWREN, high performance wireless research and education network, in San Diego area that is using unlicensed bands, that is using line of sight microwave links that are commercially available. It is providing 45 megabyte connectivity wirelessly in areas that are difficult to get to. It is providing a 45 megabyte, for example, for Mt. Palomar Observatory.
It also provides wireless connections into a number of Native American communities in the area there east of San Diego. It provided for the first time Internet access into those communities. That has turned out to be exciting. It is providing wireless access for instrumentation associated with the Scripts Oceanographic Institute there to bring data from remote instruments back into the laboratory. Some of that, in fact, is actually off shore at this point.
So, over line of sight links demonstrating the capability of wireless to get us some moderately high band width connections into remote areas. There is also a Web 100 Project, which is focused on determining the -- making network measurements to determine the state of the network and use the information from the network to fine tune the TCPIP parameters to allow applications to get the maximum performance that they can for particular applications on a high performance network.
So, those are just a couple of examples that focuses on innovative access technologies right now as we think that is one of the key areas. Vint mentioned Gigabyte Ethernet and we have seen the crunch in the dot com industry. We made promises that we haven't fulfilled, I guess, but if you stop and think about it, if you had a gigabyte in and out of a hundred million homes, that is a lot of zeros. I am not sure there would still be any band width excess fiber glut if we could manage to pull that off.
Internationally, we have connectivity through StarTap facility, which started out as an ATM switch in Chicago. It is now gone to a 10 gigabyte Ethernet interchange point called STARLight, where people are bringing wave links from Europe and from Japan. The international connections that we support financially are through programs called the TransPac, which has partners with the Asia Pacific network. There now are two 622 megabyte links to Asia going into Japan.
The EuroLink Project provides connectivity into Europe and there are, I believe, three 622 links involved there at this point. NOCANET(?) is a high speed -- I think right now moving toward 155 megabyte if it is not already there, connections into Russia, into Moscow. There are other international connections. These are the ones that we have been providing some financial support for.
We are moving into the future now into a three-tiered network model that the top level is the operational research network. This is the Abilene Network that Doug has described and the new efforts that we are just getting underway of what we call experimental infrastructure network and a network research test beds.
Abilene is an operational network and it is supported by the community. As a matter of fact, we provide support for focused activities originally to enable universities to connect and now to through things like the Web 100 Project and the middleware activity, to try to accelerate the development and enhance the performance of it, but the Abilene Network is an essential tool. The research and education community really couldn't function in today's environment without it. So, it is really a 24 by 7 network and it has many industrial parties, including one state partner. The State of Indiana supports the network operations center financially in Indianapolis.
The experimental infrastructure network is something that we don't anticipate interconnecting, the large numbers of sites that the Abilene or Internet 2 does, but we anticipate it being driven by advanced applications and cutting edge technologies. There are many applications that can take advantage of network features that can't be provided in a 24 by 7 operational network today.
Some of those features would be ultra high speed rapid reconfiguration. Some of the applications are in high energy physics or in astronomy or atmospherics and we anticipate the network being driven by those high end applications, other high end applications other than just those.
And we anticipated being a partnership with industry so that vendors would be able to have an opportunity to try out new equipments that are not really ready for prime time, not ready for sale, not ready for 24 by 7 guarantees, but nevertheless working it well enough to be in support -- be used in support of applications and used in a trial basis then.
The research test beds on the other hand, there might be a number of these. We will see what we get in response to the announcement this year. We want an opportunity here for things to be driven by cutting edge networking research, want an opportunity to experiment with technologies that may be disruptive. We anticipate wireless and optical technologies, perhaps some Sensinets(?) and things that will improve not only things in the long haul area but also in the access area.
So, that is the picture of where we are moving with things at NSF now and I thank you for your attention.
DR. LUMPKIN: Thank you very much. It is very interesting and thought provoking series of presentations. I think I will probably pass on the socks for right now, but, Jeff, you had a question?
MR. BLAIR: Could you help me -- and I guess, Doug, you addressed this and our last testifier addressed this a little bit as well and maybe, Vint, you can chime in also. I get the impression that work is being done to help include aspects of authentication with public key infrastructure technologies for -- which might be helpful for accreditation of health care professionals, maybe even accreditation of other caregivers and maybe even patients, where you are beginning to look at seeing how you could integrate that into the upcoming infrastructures like for Internet 2. Could you just help me understand what is happening in this area better?
MR. VAN HOUWELING: The fundamental effort here is to make sure that we have a standard way of authenticating users of the network and then allowing that authentication to be also on a standard way to support authorization to use particular facilities.
MR. BLAIR: Since I can't see you, is this Vint or Doug or --
MR. VAN HOUWELING: This is Doug.
MR. BLAIR: Doug. Thanks.
MR. VAN HOUWELING: Vint stepped out here for a moment.
MR. BLAIR: I never knew.
MR. VAN HOUWELING: He was quite quiet about it.
The fact is that as all of who use the Internet know, we have to authenticate ourselves to almost every service we get on the network separately.
Vint is back now.
So, you have to give every new person that you are going to buy something from your credit card information over and over again. You have to have a separate password, a separate user I.D., essentially for everybody that you work with in the research community.
We also suffer from that. If you work at several different institutions, chances are you need a separate identification and a separate password for each institution you work at. This is all a barrier to using the network in a convenient fashion. Furthermore, the fact that you have to keep all this stuff around means that you don't secure it very well because it is impossible for you to remember it all. So, you inevitably put it someplace that somebody who is diligent enough could find it or you use simple kinds of passwords and so forth that are fairly easy to discover.
So, the work we are doing together through sponsorship with the National Science Foundation and its network, this National Science Foundation middleware initiative is designed to create a secure and standard way to both authenticate and then share the authentication across organizational boundaries. So, we now have in place at a number of universities in beta testing a software that allows an individual at one university to authenticate to that university and then assuming the other university has provided access to then use that authentication to access informational facilities at another university without reauthenticating.
We expect as a result of the work that we are doing collaboratively also with the grid community that within -- I think probably within the next couple of years, researchers will routinely be able to authenticate at their local institutions and use facilities and access information all across the world. That is the objective that we are trying to achieve here.
MR. BLAIR: Clarification between credentialing certification and authentication. If you use a process of certifying where you have background information on individuals that they happen to be board certified in a particular area and that they are affiliated with a particular hospital, that is all certification and you can wind up using public key infrastructures for that, but there is the other piece, the other exposure on that, is would someone without some type of biological authentication be able to wind up saying that they are that person?
So, does this -- are you matching this up with some kind of biotechnical authentication to complete it?
MR. VAN HOUWELING: As a matter of fact, this technology is independent but parallel to the various technologies that are used -- ensure that you are you when you are on the network. Typically, the way things operate is that you can only be authorized through these systems to use facilities and information up to a certain level with more casual kinds of personal identification.
When you actually get to the level where you need to have very secure -- access to very secure types of instruments and so forth, then, in fact, the best practices today are to make sure that there is something that identifies you biologically, something that you carry with you that only you can have and also something that you remember so that you get multiple ways of making sure that the person who is at the network port is who they say they are.
For most uses, you don't need all of those levels to enable information access.
MR. BLAIR: How soon do you think that this capability will be available to the health care profession at a reasonable cost?
MR. VAN HOUWELING: As I said, my hope is that we are going to see -- there is work going on. Also, I have to say in industry, the Liberty Alliance, various Microsoft efforts, those activities are beginning to converge with regard to standards. So, I am hopeful that within the next two or three years you will see these kinds of capabilities widespread in the Internet. I want to make sure to caution you, though, that simply being able to share authentication information does not necessarily mean that we have solved all the problems of securely identifying people who are at terminals.
Vint was going to say something about that.
MR. CERF: Thank you.
A couple of comments about this. We struggled for almost -- well, let's see, since 1977 with this whole public key cryptography mechanism, which looks so attractive and yet it has not taken hold, except in certain areas that turn out to be almost invisible. Whenever you are surfing the net, for example, your browser will often exchange public keys with a server site in order to secure the communication. However, that is not something you have to engage in because the only purpose behind the exchange is to make the information flow over the net in a confidential way. It is encrypted. That is very different than trying to figure out who is on the other end.
I came in a little bit late in the conversation but I want to draw a distinction between authentication an d authorization and I am pretty sure you made those distinctions.
With regard to biometrics, let me caution you about some examples of things which would be mistakes in my estimation. For instance, some people would say, well, let's take a fingerprint and let's digitize the fingerprint and let's use that as your identifier or two fingerprints or something like that. There are occasions when these identifiers get compromised. I mean, this happens with the non-biometric identifiers.
Usually, what you do when an identifier is compromised is you discard it and you just generate a new one. You get a new I.D. in effect. Getting a new finger is a bit of a problem. So, what we would suggest is that if you are going to use biometrics as a primary source of identification, that you combine it with some other thing, like a random number, which can then be dissociated from the finger and a new one associated with it so that you don't wind up having to cut people's fingers off to claim that their identities have been compromised.
MR. BLAIR: It seems to me, unless -- you know, correct me if I am wrong -- that we need certification. We need then in addition to the certification for real authentication, we need biometrics, plus we need verification, which is what you are saying, some other independent form. So, we virtually need three elements. Is that correct?
MR. CERF: I am not sure that I would put it that way, although I certainly agree that if you are going to decide that someone should be authorized to do something that you want them to be certified. You want to know -- so, having done that, now the question is is the person I am dealing with the one that got certified. You need authentication for that and the authentication mechanisms are much as Doug mentioned. Sometimes it is what you have and what you know; for example, a Smart Card, with cryptographic information inside, which has been generated by the card itself.
It is very common now to take a Smart Card and have it generate both halves of the public and private key and keep the private key inside the card and never let it go anywhere. That card now becomes a mechanism for digitally signing things, which is a way of authenticating yourself.
MR. BLAIR: Exactly. Could I just add one last question to my thread of questions here? That is the extension that you just is to, which is digital signatures that are authenticating that a particular document has not been altered or changed in any way, tied to the authentication, which some people would call an electronic signature. Could you help me understand where we are in achieving that threshold?
MR. CERF: Actually, the digital signatures for protecting document integrity is a very developed process at this point is being used in a number of system, one of them in the copyright office where you submit digital objects and they get digitally signed, so that later when you want to find out if it is the copyrighted object, you can verify it by doing the computation on the digital signature. That technique, the mechanisms for it, are very well worked out.
The thing which I found fascinating is that companies like Verisign(?), who have made a business out of doing various kinds of applications of public key cryptography, have been able to make a business out of this under certain circumstances, for example, with Internet or Prize(?) or within a given government agency or within the military, but they have not been able to make a business out of it for the general public.
When I asked what was the barrier, they said the real problem was the validation of the individual's credentials. How do I know who this person is? They walk in, you know, with a fake driver's license and what have you. So, the cost of doing the validation of the credentials of the individual before you issue a public key certificate that allows the individual to claim he or she is who they say they are, it is the cost of doing that that is the barrier so far for the general public. It is terribly frustrating because we could be using this technique in many different ways if we only had authentication for individuals as well as, you know, within a certain context.
Which by the way it leads me to one other observation, when Doug was describing, you know, having multiple identities, which you have today with your multiple credit cards and so on, it may turn out, whether we like it or not, that having multiple identities is going to be the only way it works. In other words, a single identity has the risk that if it is ever compromised, everything is compromised. So, there is an interesting tension between convenience, which you clearly point out would be improved if you had one to authenticate and then everything else works on that basis.
But that also is a potential risk factor. So, I am tempted to believe that in the end we will wind up with having more than one identity, partly to mitigate that risk.
MR. BLAIR: Interesting.
DR. SHORTLIFFE: I think a line of questions that we have just heard from Jeff can be generalized to the issue that I would like to address -- you know, Vint, you have painted a picture of the future and the directions we are headed in, which for anyone who works in biomedicine resonates well with notions of what the health care community and the health promotion communities are going to need over time in order to better leverage networking, especially the wireless components, given the nature of clinical practice and the mobility of people as they take care of patients and the like. We see this happening a lot.
So, both Doug and Aubrey talked about current activities at the federal level or the national level that are pushing this envelope in directions consonant with the view of the future that you were drawing. Doug, you mentioned, you know, there is the health-related subgroup that Mary Kratz oversees, but, of course, Internet 2 and UCANE(?) are much bigger than biomedicine and health care.
In the case of NSF if there is at least a perception that it would be a rare networking group in a health science center that could apply to and get support for the kind of things that are going on in health science centers with any of the NSF programs because of a sense of mismatch between the -- you know, the goals and the activities in the health care community and what NSF supports in the way of sort of basic research in networking.
I would like to get a feel actually from both Doug and Aubrey about to what extent you really see the kinds of health care issues that clearly are on Jeff's mind front and center in the way in which your organizations are anticipating the future, doing your planning. Is health care really a kind of interesting set of applications or are the needs of health care actually influencing day-to-day planning decisions or operational decisions with the way in which the technology is evolving or how you prioritize things?
I would like to know to what extent you believe that those needs are really reflected on a day-to-day basis as opposed to as an interesting application.
MR. BUSH: I guess one of the constraints early on was that when the National Science Foundation was created, of course, anything that is clinical is not within our sphere of effort. Anything that is not clinical would be. There are a lot of things in networking that are not specific to a medical environment. I would have to say in direct response to your question that, you know, we don't -- I don't think we are driven in our day-to-day activities by the special needs of the medical community, but as boundaries there between the biology group, which is a part of the activities of the National Science Foundation, and the specific needs blur, then certainly there is an overlap.
For example, we are highly interested in all aspects of collaborating at a distance, telecollaboration, and I think that is a key element for medicine as a distance as well.
So, while there are special needs of medicine that would be resident just in that community, I think, some of that just is special needs of authorization and access and privacy that most of us would like to have, I guess, about our medical information. Probably more private about our medical information than we are our financial information. So, there are some special needs there that would not be on our day-to-day activity list.
The privacy elements are because personal information of students and so on, the public key infrastructure that we are trying to get rolling now with the middleware project, with the Internet 2 group as a partner in that. We are also hoping that the National Science Foundation itself in its operational networking aspect, as we do business electronically with universities across the country will be a leader in adopting some of that as well.
Right now we can do a lot of electronic interaction, but it is with passwords that are not all that secure. We hope George Strawn, who will in the next panel, is a CIO for the foundation and maybe he can comment on some of those directions.
DR. SHORTLIFFE: Just to follow up on this before Doug responds, there is a perception of if not active collaboration and a great deal of interchange and communication in the networking interested groups between NSF and other key federal agencies involved in networking work, ARPA and DOE and NASA in particular. Has there been a movement towards similar kinds of interactions with either HHS or NIH in terms of networking planning design, networking policy, et cetera.
MR. BUSH: I think the primary vehicle for interagency coordination has been the National Coordinating Office and Dave Nelson is here with us this, is the new director of that office. There is the Large Scale Networking Committee, which works under that umbrella. George and Dan Hitchcock co-chair that group. There is representation in the Large Scale Networking Committee from NLM and NIH.
That is the primary point of exchange of information and coordination of activities. So, we try to stay aware through that vehicle of what the various agencies are doing and try to have the agencies aware of that and, therefore, able to complement and enhance each other's activities rather than overlapping in a destructive way or wasteful way.
DR. SHORTLIFFE: I am sure we will talk more about this in the next panel. Thanks.
Doug.
MR. VAN HOUWELING: In Internet 2 we have several application-focused activities. One of them is in the medical area. One of them is in the engineering and hard sciences. Another one is in humanities and the arts. We typically because as -- despite what you said, we are not larger than the biomedical. It is certainly the case that our span of interest and applications is large, but, you know, we know we are an organization that has a budget somewhere in the neighborhood of $20 million and it is all contributed by our member universities with the exception of some very welcome grant money from the Foundation.
But that is probably only about 1 percent of what we do. What we have tried to do is be a good partner to institutions like the National Institutes of Health and the Food and Drug Administration. I have myself and Mary Kratz has done more of this, we come to meetings like this and we try to help folks understand the capabilities that we offer and then we ask for advice on how we can collaborate with them. That collaboration usually means that through some grant, they provide some resources to either us or to our members so that we can actually do the work on their behalf.
At this time, we have one relatively modest grant from the National Library of Medicine to do some survey work on Internet access and so forth in health centers. But other than that, we basically have one FTE working in this area in Internet 2, which is not -- I hasten to say is not allowing us to meet the needs of this community for advanced networking.
Now, fortunately, it is the case that a number of our members have done significant work, but at a meeting of the research resources organization at the National Institutes of Health -- I think you were there -- a key problem was identified that a number of the medical centers and research centers do not have good infrastructure within the centers to connect to this high performance network. So, there are a number of barriers to actually using these technologies that need to be addressed.
DR. LUMPKIN: Kepa.
DR. ZUBELDIA: One of the things that the National Health Information Infrastructure white paper talks about is the possibility for the medical records to be held or controlled, at least in part, by the individual, the patient, perhaps with a little -- ISP or another mechanism like that. In order for NHII to work in that environment, we would need some sort of naming infrastructure that could identify where your record is located, based on your name or identifier or whatever identity you take at that point.
We have a similar problem today with HIPAA. In order for the providers to send the transactions to the payers, they need to know how to reach that payer, in fact, how to reach that payer for that specific transaction because a payer may choose to go through a clearinghouse for real time transactions and through a different clearinghouse for batch transactions.
Could you talk a little bit about the naming infrastructure that supports the Internet today and how it could be used for this environment?
MR. CERF: First of all, I would not recommend using the domain name system for this purpose. The reason for that quite simply is that that binds names to machines, computers and that is really not what you want. There are some systems that are being explored now. One of them, my colleague, Bob Kahn(?), has been working on for quite awhile and has instantiated with the music industry and with the copyright office at the Library of Congress. These are essentially object identifiers.
They are potentially just very large random strings of digits but they can be -- because they can be a very big string, there can be a very large number of these object identifiers, which you will need because you will need for things to be unique. You will need for them to be known for quite a long period of time because even though a transaction may not take very long, all of the accounting that goes along with it, perhaps even keeping an audit trail of a particular procedure so that you know what was done with a patient.
That information has to be known for quite a long period of time, perhaps as long as the patient is alive. So, having the ability to assign these kinds of identifiers that are very, very big and then process them so you can find the information that is associated with them is exactly the problem that I think you are describing.
Let me give you an example from the copyright world. There is this notion of international standard book number, SBN, a similar thing for journals. You could imagine extending that notion if you have an identifier for virtually any document. That is what Bob has developed with the system that he is running right now.
What is interesting about it is that you can take this very long, almost random number and we would perform a function on it called hash coating. All that does is map the string to a particular server that knows something about where that information is actually kept. In some ways, it is like the domain name system because in the domain name world, you take a string like www.nih.gov and you look it up in a table and it points you to someplace else that says where the actual physical machine is with that name, that serves that particular domain name.
Hash coating allows you to take a much larger space and uniformly spread it across a bunch of servers. So, I would -- in the case of the copyright, the interesting problem is that copyright information has to be maintained for, let's see, something like 75 years plus the life of the author. So, this is a time period, which is more than sufficient to deal with most of the medical information I can think of that is associated with a given patient.
I don't know how to -- I don't want to spend a great deal more time right here trying to describe the details on how this stuff works, but, in fact, there are object identification systems that are being worked on in other contexts, which I believe are directly applicable to the problem of medical record keeping.
Moreover, the information can be maintained in a protected form. It can be digitally signed. It can be encrypted. Different parts of it can be protected separately so that you have some control over who has the ability to look at various parts. So, you can imagine having a lot of the information, medical information, being accessible to a broader community of people, those who are doing research worried about epidemiology and the like, while the specifics of an individual's identity might be withheld and separately protected.
So, I believe that there are some methods for dealing directly with the problem I believe you are describing.
One other comment to make about just transactions in general in the medical world. One of the problems we have is understanding what is in these documents and something that Tim Burners-Lee(?), who developed the world wide web has been working on is called the semantic web. He is using extended market language as a way of labeling the various components of a digital object.
Although this is not a trivial exercise and it is fraught with all kinds of potential traps, the ability to recognize content based on these labels and, therefore, be able to process it, could be quite helpful, especially if you start standardizing the information that is those documents in the same way that you have standardized descriptions of medical procedures for purposes of billing.
So, there are technologies that I think would address your problem.
DR. ZUBELDIA: But this doesn't address the issue of how do I find the host that serves my data. If I go to my pediatrician and they want to get the medical records of one of my children, they need to get to my computer at home. How will they know that?
MR. CERF: Actually, I would argue that that is a very bad idea. If someone thinks that by keeping your information in your computer at home that this is (a) protecting it and (b) making it accessible at will, you have now saddled every human being with the requirement to become a system operator and to assure that the infrastructure is so resilient that this potentially life saving information will be directly accessible from that person's home. I think that is a terrible mistake.
The right thing to do is to make sure that the information is properly protected, but put it into multiple 24 hour a day data centers that are backed up with power and all the other -- you know, the whole storage systems and everything else, disk arrays, that are deliberately set up to protect the information and to maintain its accessibility.
DR. LUMPKIN: Kepa just made a side comment. He is describing his home environment but Kepa is different than most of us.
Mary Jo.
DR. DEERING: I know over the past 15, 20 years, there has been a lot of discussion, first, about, you know, the last mile, then from the curb to the house, et cetera. I think we have all probably finally encountered what may be the last leap, which may be the most insurmountable of all, which is the distance between the screen and the brain.
I think for medical care, both for health care providers and for patients and consumers, it is knowledge management, rather than just the transmission of the volume itself, which is the barrier. Is there any work that any of you are doing in your middleware areas or other -- I mean, this is a problem generic to any knowledge industry. So, is there any really exciting work about knowledge synthesis and presentation that would be particularly interesting to us?
MR. BUSH: I guess at the National Science Foundation the term that is being coined is cyber infrastructure. Cyber infrastructure would involve high performance computation, high performance networking and also very large data stores and the management of those very large data stores.
The cyber infrastructure initiative is still in the talking stage. It will, I think, become a very important initiative at the Foundation. I think it is the
-- it is the thing that I feel personally that is going to be the next biggest thing that we will be doing and it will involve the data storage management. There are a number of projects that the various scientific disciplines are embarking on that will depend on that, particularly in biology and in physics. It involves more than just -- middleware involves more than just networking. It involves more than just high performance computation.
The data storage and management and availability is the next leg of things that support the whole enterprise, which I think is going to become known as cyber infrastructure.
MR. CERF: Aubrey, don't you think that that sort of gets down in the bits and the bytes and the management of large amounts of content, but I am not sure that that gets to the question that Mary Jo was trying to work on. I don't mean that as a criticism, but I think you are going after the semantic content here.
I have had these conversations before and they are very frustrating. There are for purposes of classifying documents, research documents, the medical subject headings, but those don't talk about procedures very well. They talk about, you know, the description of various pathologies and the like. But what you are also after is information about the actual protocols that are being applied for treatment.
You also need to communicate with ordinary human beings, who have not had exposure to a full vocabulary, medical vocabulary. So, now we have the problem of trying to translate in and out of common terminology. I know that Don Lindbergh has worked in that area or has sponsored work in that area in the past, as a member of the Meta-Thesaurus is one of them and MASH(?) is the second.
I believe that controlled vocabularies may be the only way that we will make much progress there, together with some very, very carefully constructed ontologies that help mitigate this gap between common usage, common language and what the medical terminology is. Then you have the problem of getting machines to understand, you know, what that all means.
So, the answer is that the AI community is I think still wrestling with that problem. Ted, you probably are closer to that than almost anyone, given your background. My sense is that we are increasing our ability to understand natural language text, but when they become specialized like this, it gets harder.
MR. MC DONALD: Yes. Mike McDonald from Global Health Initiatives.
On this point, I think there is some progress being made. Perhaps the intelligence community is doing the most amount of work there and there are substantial amounts of money going to it.
In DARPA now with the total information awareness program, a lot of it is focused in this area. I think one of the debates is the traditional AI approach is fundamental. There are fundamental barriers there and is it better to look at human machine networks so that we are using some of the pattern recognition in the networks of humans along with what we can use from the machine site.
DR. SHORTLIFFE: I would just add that nobody has commented on another element of this machine to the face problem, which is the whole issue of the cognitive perceptions of the user, which in the case of some of our user community in biomedicine -- we will leave out the professionals, although it may be just as big an issue there, but for many patients the mismatch between the way in which information is being presented today and what it is going to take to really match their mental models of their own health and their social and economic and other issues that create their perceptions is huge and it suggests a need for a great deal of cognitive work about mismatches and mental models between presentation and perception and the like.
MR. CERF: Ted, could you say something about this persistent rumor that in spite of all this, given all the information that is available on the net now and the misinformation, too, I hear patients -- doctors saying that patients are better informed than they have been in the past. Is that a fair --
DR. SHORTLIFFE: I don't think there is any question about that. They are also perhaps more misinformed than they have ever been in the past. Both things are happening and it is addressing that latter issue, of course, which is part of the concern. Physicians certainly see patients coming in with a lot more knowledge of their diseases, but they also come in with tremendous misperceptions and long lists of really rather silly questions at times because of the inability for them to adequately process what they are getting at.
NLM has tried to tackle that by a whole MedLine Plus effort just to try to produce more consumer-oriented health information than MedLine, which is, of course, intended for scientists and practitioners. That helps a bit, but that doesn't necessarily address all these cognitive issues either. It is just more trying to find the appropriate educational level and background knowledge to use it.
DR. LUMPKIN: Well, at this point, I would like to thank the panel. It was a very interesting and thought- provoking discussion and certainly raises some clarification for all of us, except for people like Kepa who have their own secure networks at home. But we do appreciate you coming in.
We are going to take a 20 minute break at this time.
[Brief recess.]
DR. LUMPKIN: We are going to go on to the next panel. I am going to ask the panel members to introduce themselves and contrary to popular belief, where the committee sets the agenda, the panel has reshuffled themselves and seeing as how they know their topic better than we do, we think that is a good idea. So, thank you for that.
So, let's start off with the introductions. Dan, we will start off with you and then I will announce your other panel.
MR. HITCHCOCK: My name is Dan Hitchcock. I work at the Department of Energy in the Office of Science and we are responsible for advancing computer research and large scale networking there.
With George Strawn, I am the co-chair of the Large Scale Networking Coordination Group, where NIH actually also participates from this community and we coordinate network support and some deployment of things and test beds for the research community across the Federal Government.
DR. LUMPKIN: Okay. Ted, we know.
DR. NELSON: I am Dave Nelson, director of the National Coordination Office for IT Research and Development in the Executive Office of the President and on loan from the National Aeronautics and Space Administration.
The what we call NCO or National Coordination Office helps to keep the program that we are going to be talking about on track.
MR. STRAWN: I am George Strawn from the National Science Foundation where for the last few months I have begun operating as the CIO. Prior to that, I was in the Computer and Information Science and Engineering Directorate as the executive officer. Prior to that, I was the director of the Networking Division that my colleague, Aubrey Bush, is currently holding.
DR. LUMPKIN: Before we start off, we are going to start off with Ted first and then we are going to go to David Nelson, then to George and then Dan is going to bat clean up.
I have a meeting elsewhere in the building and I will be leaving a little bit before the panel finishes. So, if I get up and pass the gavel over to Kepa, it is not an editorial comment.
Ted.
DR. SHORTLIFFE: Bill Yasnoff suggested that I kick this off with just a few words to set the scene because twice in the past, actually before I was a committee member, I came and reported on some other activities I was involved with that are quite relevant for this topic. Since some members of the committee are new, we just thought it might be good to quickly refresh your memory about those past studies. It is particularly interesting -- they are not that old. One is from 2000 and the other came out in 2001, but already I think we can see some change in the department here and in the mindset of the health care community that is consonant with some of the recommendations in the report.
So, for my summary, I chose to focus on ones related to the Internet, in particular, and as you will see, some of them I think still are areas for ongoing activity. This was the report that Doug Van Houweling mentioned that was produced by actually the National Research Council at the National Academy of Sciences.
It is available on the web at the URL that I have on there. I have a copy here if anybody wants to flip through it. Of course, you can order it from the National Academy Press if you are really interested in the content, although it is all on the web.
This was a report that was commissioned, I believe, by the National Library of Medicine primarily in which they asked us to look at the Internet and health care, which, of course, is the topic for today and, therefore, this report is particularly pertinent. There were four objectives that we were asked to address for this study. What are the necessary technical capabilities in the Internet to meet the needs of the health care community going forward?
What are the applications that will, in fact, drive those technical requirements with the right strategy for helping to assure that the health care community properly capitalizes on those capabilities and is involved in assuring that they are, in fact, part of the Internet of the future?
What, if anything, are capabilities of the Internet that are uniquely required by health as opposed to other segments of society that have their own requirements that they place on the Internet?
As you might imagine, in ten minutes I cannot summarize this book in any detail. So, I am going to just pick a few key points that seem most pertinent. Pretty much there was a conclusion by the committee that there was absolutely nothing we could think of that the health care community needed that somebody else couldn't also capitalize on and that the notion that we were totally unique in any area from an Internet perspective is probably simply not true.
On the other hand, health care introduces a whole set of complexities that are somewhat unusual and not the least of which is the rapid changes in topology of organization. Patients coming on and off. They need to be added, et cetera. VPNs won't meet the needs, for example, of lots of secure requirements that are out there because of this changing topology.
So, we do place some interesting additional constraints, although you can think of others who would benefit from solutions to those constraints as well. There is a large portion of the report that deals with the major areas of application. These are reflected in our own working group report, these six areas all being important consumer health clinical care, administering financial transactions, public health, professional education and use in the biomedical research community.
There is a lot of discussion about the different requirements of these different areas and interesting, the most intense requirements tend to fall on the clinical care area where you have broad requirements both in the band width latency technical end of the spectrum and also in access, ubiquity and reliability areas.
The findings of the report are broken up into these categories, a whole bunch of general observations and discussion about some of the organizational challenges because of the nature of health care and its fragmented distribution in the U.S. system at least, without much centralization. A discussion, a whole chapter on the technical issues with a focus on band width and latency requirements, quality of service as a major additional issue and, of course, security.
I have got organizational issues on here twice. I don't know why I did that, but policy issues is the last one and in a way the policy issues may be more pertinent today than any of the others for our discussion in this panel.
During this report, we became very aware that these are the books that the NCO produces every year, the National Coordinating Office for IT R&D. This is last year's cover and these were smaller little blue books back in the early nineties, but ever since HPCC legislation was passed and these books were produced, it has been fascinating to see the extent to which health care or biomedical topics have been part of the grand challenges that are summarized as one looks to the future of networking and high performance computing in this country.
But in spite of that, although there have been all these agencies involved in high performance computing and communications program in the 1990s, more recently in the next generation Internet or NGI efforts, the vast bulk of the focus has been on four of the agencies, both in terms of funding and in terms of intellectual participation, the activities.
If this is no longer true, I look to Dave Nelson and the LSN people to tell us that, but the rule of HHS, NIH, NLM, although they have been at the table, has been quite different I think or at least it was our impression when we did this study than the way in which DARPA, NSF, DOE and NASA have played roles and been funded through this set of activities.
There was a small amount of money that went to each of these other agencies, but the vast majority of the money went there, but also the intellectual connection was much stronger in the four agencies that is mentioned. You may say that makes sense, but I would ask why? What makes health care any different from the military or defense when it comes to the relevance of the Internet and information technology in general to their mission.
That was sort of a question that the committee asked a great deal. So, there were recommendations made in these four areas. I don't have time to go over all the recommendations but I thought I would make -- I would just summarize in two categories some of the key points that are relevant for today's discussion under technical requirements and public policy.
The key technical recommendations were, first, we should ensure that the technical capabilities suitable for health and biomedical applications are incorporated into the test bed networks being employed under the NGI initiative and eventually into the Internet. In other words, we have to make sure that our needs are understood and reflected in the design process.
And, second, that the health community ought to work with the networking community to develop improved networking technologies. In other words, we ought to not only be at the table, we ought to be intellectually engaged in the process because there is no resistance to biomedical considerations being taken into account on the part of those that are doing network planning and design. It is simply one could argue that these issues are not being raised front and center and, therefore, aren't part of the mindset as decisions are made.
The policy recommendations largely were thrust at HHS, a feeling that it should much more aggressively address a broad set of policy issues that influence the development, deployment and adoption of Internet-based applications for health care. Many of these ended up in our NHII report as well. It needs to provide strategic leadership for Internet-related efforts. It needs to assure departmental participation in national network design, technical implementation and policy setting. It needs to convene public and private bodies on Internet and health.
The four cross cutting issues affecting the various agencies within HHS -- by the way, there was a lot of discussion with HHS people that led to these conclusions. So, there are many folks within the department, who recognized that this has not been optimal in the past. These issues have not been optimally addressed within HHS.
Explore cross cutting issues, encourage information sharing among the agencies, advance the national debate on IT issues in health and create the organizational structures to ensure that policy issues are properly addressed within the department. So then, the second report, I also have a copy here -- I see a few others around the room -- came out in February 2001. Now, this was from the President's Information Technology Advisory Committee. It was not focused solely on the Internet. But as you might imagine many of the issues that came up were quite pertinent to any discussion of the Internet.
This, too, is available on the web at the URL up there at the top when you look under their various committee reports on the PITAC web pages.
I am just going to summarize the findings. I think the recommendations will be obvious as I look with you at these findings, that the United States lacks an accepted national vision for the role of IT in health care and this is in contrast to some other countries that have made a major effort to try to develop national visions in this area. Now, you could argue there is an attempt to address this very active right now that is going on but I think at the time that this report came out, it was a pretty fair statement.
A critical and enabling investment in biomedical computing infrastructure and enabling technologies has not yet occurred. The closest thing we have had arguably has been the network connections program of the NLM to try to get more hospitals on the Internet. Most of them, academic medical centers, have ended up on the Internet because their universities got connected and they get sort of back door access that way.
A number of difficult public policy and regulatory issues constrain the adoption of IT health applications by health organizations and consumers and advances in IT are critical in order for DHHS to accomplish its mission, to improve the quality of U.S. health and health care.
Next set, biomedical community relies on IT innovations produced by other parts of government. We don't do it ourselves. We wait for others to do it and then we use it. This may have adverse effects on the pace at which biomedicine benefits from IT research and the solutions may never adequately reflect the needs of the biomedical community because they are not part of the design consideration.
There is a line in one of these reports that basically says somehow or another NIH has to realize that biomedically motivated information technology research is biomedical research and, therefore, is part of the charge at NIH, not really a well-recognized notion at NIH right now.
We need a larger cadre of researchers and practitioners who understand both health and computing and communications, just the people problem that we are facing. Recognition of a suboptimal role in management of IT within the department because decentralized management constrains both the development of coherent IT vision and departmental activities applying IT and health care of biomedical research and that there is a lack of any kind of coordinated IT effort and leadership across the agencies. I think, again, we are seeing changes in this area in the last year or so.
For example, we have heard testimony about the Consortium for Health Care Informatics -- is that right? -- Consolidated Health Care Informatics. That has all happened since that observation was made.
Finally, a lack of central leadership within HHS or a centralized budget, which leaves agencies in DHHS functioning without coordination and guidance and the individual agencies in the department do not accept a mandate to support IT research and I think that applies to all those agencies within HHS with the exception of NLM that support extramural research, even if it is fundamental to their mission.
I think this has been raised very interestingly in the last year because of the issue of public health infrastructure and the role the CDC might play in supporting IT research that will support the infrastructure for Public Health in the Future, a lot of which is related to the Internet and the NHII and then the issue of whether CDC is well set up to actually provide the kind of research funding that would be necessary for that kind of work.
And you can comment on whether this is still a correct observation but it has been the observation that the department is perceived as a pretty minor player in federal information technology policy development. So, there is a quick start to what I hope will be a very interesting panel session. We have just the right people here to talk about these things.
DR. NELSON: It is a pleasure to be here. The hand held mike may slip away and if it does and my volume drops, if somebody would let me know because I know especially for those who are listening rather than listening and hearing a good voice is important.
My name is David Nelson and I am director of the National Coordination Office for Information Technology Research and Development in the Executive Office of the President. You have heard in the first panel and with Ted Shortliffe's remarks a number of the issues, the challenges.
I am here mainly to issue an invitation to HHS to work with us. The three talks you are going to hear now are myself, giving sort of an overview of the networking and information technology research and development program and then George Strawn, who co-chairs the Large Scale Networking Coordinating Group will talk about how that works and how HHS can work within it and to some extent is already.
Then Dan Hitchcock, who is also co-chair of the Large Scale Networking Group will talk about some of the technical issues that not only the Large Scale Networking Group is working on, but the other research coordinating groups within the -- and I will start using the acronym NITRD. It stands for Networking and Information Technology R&D Program.
So, my bottom line message is this is an invitation, that this is a mechanism whereby HHS can become more of a partner and participant. As a submessage, it is more than just networks. I think you already heard this morning that getting the bits through the fiber may be the easiest part. Getting them useful, reliable, available, secure, ease of use for those who aren't experts, those are still research topics and though many of the elements are in place, the final answers are far from being in place.
The Federal Government has played a critical role in supporting fundamental research in networking in IT and even though we think of fundamental as 10 to 20 years out, the Internet and computing has been so rapidly moving that many of the things we have been working on turn into practice within five years. It is kind of amazing. It is hard to see the boundary between fundamental and applied. I think as we go through this, you will see some of that. I would say that is a creative difficulty. Would we have those problems everywhere?
Federally sponsored research has helped to build the technology base for the IT industry. It is typically funded research that is too long term or too broad reaching or things that can't be captured with intellectual property or with invested capital. So, it builds the basis on which the private sector can carry on.
The Networking and IT R&D program, again NITRD, provides a mechanism for focused long term interagency R&D and information technology and as I said, in some cases, long term is five year payoff, rarely but not in an empty sense, even less, two, three years. It is about a $2 billion agency program. Each agency brings its own money to the table.
This is not a program that is like a federal giveaway for agencies. Come, there is money. Instead, it is a program where the agencies, and as long ago as in the 1980s, realize that their individual capability did not match their individual needs, that as an agency the mission needs, the research needs transcended what they could find in terms of not only dollar resources but intellectual resources within their agency.
They felt it worthwhile to come together -- this was originally in the HPCC, high performance computing and communication program, but that tradition has survived. The intellectual involvement is arguably as important as the dollar involvement. I offer that observation to HHS.
NITRD program is assessed and advised by the President's Information Technology Advisory Committee and Ted Shortliffe, the prior speaker, was a member of the President's Information Technology Advisory Committee, again, in a jargon sense known as PITAC and he mentioned this report. You have in your package also, I think, what we call the blue book and, again, Ted referred to it. And this is a fairly high level description of how the program works.
This is a wiring diagram and some of these charts are going to be fairly dense. I will not read all the words and I commend you to look at your hard copy, but at any rate it starts with the White House and the Congress and then the National Science and Technology Council and then under that is what is called the Interagency Working Group on Information Technology R&D. We could think of that as the board of directors for the NITRD program.
I co-chair that. Peter Freeman of the National Science Foundation co-chairs that. Then off to the left is the National Coordination Office, which is what I also chair and then up in the upper left in yellow is PITAC. Now, down below you have six, both coordinating groups and what we call program component areas. The program component areas, what you are trying to do in the coordinating group is how you do it.
As I said earlier, Dan Hitchcock is going to go back to some of the specifics of those. So, I am going to brush through them fairly quickly. High End Computing on the left and then Large Scale Networking, outlined in red, because that is the focus of many of our comments today; High Confidence Software and Systems, Human Computer Interaction and Information Management, Software Design and Productivity and Social, Economic and Workforce implications of IT, an IT workforce development.
I would suggest that the role of health care fits into each of these in a non-trivial way and I say health care in the main research all the way through clinical and financial and databased and so on.
The National Coordination Office, what do we do? We try to keep the CATS(?) going in approximately the same direction, even though we listen to them very carefully to hear which direction they want to go in, but they sometimes veer off. So, we help to formulate and promote federal information technology research and development in order to meet national goals.
Let me digress a bit. I could put this comment in almost anywhere, but those of us who participate in the program are all human beings. We all have health problems. We all have relatives and friends who have had serious health problems. I think Ted alluded to it and so I can only underscore it, but when we meet to figure out what the important research issues are, the things that the agencies should be working on, health care is always in our minds. We always try to think of health care issues.
For one thing, they are fascinating problems. For another, we all suffer or are for the benefit, but we don't necessarily have -- and, again, I think, Ted mentioned this -- we don't have the intellectual heavyweights at the table, who can speak authoritatively about these. So, we are trying to do them as amateurs and we would welcome the help.
I report to the director of OSTP. I co-chair the Interagency Working Group. We do coordination. We do planning. We keep track of budgets, assessments and so on and we support the technical groups where the work actually gets done.
This one really is an eye chart and I am not going to go through it, but you have it in your package. What these are some of the things that those coordinating groups work on. Just to pick one, we will do large scale networking. The issues that the large scale networking group deals with as a program component area, network access, reliability, security, scalability and management, active and intelligent networking, networking in extreme environments; applications, such as networks of sensors, grids of networks, collaboratories. You already heard discussion about collaborative technology test beds.
So, all of these are things that have a health care component and we invite HHS to be involved. I could go through the others. I will leave you to read them. We will talk a little bit about some of these issues in Dan's talk.
I always have to have a budget description. How big is this? What are we talking about? If you look, the total for 2003 -- and, of course, we are still on a continuing resolution. So, this is still not quite final, but, hopefully, it is close, about $1.9 billion. If you look in this, at the HHS participants now, we have NIH as a substantial collar participant, totaling slightly over $300 million dollars and we have AHRQ as a smaller participant, totaling just about $9 million.
The distribution of the columns is by those program component areas and I will leave it to you to do the mapping. This is just the initials. But as you can see, the biggest one is high end computing infrastructure and architecture and if you take high end computing research and development as well, it is definitely the biggest activity. That is partly the nature of what scientific research is all about. It is also partly a legacy of where the program came from, but you see a large scale networking at over 300 million as a very substantial component and then human computing -- human computer interactions and information management next biggest and the others somewhat smaller, but growing.
If you were to look at this equivalent of ten years ago, you would see quite a different program. It is a munch broader program now, much more equipped to deal with some of the social, personal, security, reliability issues that are of such importance in health care.
Let's talk a little bit about how you participate in the Interagency Working Group and through that, in the coordinating groups. The IWG member agencies, former board of directors at the IWG itself, but then they participate in the program component areas to coordinate on specific R&D goals, ensure adequate investments and maintain necessary budget visibility.
Now, the kind of things that are actually done in both coordinating groups are identifying research needs, what needs to be done. That is done together. That is it is not what are research needs for NASA, what are the research needs for NSF. We try to sit down and cover the whole waterfront. We often form workshops, work with outside groups, with our partners, Doug Van Houweling, Internet 2. They are often at our table. It is a broad net. PITAC is helpful.
It is what are the research needs. All right, given the research needs, what are the research plans? What are we going to do as a Federal Government? Of course, eventually that comes down to what agencies do, but we keep looking at each other to try to get a balanced program.
As I said, when this first came together in the mid-eighties, it was that the agencies realized their needs exceeded their capability and only by planning together and executing together could they accomplish what they individually needed. They almost got for free what the nation needed. Exaggeration, but almost for free.
Solicitations and application review. We trade reviewers back and forth. This was the intellectual involvement. So, if NSF does a solicitation in an area DOE might offer reviewers, that is mutually beneficial because NSF gets a stronger review process, but DOE knows what NSF is doing and can bring that back home. The same can happen to HHS.
Program reviews, how well are these going? What things aren't working so well? Maybe there are some outside resources that could be brought to bear. Maybe there is another activity somewhere that if you knew about it, you would do a better job. So, the interagency program review helps. Workshops of all sorts, I think research is a workshop rich environment. We don't know how to proceed other than do workshops. But they are often more informal. People sit around over the coffee or the beer. It can especially valuable. That sets the sort of raw ingredients that go then into research needs and research plans.
Also, outreach and partnering, partnering with other non-profit organizations, such as Internet 2, partnering with industry groupings, such as ITAA and as appropriate partnering with individual profit-making organizations. All of these activities are done across agencies, obviously not all agencies participate in all activities. There is a self-selection that goes on, but the availability is always there. The invitation is always there.
Now, in addition to those IWG members, we have some federal agencies -- and I just mention here FAA, Federal Aviation Administration, and USGS, U.S. Geological Survey, participate primarily as users of research as opposed to suppliers of research. Why would they do that?
Well, to identify the NITRD research that is applicable to their agency program, so as it comes along, they can harvest, but also to leverage agency research and development by offering their research needs. In other words, let's take one, FAA. FAA has serious needs for reliable software for managing the air space. Even though their research budget is very small, they influence the decisions by other agencies as to how to do reliable software.
So, this is an area where HHS can become involved as a consumer of research, as well as a generator of it. Then finally participating in test beds in standards development, test beds need people as testees. So, sometimes an advanced user, even though they are not knowing the research can be a vehicle for figuring out whether the research has actually borne fruit.
Here is contact information if you want to reach me directly. It is just nelson@itrd.gov. We have a generally answered e-mail address at nco@itrd.gov and there is a web site, www.itrd.gov.
With that, I will stop and turn it over to my colleague.
MR. STRAWN: Well, good morning from me also, ladies and gentlemen. I am very happy to be here discussing these interesting and important matters. I guess I would begin by seconding several things that have been said. I think Ted is right, especially in the early days. There was a big four agency grouping of DARPA and NSF and DOE and NASA, which short of put their heads together really early on, in the eighties, to sort of begin popularizing the ARPANET and spreading it beyond just the small research project that it had been.
I think it meant a lot to those of us who were there at that time because we did get our oar in the water. We did know what each were doing. We did badger each other to do things that would be compatible and make a broader sphere of activity and it worked. It really worked. Maybe it is because networks have to work. You have to interconnect federal networks. They have to work and so on and so forth.
The job of the Large Scale Networking Group has always had a very practical component, as well as theoretical research. I suppose there are areas of interagency coordination that work better than the LSN activity, but I am not aware of what they are. I think we have a 15 year track record. In the early days we called ourselves the Federal Networking Council, rather than the LSN, but other than going through that name change, this is a longstanding group and I certainly echo Dave's invitation that we would love to see more HHS participation.
That is the way you find out what people are doing. That is the way you get your agenda in front of us. You have heard all the interesting stuff up to now, what is happening, what the important futures and present products are. I am just going to briefly talk a little about process, about the process of our getting together, what some of the things we have done together in the past and why you might find it interesting to increase the amount of HHS participation with us.
The LSN works by forming teams of special interest. We have had a thing now called the JET, the joint engineering team, which is the practical group, which makes the activities, makes the actual network work together. We create interconnection points that have been called different things over the years. We have promoted use of comparable standards and the like and this has been our practical test bed coordination mechanism.
The other end of the spectrum, our networking research team, brings our program officers and others, who actually run the fundamental networking research programs of the various agencies together and that is where we make heavy use of the evaluate each other's programs, provide reviewers for each other's solicitations, et cetera, et cetera.
Of course, you can pick up an awful lot of information about what is going on and what needs to be done by participating in that type of activity. Our newest working group or at least newest named and most cleverly named, I guess, the MAGIC team, Middleware and Grid Infrastructure Coordination, since middleware and grid technologies and terminologies have become quite prominent over the last several years, these turn out to be highly beneficial to coordinate among the federal agencies and the university networking activity. So, we have a very active group, which is promoting doing things in a comparable fashion. This group has attracted quite a bit of interest from the private sector, who also sees a good thing and sees middleware and grids as an important future technology and we invite them in to watch what we are doing.
That is, I guess, a good point to remind you that the Internet itself is a good example of the government/university complex as opposed to the industrial military complex in terms of where these ideas came from, how they were first promulgated within ARPA and then within the other agencies for the support of basic science and after we have demonstrated the general applicability in a broader university and government laboratory environment and then the private sector said, whoa, there may be money to be made here. Step aside, folks. Don't hold us back. And the Internet industry was created.
This is a subset of the information that Dave had on his chart. So, I won't belabor it, other than to point out that all of these agencies do participate in LSN and that NIH, NLM levels have gone up quite a bit in the last few years. So, their participation is definitely increasing.
The next generation Internet was mentioned a couple times. I will mention it just briefly for historical purposes. This was a five year project between 1998 and 2002, where the agencies marched together under this common banner with the goals that you see here below. We wanted to enhance the fundamental networking research and QoS, multicast, measurement, security, reliability, some of the now fairly traditional topics were on our minds. The test beds were the most visible portion of the activity because we wanted to increase -- actually, we stated it originally, at least a hundred sites, operating a hundred times faster than they had been because at the point of time we began, a typical university connection was a million bits per second and we did end up with a hundred million bits per second, some more than that before the process was over.
DARPA led the ultra high group on their SuperNet, connecting more than ten sites at at least a gigabyte per second. By the time we were done, we had pretty well doubled those goals of 200 sites at the hundred megabyte or above level in 20 or so sites at the gigabyte. We did also focus on applications to be run over the test beds to take advantage of the fundamental network research and so forth. You will note we categorized them into both the general enabling applications, collaboration technology, digital libraries, other general topics and then looked specifically to science applications, basic science, crisis management, education, environment and we generally have health care as one of the important components that we always mention.
I think we always sort of wish we had more participation. Just as Ted mentioned, we -- and I will tell you that the agencies tend to go out of their way to say now how can we make more seats at the table for the health care
community. It is such an important area and, obviously, so related to an important application of information technology.
Specifically, there were several ways, I believe that the NIH community benefited from next generation Internet. You have heard some of these networks mentioned before. As the next generation Internet began, the NSF had a network called cryptically the vBNS, run by Vint Cerf and his crowd for us and we began expanding academic access to that network until we had more than a hundred universities connected to it.
At that point in time, NIH didn't run, doesn't run an independent network, but at that point in time we str