Distinguished Guest Lectures

The World's Libraries at our Fingertips through the Net

Salam! I am sorry, but that is the only Arabic you are going to hear from me today. First of all, I understand that I am interrupting student exams, which means that some students are not here because they are still taking their exams. However, some of you have escaped, so I hope it will turn out to be worth it. First, I have to thank Hisham El Sherif, as well as Professor Serageldin, for a wonderful introduction and a wonderful welcome on my first visit to this city and to this library.

It is a personal point of honour for me to speak at the Bibliotheca Alexandrina. As Hisham pointed out, when I think of libraries I always think of the library of Alexandria , which was a focal point of the ancient world's knowledge, and I think we have an opportunity with this new wonderful library to be the focal point of 21st century knowledge. So, I hope one outcome from today's discussion will be that we can move forward in that direction. Hisham asked a question: "Where would we like to be in the year 2020?" My answer to that is "alive!", especially considering how long I have been around and how many years we have put into the Internet. My task this afternoon is to try to convey some ideas to you about what the world can be like, and what it is becoming, as more and more information arrives in digital form. So, the title of today's talk is “All the World's Libraries at our Fingertips through the Net ".

The Internet: First Decade

I want to start by sharing some reactions with you about what I think of when I think of Egypt . Now I am sure that this is coloured by my history and background and everything else, but I can tell you that even as a child the history of Egypt fascinated me. And as I have grown up, and as my children have grown up, I have noticed that every child seems to be fascinated by the Pyramids, mummies and the Sphinx, and so we should take advantage of this. Every child everywhere in the world is fascinated by that particular part of your history. And the reason that that is so important is that once we get children's attention, then that is an opportunity. We want to teach them to read, to learn more about what they are interested in, and to open up their interests as broadly as possible. I think that because of this special opportunity that today's age affords us, highlighting the opportunity to put information online, this means that we can make the Bibliotheca Alexandrina the focal point of that initiative. The library is really an icon in history, and it can become an icon for today .

I want to move to a bit of statistical information about the Internet, just to give you some sense for how it is growing. This is a seven-year comparison, so we are going from the middle of 1997 to early or mid 2004. During this period, the number of servers on the network, the machines that do Web services or e-mail and the like, grew by over a factor of 12. This does not include things like personal digital assistants and the like, things like the blue tooth I am wearing: these devices probably add another 300 to 500 million items to the network. The number of people on the Net is not a known quantity. There is no one place where everyone has to register, so we do not really know who is on the Net. But the estimates are that there are at least three quarters of a billion, possibly as many as one billion, users.

While on the one hand that is a big number, on the other hand it is still less than a sixth of the world's population. So suddenly we realize that as big as the Internet is, it is still small compared to the population that should have access to it, but does not yet have such access. Just for comparison, the world telephone system has grown by over one billion terminations over the last five years: it had about 1.2 billion terminations five years ago, and now it has 2.3 billion, about one billion of which are wireless or mobile devices. Sometimes you probably feel as if all these are right here in Alexandria ! But the point is that these devices have expanded the telephone system dramatically. They are likely to expand the Internet dramatically as well, because many of them either already are, or will later become, Internet-enabled devices. We will revisit that idea a little later .

I want to tell you a story about my personal library. I have a few thousand books; it is not a big library, but I can tell you that when I am looking for something, and I do not know in which book it is, then when I stand in front of my book shelves I still think I do not have time to look through every book to find the thing I am looking for. Even if I know which book it is in, I still may not remember what page it is on, and the book might not be indexed. And so I stand there thinking: I love books, but right now I wish they were all online because I would be able to get help searching for the information I want. So, I really look forward to the time when all the publications that are available in print now are also available online and in searchable format.

There was a time when I was about ten years old when I wanted to do just this. When I was ten, the world was young and dinosaurs still walked the planet… We did not have the Internet, and we did not really have any computer technology at all, and the only thing that was available was microfilm. I had a smaller book collection then, but I used to be fond of reading books lying down in bed and holding them up above my head. The problem was that these were hard back books, and when I fell asleep, the books would drop and hit my nose, painfully. I wanted to have my books moved over onto microfilm, so I could project the books onto the ceiling. This was in about 1953. I went to the Kodak Company because they made microfilm, and I asked them how much it would cost to have my 500 or so books microfilmed, and they said US$3000. Well, my allowance was 50 cents a week, and so I was not in a position to take them up on this offer. Some 50 years later, I am still eager to find a way of getting my library online, but it will not be with microfilm .

For many people the Internet has always been around. If you are no older than, let us say, 12 years old, there is no such thing as a world in which there is no Internet, because it became visible to the public in 1994. On the other hand, the online searching technology is actually relatively new. I did not get introduced to this in any very profound way until 1992, and I have been using the Net all the way back to the 1970's. However, even in 1992 the idea of putting information online and having it searchable was still in its early stages.

In 1992, I was president of the Internet Society, and I was going through my e-mail when I found an e-mail from an irate Slovenian complaining bitterly that the suffix ".si" had been taken or reserved by the United States for something called the Spratly Islands, and he wanted me to fix this so the ".si" suffix could be used for Slovenia. My first thought was: I wonder what the Spratly Islands are, and why is the United States interested in reserving ".si" for them? I did not have the answer to that, so I put the message aside. The next message was an invitation from the University of Minnesota , saying "Why not go to our Web site and try out the Gopher application?" “Gopher” is a double pun, because in English a “gopher” is a little rodent, and this is the mascot of the University of Minnesota . However, gopher also sounds like to "go for" something, and the application essentially allowed you to work your way through a series of menus on the screen.

The Gopher application was not a search engine of the kind that you are familiar with today, such as Google or Alta-Vista. It was just an online information system that you could work your way through in a hierarchal structure. So, I decided to try it out, and I went through and found something called "CIA World Fact Book". I remember panicking, thinking, "Oh my God, they have put classified information on the Net, and we are all going to be in trouble". However, it turns out that this is an unclassified publication, so my next reaction was, well if anyone knows what the Spratly Islands are, the CIA should. I looked them up in the World Fact Book , and discovered that they are unoccupied islands in the middle of the South Pacific, and that they have three air fields and some fortified places, and that nobody lives there. There was no arable land of any kind; there was no land devoted to living space or anything; there was 100% “other” land (whatever that meant), and there were large quantities of guano on the islands. So, I thought that maybe that was why the USA was interested in the islands: they were a source of fertilizer. It turns out that the islands also sit on some important sea links, and the US also had an interest in those.

M next step was to ask the US State Department why they had reserved ".si" for the Spratly Islands . I never found out the answer to that, because a week later I got a thank you note from the Slovenian saying, "Thank you, the ".si" has been released and is now assigned to Slovenia as its country code on the Internet." I thought at the time that I get blamed for a lot of things that I have not done, but I sent him a note back saying "You're very welcome, I'm glad I could be of some assistance." I still have no idea what happened. But that was my first introduction to an online information system of any significance, and of course not long after that, in 1994, Tim Berners-Lee's World Wide Web burst onto our scenes in commercial form, when a company called Netscape Communications launched a browser that had been built by Marc Anderson & Eric Bina at the National Centre for Super Computer Applications in Illinois in the United States. This browser was Netscape Navigator .

The Search for Digital Objects

This history brings me to a general point that I would like to make, and which I would like to convey the importance of to you. Today, we are beginning to see an increasing amount of digital information on the network. Some of this is in very well-structured form, but some of it is not. We sometimes speak of this information as “digital objects”, which are objects that have a well-defined structure and are understandable to computers as content. The content is actually encapsulated in the middle of software, and you can think of these digital objects almost like viruses. If you are familiar with how a virus works it may be easier to understand the idea of digital objects.

A virus is a bit of DNA wrapped up in a protein coat, and we can think of digital objects as information wrapped up in a coat of software. The external software coat has standard interfaces, through which the object can be requested to do something with its content. For example, you can give the instruction, "Show me your Arabic rendering", or "Show me your French rendering", or "Show me your English rendering". Sometimes you may get the answer back that "We don't have one"; sometimes you might get the answer, "Here is our best estimate of that rendering": you may have gone through an automatic translation system, of which there are now several, and these produce varying results. But the main idea is that you have a standardized ability to interact with this digital object in order to get its content out in a form that you can use.

In some cases, this content might be rendered as sound. For example, you might have a musical composition stored away, and when you ask it to render itself as sound, what you get back is an orchestral rendering of the music. However, you might equally have asked it to provide you with its print rendering, and in this case you get sheet music from the same object. The idea is that these objects can be represented in a variety of different ways, can be presented and rendered in a variety of ways, and it is the software surrounding them that produces these different presentations. Of course, once we start producing these digital objects, we have the problem of where to store them and how to find them again. And, in fact, this is the beginning of the reintroduction of the concept of libraries, except in the online environment we have digital objects not physical ones. However, these digital objects, just like their physical counterparts in the past, have to be catalogued and indexed so we can find them again. Instead of using a card catalogue, we use computer programmes like the search engine Google to do this. Remember that here we are talking about deliberately structured information .

I want now to talk about another kind of information, which is the kind we have most of on the Internet. Today, we have a standard representation of content on the Net, this content showing up as Web pages and being typically encoded either in hypertext markup language (HTML) or in extended markup language (XML). This means that the information is not entirely structured, but that it does not have a rigid format either, and it has a lot of tags on it to help us decide how to present the information. In the case especially of extended markup language these tags help a computer to understand what the content is: it could be an invoice, for example, or a bill of some sort. If so, then the material is tagged in such a way that the computer knows that it is a bill just by looking at it, and it knows that you are supposed to pay it, and it knows it should ask you for your credit card to do so.

The result of having this standardized representation in HTML or XML has led to the creation of things called web crawlers that go through the entire Internet, looking at as many pages as possible. In the case of the popular Google search engine, for example, the crawler will go through some 5 billion pages, which is only a fraction of all the content on the Net. Those of you who are familiar with current physics will know about things called "black holes", stars of so much mass that their gravity field prevents light escaping from them. I think that there is a lot of “dark information” in an analogous way on the Internet today, sitting in databases that are not represented on the Web pages searched by Google or other engines, or sitting in databases that have to be “quarried” to extract information.

Often, when you are doing some sort of Web search, or interacting with a Web page, what is actually happening is that as you interact the corresponding Web pages are being generated by information from the relevant database producing the rendered Web page and then presenting that information to you. This is where the “dark information” on the Net is held, because this information is not visible to the search engines. Some estimates suggest that there could be as much as a thousand times more information in that form on the Internet than there are Web pages. So, we are only scratching the surface of today's information content on the Internet, even with the immensely useful search engines that we use today .

One possible way of accessing such “dark information” -- information that cannot be easily searched – is by associating it with information that can be easily searched. An example of this is what happened in the United States as a result of the Americans with Disabilities Act.

According to that law, television programmes had to be captioned so that they could be understood by the hearing impaired. I happen to be hearing impaired, so I was very happy to see that law passed. However, people at the MIT Media Lab then came up with a very interesting idea: they saw that news programmes were being captioned as the law said they should, and they recorded the programmes on video and separated out the captions as text, using this text as an index for the video. The result was that if you were interested in commentary on a particular subject that might have occurred in a news programme, you could search the captioning information, which would then take you to the appropriate frame and allow you to find the point in the programme that interested you. Google has a search tool that lets you find images on the same page as a text you are looking for, associating these images with the text. Sometimes, of course, the image has nothing to do with the text you are looking for; it just happens to be on the same page as the text. But often the image is quite relevant. The result here was the development of a kind of media confluence, using indexable material as a way of finding things.

The most recent introduction of unstructured information on the Net has been in the form of things called "weblogs". These are like diaries, in which people share their daily thoughts with anyone who might be interested. Some of us might wonder why anyone would be interested if you look at some of the weblogs that are currently online. Nonetheless, this kind of information is accessible by doing text searches, and this kind of unstructured content is becoming accessible to us in a way that it could not have been in any other environment.

The reason this is important is that in some cases these weblogs may be useful to us, if we are doing research for example. If the weblog is written by someone, a biophysicist or biochemist for example, who is using it to make notes about his work, then the weblog can lead others interested in the same kind of work to him or her. The Net has thus become an avenue for linking people together who have common interests, which is another important aspect of online information .

Tim Berners-Lee and the Semantic Web

Today, Tim Berners-Lee, the inventor of the World Wide Web, has another very powerful notion. I should say “Sir Tim Berners-Lee”, because he was knighted about a year and a half ago. Tim invented the World Wide Web in around 1989, when he was at CERN in Switzerland , and today he is enthusiastic about what he is calling the “Semantic Web”.

What he is getting at here is the idea that proper labelling of the information on the Internet can help us search through it more carefully. To give one example: if you are looking for books by a certain author, you could conceivably just put the author's name into the Google search engine or the Alta-vista search engine and then see what comes back. However, if the information on the Net included the kinds of tags that you can have with extended markup language, then there might be tags associated with an author's name that say: "This is an author's name". This constitutes the meaning of this text, as opposed to just a reference in it to a person's name. By tagging this information as a reference to an author, the search engine will only look for, only respond to, and only tell you about those pages that contain that name with the tag "author" associated with it. This will make Web searches much more effective.

When you do a Web search today, you typically get back 10 pages of responses, many of which will not relevant. The point here, based on Tim's idea, is that by using tags, and especially by using extended markup language, the codification of searches can be made much more effective. This is also having a side effect, which is as important as anything I can think of, because you can also use these same tags to represent information in business documents and in scientific data, or in other kinds of useful exchange. Because these things can be tagged, or recognized, as business documents, for example, a computer can understand that what it is looking at is an order, or an invoice, or a payment confirmation, or a bill, or some other kind of business transaction, and not something else entirely. All the information is tagged, so the software knows what it means semantically.

This system could be taken further, and one can imagine using the same structure of carefully tagged information as a way of controlling scientific instruments through the Net. Once again, the idea is that the information contained in commands given to the instruments is labelled, or tagged, such that the remote computer that is actually controlling the instrument understands what it is you want to do. Of course, when data comes back, let us say, from a microscope or a sensor system out in space, we can use the same labelling system to mark the kind of information that is being sent back. As a result, when we analyze this information we have a good tool at our disposal to understand the information received .

Side-effects of Online Information

Clearly, there are already a lot of important side effects of putting information online. One of them is already very, very clear, and that is the accelerating pace of scientific understanding and discovery, and particularly when it came to genomics, the study of human or other genes. We have been able to learn more about genetic information more quickly as a result of online information, because everyone publishing information in paper journals has been required to put the same information into one of the several human genome databases, or other genomic databases, making this information accessible to the entire scientific community. A side effect of this insistence that data be made available in machine-readable form in a timely way was that scientists everywhere had a rapid ability to discover whether anyone else had seen the same gene sequence that they had just discovered, for example, or whether anyone else knew more about what it than they did, or what it was for, or what it affected.

Today, of course, we have successfully sequenced the human genome, and we have sequenced other genetic sequences from other animals. Now we are at the point of trying to work out what all this genetic information really means. One of the things it means is that for those genes that are expressed, that produce proteins, hundreds of thousands of different proteins, these genes interact with each other in animal physiology. Working out the ways in which they interact is a much more complicated problem than figuring out the gene sequence. In a sense, all that we have done thus far is to climb to the top of the mountain of genetics, but now we need to climb a much higher mountain to understand the resulting proteome , or the collection of proteins that genes express. I expect to see much more rapid evolution of our genetic understanding as a result of climbing this second mountain .

A further development happening today is that universities across the world are starting to share information online. MIT, for example, took an enormous step forward by putting all of its technical information and educational materials online, and making these fully accessible to anyone anywhere in the world. Some university presidents thought that MIT was out of its mind to do so -- after all, this is valuable intellectual property. But MIT had concluded that the fastest way to attract the best students was by saying: "Here's the material we teach. If you understand this already, you don't have to come to MIT, but if you don't understand it, this is the place to come to." They had a very positive response back as a result, and in the course of this advertising initiative, if you like, they have shared wonderfully valuable tutorial material with the rest of the world .

Today, we are increasingly communicating online, using e-mail incessantly – so do the spammers, unfortunately – and this is going to be a never-ending babble. We use instant messaging, and we use other kinds of presence-based communication. Our mobile phones, for example, can now be programmed to say whether certain of our friends are online, so we can know whether it is worth the trouble of calling them. We can even imagine having documents that are malleable, that are active, where you can interact with these documents. I expect to see this technology increase the use of e-documents in business and in government. Many of us interact with government, and we would like to do this online, so that we do not have to stand in line physically, or stay on the phone listening to music before someone replies. Online, indexable, machine-readable information should improve our ability to conduct our lives and businesses online .

Standard Identifies of Online Information

However, as we produce more and more digital objects on the network, we need a reliable way of identifying them so we can find them again. Today, there is a massive initiative by a number of groups aiming to design such standard identifiers for digital objects on the network. Anyone who has ever purchased a book will have noticed something called an International Standard Book Number that is associated with it. Periodicals have similar kinds of identifiers. These can be used as handles, so to speak, for a digital version of these things. My partner Bob Cohn and I have spent quite a bit of time working on a system called the “handle system”, which allows arbitrary strings associated with digital objects to be assigned to them, allowing us to find these objects again wherever they might be.

What is interesting here is that not only do you want to find the object so you can use it again, but you may also want to make reference to any copyright controls there are on that object. Certainly the media, the music community, and the movie community, are very interested in finding ways to make information digitally available, but at the same time to constrain and restrict that information's reproduction and distribution. These kinds of controls can be put on digital objects, as a way of helping the industry protect its information. I am sure that many of you are familiar with digital signatures, which are a useful way of ensuring that an object has not been altered, as well as a guarantee that it comes from the source it says it does.

Aside from indexing the information, however, another real concern is the increasing amount of digital information we have and our need to put it somewhere. We have a variety of places to put it: disks and tapes and things of that sort, as well as CD ROMs and DVDs. However, there is an open question about how long these media will actually last. For example, I am very fond of a little new medium that I carry about with me. I have a digital watch, which does not look digital since the face is analogue, but nevertheless has a USB connection and contains 128 megabytes of memory. I carry my Power Point presentations and the spread sheet containing my wine list and my science fiction bibliography on my watch. However, I do not know how long the information will stay on the watch, so although I am a huge enthusiast for digital information, and always push the idea of putting things online, I have to admit to concern about storage, which I can illustrate by telling a story.

I was once at a library, a specialized library for really old documents, talking to the head librarian about putting things online and making them digitally available. I was saying how wonderful the new technology was, and I was demonstrating some CD ROMs I had brought with me to show the latest high-density digital technology. The librarian looked at me, excused herself and went into the stacks, and then came back with a vellum manuscript, an illuminated Bible dated to 1000 AD, and therefore one thousand years old. She put the book down on the table and said: "Would you tell me again: how long do you think this digital technology is going to last?" I had to tell her I did not know how long it would last, since we certainly had not had a thousand years of experience with a DVD or a CD ROM. So, she put me in my place.

This story tells you that if we intend to digitize information we had better get accustomed to moving the digital content from one medium to another and copying it as necessary in order to make sure that we really do retain it. With today's copyright laws it is important that we do retain information, because copyright laws, the Berne Convention in particular, say that copyrights last for 75 years after the death of the author in question, with the result that if you have a young-ish author of age 30, let us say, there are around 45 years of life, plus another 75 years, to be copyrighted, which is 120 years that we have to preserve the content to make sure that copyright is observed. So far, no one has demonstrated any digital storage medium that lasts that long, except for paper. I sometimes have the horrible thought that we will go back to punched paper cards in the 21st century as these have been shown to last longest. This would be a terrible outcome, and I hope that some of the newer technologies will be shown to last longer .

Digital Libraries and the Future of the Book

A further project underway today is the so-called "Digital Library", which by now is a generic term. Many universities in the US are involved in this, as are some of my colleagues on the Petersburg Price Committee. One of the things encouraged in the creation of digital libraries is not only taking information naturally produced in a digital form, using word processors and the like, or material produced as output from some data device, but also transcribing material that is now in the public domain into digital form, making it accessible in that way. Two countries that have been instrumental in producing this material are India and China , because the labour involved in transcribing some of this material is significant and the costs there are reasonable. As a result, we are seeing an increasing amount of material developed in this way, and of course it would be wonderful if the library of Alexandria became a focal point for this kind of initiative.

Indeed, I occasionally speculate about the future of the book. Today, we think of books as rather static things: you open them up, and page 25 of a given book still has the same content this week that it had last week, and last year, and ten years ago. However, as we start moving towards digital versions of these things, it would be interesting if books could start to talk to each other, to update each other, so that the next time I publish a new version of something, for example, all references to it can go to the new version, pull the new information out of it and update all previous versions.

Marvin Minsky of MIT has speculated that a hundred years from now we might be saying to each other: "Did you know there were once books that didn't talk to each other?" I sometimes worry about what it would be like to live in an environment where books keep changing, because I am fond of returning to the pages I remember, and I sometimes remember where information is in this way: it is about half way through the book on the right-hand side of the page in the upper half, where the mustard stain is… A dynamic, digital book may not be memorable in the same way as a traditional one, so we will need more tools than we have today to keep track of what books are doing to each other .

The Future of the Internet

I am going to shift gears a bit here, because now I want to give you some examples of what I anticipate is going to happen to the Internet in the future. One thing that has most surprised me over the last 30 years has been the discovery of things on the network that I never thought would ever be a part of it: things like refrigerators with liquid-crystal displays on them, or picture frames that are actually Internet devices and have storage for 20, 30, or 40 images and step through them however fast you instruct them to do so. Even telephones are becoming part of the Internet environment. As a result, I have been speculating about what the world is going to be like, and I would like to share these speculations with you.

Most of the future devices will be programmable and will use high-level languages like Java and Python. Some such devices are already in operation, like Web TVs. For example, I was in New Zealand a year ago, and I discovered that in my hotel room there were four different ways to get onto the Internet: there was a telephone jack that you could use with a modem; there was an RJ48 Ethernet connection you could plug into; there was a Web TV with infra-red keyboard that you could use to surf the net; and if I held my laptop at just the right angle next to the window, I could pick up a hot spot down the street in an Internet café. I have to admit that I could not type on the laptop very well at that angle, but that was a fourth way of getting onto the Net. We are seeing more and more of these kinds of devices showing up in a networked environment .

Video games are also becoming Internet enabled, and this has led me to another interesting speculation. The kids that play these games like to be in the same room together, screaming and yelling as they shoot at each other on the screen. However, sometimes they cannot be together, and they have to be in different locations, or in different cities. What is now happening is that when these games are put on the Net the kids can still play together as a group even though they are no longer in the same physical space. Some of the devices have films as background, and of course they already have speakers so that you can hear the sound effects of the video game and talk to one another while playing. This turns out to be important, because these game devices have about 20 buttons and no keyboard, and there is no time to type or anything, because the player is too busy pushing buttons to shoot, or whatever it may be. Therefore, players use voice information to coordinate their moves.

It occurred to me that with an inexpensive television set, or rather with a small television camera, you could put the camera on top of the television set and pick up an image of the other player, enabling you to see your friends and talk to them while playing the game. That would be interesting because you could have a collection of people physically separate from each other who could nevertheless see each other, hear each other, and shoot at each other, which sounds like a good definition of a video conference! Therefore, I am beginning to think that what will happen in the business world in the future is that these video games, which are inexpensive consumer products, will be brought into the business environment, plugged into high-speed business local networks, and used for video conferencing. Of course, if the video conference gets boring, you can always go back to playing the video game .

Unfortunately, I do not have the time to go through every example of what may happen in the future. However, it is surprising, for example, that there are Internet-enabled washing machines being test-marketed today: college students like them, because you can put your clothes in and go off and have a beer and the machine will send you an e-mail, or page you, to tell you that your clothes are done. You no longer have to wait watching your clothes turning round in the little window. There are also commercial refrigerators that are Internet-enabled, and these have a touch-sensitive liquid-crystal display. I do not know how it is in Egypt , but in many American homes the family communicates with pieces of paper stuck on the fridge with magnets. Now we have been able to improve family communication by using e-mail on the refrigerator .

Another thing happening in our part of the world is the introduction of a new device called an RFID, a radio-frequency ID device. These are now frequently used on automobiles, so that when a driver goes through a toll-booth an automatic payment can be made by the device. These devices are radio-responsive: in other words, when a radio pulse is sent to them, they send back whatever the customer's account number is. There are businesses that want to put RFID slips in the packaging of products in grocery stores: this way, customers will be able to get through the checkout faster. If there is an RFID device in every package, the customer simply goes through the checkout and the machine will read the radio information of everything in his or her basket and then calculate the total amount owed.

What would happen if a refrigerator had the same ability to read RFID packages as supermarket checkouts? If it could read these packages it would know what it had inside it, and while you were away at school, for example, or at work, the refrigerator would be able to search the Net for recipes using materials that it had inside it. As a result, when you came home the refrigerator's liquid-crystal display would have a list of suggested dinner menus for you. Some people have suggested that this is not going far enough: the refrigerator should also automatically cook the recipe for you !

Imagine also that you are away on holiday. You get an email from your refrigerator saying: "Hello, I don't know how much milk is left, but you put it in three weeks ago, and it's going to crawl out on its own if you don't take it out soon." Or maybe you are shopping at the grocery store, and your mobile goes off. It is your refrigerator again, this time saying: "Don't forget the marinara sauce. I have everything else I need for the spaghetti dinner tonight " .

On this theme I am sorry to have to tell you that the Japanese have spoilt things for us by inventing an Internet-enabled pair of bathroom scales. The idea is that when you step on the scales your weight gets sent to your doctor, becoming part of your medical record. This is probably OK, but what happens if the refrigerator gets the same information? This might mean that when you come home in the evening you only see diet recipes coming up on the display. Or maybe the machine will just refuse to open, because it knows that you are supposed to be on a diet.

One last thing I want to mention to you as a possibility for the future is Internet-enabled clothing. This is not a joke: when astronauts are sent into space they wear clothes that have sensors in them to keep track of their vital signs. This same technology could be extended to earthbound medical conditions, where it is important for doctors to be able to monitor the patient. Instead of the patient being tied to his or her hospital bed with wires, that patient could be moving around in the normal way, the information necessary being captured from the clothing he or she is wearing and transmitted, under suitable protection, to the doctor or hospital.

I sometimes think about what the world would be like with Internet-enabled socks. The first thing I would do is to interrogate my sock drawer. I could send a question to the sock drawer saying: "Please report" and it would send back a report saying: "There are 17 matched pairs of socks, except one sock is missing: 144L is not in the drawer." I could then send a radio multicast around the house and get a response from sock 144L: "Hello, this is sock 144L. I'm under the sofa in the living room."

We have now solved the problem of the missing sock: an enormous contribution to society .

The Interplanetary Internet

At this point it is only fair to ask what else will be at our finger tips, now that it is possible to have the contents of the world's libraries online and in searchable form. Here is another piece of speculation that comes from straightforward engineering and has to do with the expansion of the Internet across the solar system. Many people laugh when they hear me introduce this idea, thinking that it is just a joke. However, it is not a joke: it is actually a project I started at the Jet Propulsion Laboratory in the US six years ago. It is a serious piece of engineering, because we need standards for space communication for the same reasons that we needed standards to create the Internet in the first place. I want first to acknowledge the other engineers who have been working on this project with me, and then to give you an idea of what is currently happening.

You probably know that over the last four years we have been exploring the solar system by sending robotic devices out into orbit around the planets. The deep-space network antennae used to communicate with these devices are 70 meters wide and can communicate with spacecraft that are either flying past the planets or are in orbit around them. There are three such installations of deep-space network antennae: one in California , one in Spain , and one in Australia . As the earth rotates, these antennae look out into the solar system, the three locations being arranged such that at a given time at least one of the three can see the spacecraft we need to communicate with. What you may not know is that in the past every space mission tended to have a communications system tailored to the instrument package on board the spacecraft. These communications packages were not standardized, and when a new mission was launched it frequently could not reuse the communications assets of earlier missions. Our first target, therefore, was how to standardize communication, such that when a new mission was launched any existing communications capabilities out there could be used to help support that mission.

The planet Mars has been a focus of attention for some time now. The Viking system landed on Mars in 1976, and more recently, in 1994, the Pathfinder mission sent back quite a few very interesting pictures of Mars. More recently still, two rovers have landed on Mars, and these are still functioning well beyond their anticipated lifetimes, producing enormous amounts of information. Some of this is stunning, because it has been shown that there was once liquid water on Mars, which is very significant because it means there might at one time have been life on the planet. Perhaps there still is life on Mars in some parts, only it is buried underneath the sand where the water has turned to ice. Perhaps Mars could sustain life, and this would be very exciting if we ever sent astronauts to Mars.

What we are doing in the course of our work today is to design an Interplanetary Internet, really a network of Internets, since we discovered very quickly that the standard TCP/IP protocols that make the earthbound Internet work do not work in outer space because of the distances involved. The problem is that between Earth and Mars, for example, the delay due to the speed of light ranges from 5 minutes to 20 minutes depending on where Earth and Mars are in their respective orbits. This means that it takes between 10 and 40 minutes for information sent from Mars to reach Earth. TCP is a protocol designed to do what is called "flow control", among other things. This means that when you cannot take any more data, you say: "Stop, I can't take any more data." On Earth, it takes no more than a few hundred milliseconds for a stop request sent out from one machine to reach another, and that is quick enough to ensure that no more date flows until the receiving machine has had time to adjust and can take more.

However, if the machines are 20 minutes away at the speed of light from each other, and one says: "stop I can't take any more", the machine on Mars will still transmit data for 20 more minutes. Where will that data be put? It is raining down from the sky, and there is nowhere to put it. This means that the basic TCP/IP protocol does not work for interplanetary communication, and we have had to invent a whole new set of protocols. In future, on the surface of planets and in spacecraft we will use standard TCP, but between planets we will use a deep-space network providing long-haul communication, and we have invented a new set of protocols that will allow the long-haul system to work properly .

To give you a sense of how we have had to rethink our normal intuitions about protocol design for interplanetary communication, think for a moment about what it means to communicate with someone on another planet. One problem is that the two planets are rotating, so if you do not send your information out at the right moment by the time radio signals sent from one planet reach the destination planet the receiver has rotated to the other side and you cannot talk to it. Imagine trying to throw a baseball to someone on a satellite orbiting Mars: if you do not throw it at just the right moment the ball will fly by before your receiver gets to the point where he can catch it.

This problem caused us no end of trouble, and we had to invent a system a lot more like e-mail than anything else. When you send an e-mail message to someone else, you do not know if that person is on the Net at the same time as you are, and you do not really care because your mail is stored and forwarded and then held until the recipient is ready to receive it. In the new programme, "the system for interplanetary communication", the same idea is used. The full stack of protocols for the interplanetary Internet we hope will be on board a 2010 orbiting satellite designed specifically for telecommunication support on Mars. Thus, by the end of this decade we hope to have an interplanetary Internet in operation.

What does all this have to do with libraries? That is simple: libraries will be the source of the enormous amounts of important scientific data we want very much to make available to all the scientists of the world. To be available it has to be in a digital library in a form that we can find again, and in a form that is sufficiently self-describing; that is to say, it has to be in a form that in 10, 15, 20, or even a 100 years from now will be understandable to the scientists of the day in case they want to reanalyze it. More that once we have found information that we did not really understand until 10, 20 or 30 years after its collection, and we needed to go back to the original source data to reanalyze it to discover what it had to tell us.

This is just some of what the next ten years or so holds for us on the Internet. Beyond that, you may have the answer, and I probably do not So, I am waiting to hear what your ideas are.

Thank you .