David Clark

Jonathan Zittrain

On the Issue of Domain Names...


This is an industry in which Bill Gates said 64K of memory should be enough--or just under 40K should be enough for anyone.
It's funny, because in 1978 we fought about the format of those addresses, and we just had this fight again, as part of defining IPV 6--and we are going to run out of IPV for addresses, there's no doubt. We have run out--
...
That gives you a hint that there was an IP version 1, 2, and 3 that we never saw, that we don't even care about. [laughter]
The first version was like 1 through 10, 10 scientists. [laughter]
You know, there was a network at MIT which was very popular which had 8 bytes for net and 8 bytes for post. We got a long way with that. But we had a fight with IPV 6, and I should say we've made the addresses bigger. The addresses now are 32 bytes and we've made them 128 bytes, which if you stop and think about it, is somewhat larger than the number of, I think, molecules in the universe...
So if you were to assign every molecule in the universe an IP address...?
Right. But then of course that's not the way we use it. We impose a lot of structure on it, which instantly wastes many bytes. In 1978 we said, "You know, you can make the addresses variable lengths, and then we wouldn't have to argue about how long they were." And some of the people said, "I cannot imagine writing a computer program that's sophisticated enough to deal with variable length addresses." And other things like that, which are major policy decisions ...
Okay, so now one thing to notice is we have the word "we" there. "We" got together and here "we" are designing version 6 of the IP address, and this is the stuff that we are going to be living with two versions from now. We're already living with version 4, and something to explore in the very near future will be--who is this "we"? Who owns the horses that are being traded? He says, "Come on, you couldn't handle variable length addresses," and he says, "No, no, we certainly can." Who are these people? That's a question that we'll soon be exploring.
   All right, so you've got these numbers. Now it's clear that if you wanted to get to one place on the Net from where you are, the numbers are not that useful. As easy as it may be not to have to say 274 billion, 362 million, etc., even a series of four numbers is kind of clumsy to remember. So we have these names, we retain the dot, and we retain some sense of structure going backwards, starting I guess from the right and going left: you have the .com domain, the .edu domain, the .net domain, and then you go back and you have harvard.edu, and then maybe you have law.harvard.edu, and perhaps any name you come up with gets mapped somehow to one of these four numbers so that if you type the name into your computer, the computer figures out what number it should actually be visiting in order to properly, as you might say, resolve the name.
   Okay. So now we have .edu, .com. At some point people sat down and said, "How about .edu? Why wasn't it .education or .school?
Let me give you just a little bit of history, because we always had names, but in the beginning it was simply a table which was stored on a computer and everybody loaded it down. So there was a machine at MIT, a machine we did our research on, and it was called Multix. So there was a table, which we always like--we wrote it on the wall, you know, Multix, and the address was 18.something ... (inaudible). And more about the mid-80s we recognized that this didn't scale--that's a classic computer science speak--we said this doesn't scale.
   And we had a few hundred hosts. It was easy. You could have a table on your computer and any day you might load a new one down, but someday, we saw, we might have a few thousand hosts and it just wouldn't work, so somebody said that we needed to do something. The standard computer science response when something gets too big is to make it into a hierarchy, which is one of the only powerful organizing principles that really does scale.
   And so we said, "Well, why don't we break the names into pieces and we'll have a name at the top and that'll lead you to a substructure, which will lead to another substructure, and you can have lots and lots of names. And the whole idea was to diversify so that MIT would only keep MIT's names, because the computer at MIT, and a computer at Harvard, it keeps track of names like law, and there's a computer at MIT that keeps track of names like LCS, and then there's a computer someplace that keeps track of MIT and Harvard, and that way we have this hierarchy both of names and the physical computers, and the whole point about the names, it actually lets you find the physical computers.
   And I knew we were in trouble--it was about 1985 or 1986--when we had the idea of a hierarchy, and we were ready to explain this to people, and all of a sudden the war broke out, this is what the name should be. And all of a sudden I realized we had made a terrible mistake, because people were assuming that the names had meaning. It's a very human trait to think that names should have meanings; people always look for meaning. And I said, "No, they should be meaningless names." And they said, "What do you mean?" And I said, "Well, ethernet numbers don't have meaning. The purpose of this hierarchy was to help you find computers, so the names should just be the names of the computers."
   And so one of the names could be Fred, and what that means is that there is a server that's run by Fred, and if you think he runs a very reliable name server that never crashes, then we should set Harvard inside Fred. And there was this group of people that said, "No, no, this is our culture, we have to because the names have to have meanings."
So now you're saying that by your preference, instead of harvard.edu, you'd have fred.glom, and people would know that if they wanted to do something at Harvard, they should go to fred.glom.
Here's the argument: do people actually have to guess the names, remember the names? We started out assuming that people wouldn't even have to remember the names, because I'd send you a piece of mail--you assume that we'd get started in sort of a boot-strapping activity. I'd send you a piece of mail; my name would be in it. Never mind how Adam sent the first piece of mail to Eve. I'd send you a piece of mail, and I'd have a little program in my computer which would suck up your e-mail address, and it would associate your name as a human with this e-mail address, and I would never actually see the e-mail address, it would just say Jon Zittrain.
So it's like a bookmark.
Like a bookmark, exactly. And what's funny is how you recapitulate history, because compared to the strings we use to send mail--you know, ddc and lcs, URLs are just ugly. We had the same argument--I said to Tim Burnsley, "URLs are ugly," and he said, "Humans will never see them." I said, "What do you mean?" He said, "Well, they're just links inside pages that help you tell one page from another, and so you never actually see the URL, you just see the text that's been put down on top of it that is its indicator." And he said, "It's a self-contained world." He could not envision URLs on soda cans or on the sides of buses, or the rest of the world or a TV ad actually trying to name the Web, he just didn't see to that point, so it never occurred to him that the names would be outside.
   And it was the same with the mail names. Initially people thought you'd never see them. You know, the idea that I would print it on my business card and give it to you was a very odd thought. I mean, people just didn't see the world reaching out beyond the bounds; they saw it as a self-contained system.
   So then people began to say, "No, no, we're going to see them, we're going to put them on business cards," and then they said, "So then they have to be short, somewhat easy to remember," and that's what leads you to things like .com, .edu, because you want them short, you don't want to have to type the string "education." They're actually trying to be nice to your fingers. It was very much a human factors issue that they got shortened like that.
So now at some point people say, "All right, let's make them short and easy to remember." Clark loses his argument, and they go ahead, and you have harvard.edu. And somehow when you get to harvard.edu, we know to send that to Fred, the computer down at the science center, and Fred will worry about parceling out law.harvard.edu and FAC and everything else that are the sub-domains. But we know at least as a first instance, when somebody wants to go to harvard.edu, they should go to Fred. Now how do they know to go to Fred when they type in harvard.edu?
Well, the way the DNS works is that every computer--it's a tree of physical computers as well as a tree of names--has a name associated with it. So there is a computer at Harvard, never mind what its actual name is...
That's a secret.
Right. Well, actually its name, if you want to know the truth--
NS1.harvard.--
Or something like that, but we've put an alias on it, which is harvard.edu. But any time you configure a computer, one of the things you have to enter is the address of your local DNS server. And whenever you want to look something up, you send a message to the DNS server that's yours and you say, "Here's the name. Help." And the DNS either knows the complete answer because somebody already asked the question and it remembers, or else it says, "Well, I've never heard of law.harvard.edu." Let's say I'm up at MIT, and it says, "I've never heard of law.harvard.edu, let's try harvard.edu. Nope, I've never heard of harvard.edu. I've heard of edu." So it sends a message to the computer that's edu and says, "Tell me where Harvard is," and it gets back a 32-byte address, and it sends a message to Harvard saying, "Where's law?" And it gets back a message saying, "Here's the address," and now it sends an IP message to that address, saying, "I found law.harvard.edu." And as you walk down the tree by translating one step of the string at a time, you interpret an address.
So now somewhere there's a computer even bigger than Harvard's, at least in terms of the size of its database, perhaps, that has all of the edus listed, that says, "If you're looking for an edu, come ask me, and I won't steer you wrong. I'm going to take you to harvard.edu when you ask for it, and tell you what the number is for it, and I'm going to take you to mit.edu when you ask for that..." Now how did they decide initially who was going to handle all the edus? Was it kind of like "not it" and then somebody had to do it?
Well, at the time this was happening, the Internet was still very much a government-run institution, and so it was natural to turn to our government funding sponsors, who were ... (inaudible) and more specifically the National Science Foundation, and say, "Do something." And it was pretty clear what you ought to do, because the person who has been making all of these judgmental administrative decisions for the Internet since its beginning is Jon Postel, who is the embodiment of something more formally called the Internet Assigned Numbers Authority, IANA. People are sometimes amused--"Oh, it must be this grand thing called IANA"--when they want to get a number, and in fact Jon is this guy with long hair and a beard and sandals who lives overlooking Marina del Rey, and you call him on the telephone and he says, "Well, you want some numbers, yeah." That's the Internet Associated Numbers Authority. It is as public a service and as non-profit as you can imagine, it's just this guy.
   So they basically ... (inaudible) a contract to ISI that said, "Well, why don't you run a computer--" they had this big computer in the basement, "And why don't you put all of edu and--"
Now ISI is the parent of IN, the Internet Services Institute.
No, Information Sciences Institute. It's a computer research lab associated with the University of Southern California, just the way that SCS is a computer lab associated with MIT. So they got a contract through the National Science Foundation to get the computer, and they put all of edu and all of com and all of org on it. And then of course there are all the countries--you didn't mention the countries.
Well, we're going to get to that. But now, say you came up to Jon Postel and said, "So anyway, we want to register mit.edu. That isn't taken yet, right? Can we route that to our Fred?" What would he say?
Jon exercised, at least at the time, a non-negotiable power of judgement, and he would say no.
So we have a king. We have an authority that says, "What are you talking about? You're not MIT, you don't get that, you are not the MIT."
That's right, you want to be MIT, you can be...(inaudible).
And if they were the Harvard University of St. Louis, a very little known liberal arts college--
He would have made a judgement.
He would have said--
He would have said, "I don't think so."
You can be "harvard of st. louis.edu."
Yeah, "harvard.stl," that's right.
Okay, on the edus it's simple enough. I guess somebody else had to do the same with .com, right?
Yeah. Initially Jon did .com too. He did everything until it got too big.
And he said, "Oh, this is too much work, you're not paying me enough."
Well, actually this is too much work.
So they found somebody else to start maintaining the tables.
That's right. NSF decided to give a contract to somebody who was more organized and could run big computers and so forth, and they did a competitive solicitation for what is called the InterNIC, Network Information Center, and I guess we're jumping directly to the punch line here, right? So now we get to NSI, whatever it is.
Network Solutions Inc.
Network Solutions Inc., that was the successful bidder on this contract to take over the operation of the .com domain.
Now who was on that side? Did they have another business? What did they do before they did this?
I don't know, do you?
I have no idea. It's this sort of shadowy organization.
I don't know, I've forgotten, maybe I never knew.
So somehow some people got together and said, "Okay--"
"We're a commercial company, and we can do this better than this guy with a beard and Birkenstocks."
"So we're going to do it."
Right.
So now if you wanted to get a domain, you wanted to start--I don't know, you wanted to register hot news.com, you would call up NSI, and you would say, "Does anybody have that yet?" And they'd say, "No." You'd say, "Can I have it?" And when they exercise judgment on your request, they'll say, "Sure, you want it, you got it."
By and large. Of course, you're getting to the root of the problem here, which is they suddenly realized that they were scared to exercise judgement. In the beginning, yeah, it was first-come, first-serve. First-come, first-serve was a very straightforward role.
And if somebody came up with a list of, say, a hundred domains, they'd say, "Well, it's going to take a while, but all right, by next Tuesday we'll have them ready."
I don't actually know what the relationship was. Technically speaking, the NSI was an operational organization. It was supposed to be carrying out the instructions of the IANA, so technically--but of course we're all fighting about this because nobody can find the contracts, and that's part of the problem. Well, right now of course you understand everybody's running for cover. I mean, we haven't gotten to the end of the story, but what's happening now is everybody can find a rock to crawl under and they're all looking for justification and covering their posteriors, and so what's going on now is you can't even rummage through file cabinets. Down in Washington, D.C. If they find the cooperative agreements from the National Science Foundation that actually explain who was supposed to do it, because it's all in fact unsubstantiated by any paperwork anybody can find.
Which is interesting, because those contracts weren't written with an eye toward the problems that exist now anyway, so it's basically going to just be random language that we apply to this problem.
Right. But ... (inaudible) had a very straightforward strategy for dealing with these contracts, and NSF, when it could, did what they called cooperative agreements, and the definition of a cooperative agreement is that you sign a piece of paper that says nothing in order to get around sole source procurement, because you have to do a competitive bid. And so you do a solicitation, and then they negotiate a cooperative agreement, which means, "Okay, you win and I will go work out what you're supposed to do." And so it's not in the contract, what you're actually supposed to do.
   If you sat down with your program manager and said, "Okay, what I really think I can do is--well, I can't get the rocket to the moon, but I can throw a softball," and they said--whatever it is, it's not in the contract, so nobody can find it.
So there was a bunch of stuff done with a handshake that basically defines the shape, the structure that we're still living with today of domain names, and who has them.
The structure is actually very formally documented, the fact that it's dot-com, dot-this, dot-that. That's very formally documented. But the question of who administers--and it's actually very clear who administers--but by what authority and within what scope, that is not clear. So technically the rules in principle, as I believe they were stated, were that if NSI were faced with a serious matter of judgment, they were to turn to the IANA for guidance.
Which means Jon Postel.
Right.
They return to the stage.
That's right. So if Joe Gabondik showed up and said, "I'd like to have McDonald's," or Disney, where it's clear that this is one step away from touching the third rail, clearly NSI was supposed to be able to say, "Excuse me, I must go in the other room and call Jon Postel," so that they get the judgment. But in fact they regarded the registration process pretty autonomously.
Okay, so they've got all these addresses people are trying to register with NSI. When they register, initially they don't pay anything, is that right?
I've forgotten what the rules initially were. I think initially they did it for free, but then a fee was put in.
A very low fee.
$50.
__:Is that you?
__:Yeah, do you mind if I put that down?
__:Wait a second, there's three things here.
See, he's actually taking notes, which means I'm going to be on record.
This is like one of these agreements you can't lose. He's got a file cabinet somewhere with all the agreements.
That's right. I guess what I should say is I think this is right, but as I said, the history is a little fuzzy for me. So go on, we can get still get to the fun part, even though I don't quite remember all the details.
We have some of these addresses, and the contract that NSI has says, "All right, you guys are going to handle .com, .edu, .org, and any of these other following top level domains, tlds." Now what if I come on and I say, "Geez, all the good things are taken now. Why don't we start a .bus domain? I'm going to create a new company, Computer Solutions Inc., and I'm going to start maintaining a list of people who give me numbers and map something.bus to those numbers, and I'll charge you $100 to register, and then people can just connect their computers to my server, and if they're interested in finding out where something.bus is, I'll tell them." Is there anything to stop me from doing that?
Well, at the top of the tree, of course, is a computer whose name you can't see. There's an old saying, which is the root of the tree is always magic. So obviously if you don't know where .edu is, or more to the point you see a string--I just pulled a business card out here, and the rightmost suffix is .es. Who knows what .es is? I bet you your computer doesn't either. So you have to go someplace to find out what this is, and this is called the top-level domain server, or the root servers. So how do you find the root servers? Well, the answer is every DNS server knows where the root is.
   So if I were to type this string in, I would pop all the way to the top, no doubt, because MIT probably hasn't sent an e-mail to anything .es for a long time, and it would go to one of the root servers, since the only way you can find it is by going into the domain ... (inaudible) aid system, which is where we make sure the addresses are known, and ... (inaudible) send replicated copies of the root server around the world, and you'd go there and you'd say, "Where is .es, where is .edu, where is .com?"
   Now, you could say, "Well, how do the root servers get that information?" and the answer is they all have to have the same information or it's not going to work at all. So there is a table, which again is a file which is maintained by Jon Postel, and the top-level domains are .com, .edu, .org, .go, .mil, and then the two-letter suffixes for all of the UN-recognized countries, which in fact we decided to let someone make the list of. It's nice to have a United Nations to make these decisions.
There's the "we" again.
Right. So the top-level domain table is maintained by Jon Postel. So if you want to do .bus--
I've got to talk Jon Postel into it.
You have two choices: one, you can just start a whole new naming structure which has nothing to do with mine, but then the question is how would your computer ever find it. The other is you go to Jon Postel and you say, "Can I have .bus?"
And he says--
No. [laughter]
I say, "I'll give you $1 million."
And he says, "I'm not moved by money. It doesn't make sense for the good of the Internet. I think the answer is no."
Now this is extraordinary. How many people knew this before today's class?
__: Well, what if the man dies? [laughter]
Does he have heirs ... (inaudible)? [laughter]
You see, this is why the contractual structure becomes so interesting, because it is in a very technical sense a memorandum of understanding which has possibly now expired between an institution of the government, the National Science Foundation, and a private sector, non-profit institution, the Information Sciences Institute. Technically speaking, the responsibility for carrying out the IANA is an institutional responsibility.
Whoever fills his sandals at the institute...
That's right. But we all know that--I mean, quite apart from the contractual structure, we are tremendously dependent on what has been twenty years of incredibly good judgment from Jon, and yes, if he fell in a hole or decided he wanted to take up Zen or something, I don't know what we would do. If it was ten years ago, ... (inaudible) gotten the way we could. I mean, a hundred of us then caucused and selected a new person, but now with the world being so commercial, if we lost the personal stature he brings to this, I honestly don't know what we would do. I mean, we would have to do something that's much more either convoluted, Machiavellian or public, and I don't know which one that would be.
   I once said, we had a war inside the--you know, this is a very personal community--we had a war and lynched our leaders because they made a mistake, and so we threw them all out. We had a real purge, and I was asked after that to give a calming talk, which was a little weird, and so I invented a saying in trying to describe our community, both its strengths and weaknesses, and I said, "We reject kings, presidents, and voting. We believe in rough consensus and a running code." Somebody made that into a T-shirt and printed out one thousand copies.
   It is true that any time anybody stands up and offers to leave, this particular community kills them first and then asks why, because if you want to leave, you must have self-interest, and therefore we don't trust you.
__:Say it again.
We reject kings, presidents, and voting. We believe in rough consensus and a running code," where rough consensus was deliberately ambiguous concerning whether the consensus and its final product were ... (inaudible) or the process was ... (inaudible). The answer was both are true. But whenever anybody stands up and says, "I want to leave," this community's reflexive reaction is to not trust them. And so it kills them in a minute. So Jon is one of the few people with stature, because he's been around so long and his credentials are so impeccable. He's never made a judgment that looked as if it smacked of self-interest or idiocy. And so, yeah, the issue of literally replacing him is a fascinating one. And I thought he was going to resign last year when he got sued.
There's clearly a story there.
Oh yeah, sure, by the guy who wanted to do .bus. The guy said .bus and Jon said no and he sued him.
And I thought that was my idea.
What, of suing him?
No.
No, .bus, literally .bus.
So now this might be a good juncture at which to ask who is "we?" Who are these people that are getting together and holding these lynchings and having purges and passing resolutions or making decisions that somehow are affecting our toasters?
Well, it started out with twelve people in a room.
Twelve angry men.
And they were convened by ARPA, which is the Advanced Research Projects Agency of the Department of Defense, which is why the Internet is thought to have military roots. That committee at the time was led by Vince Surf, and there were originally twelve but the group gradually grew as we added more researchers and so forth. Vince started this in 1974, I joined it about 1976, I guess, and the meeting started getting bigger and bigger as we had more interest.
   So we decided we needed to split off a smaller advisory group and then a larger ... (inaudible) thing, and the advisory group has changed its name four times. It's a deliberate strategy. If you want to hire a committee such that people don't find you and ask if they can join, the one thing you can do is rename it every two years so they can't find you. So we used to call it the Internet Configuration Control Board. We made that term up to make it sound as uninteresting and boring as possible so we could go meet in private without anybody noticing us. Then we called it the Internet Advisory Board, and then we had to change that because the government passed the Sunshine Law.
Now it's called the Ladies' Auxiliary [laughter].
So you can't give advice to the government or you're covered by FAC, so we changed it to the Internet Architecture Board. And I chaired that from about 1980 until 1989. Meanwhile we spun out a larger group, which is where they were debating the technical details that were supposed come into play, and that got too big, that got to be a hundred people, so we started doing what all computer scientists do, we imposed hierarchy on it. We broke it into a master group and then subgroups, and then we eventually broke the subgroups into groups, and that's the structure we have now.
You were about to say something, I can tell.
Well now, this week--we have a procedural story so far, but we started out as twelve people, and then we created--which was self-selected, engineers, computer scientists, and Americans? All Americans?
No.
Some Canadians?
No. Actually from the very beginning we had a Brit, a Norwegian and we've had--we were slow to pick up Asian participation, but we've had European participation from the very beginning.
Okay, so an international group of engineers, and you have meetings every so often, renaming yourselves, and now there'd be an agenda for meeting.
That's right.
A typical meeting with a busy agenda might be what, what would be up for discussion?
Whether we should do a DNS, whether or not we should go fight Sun because they proposed something called Network File System and wouldn't give us control over the standard or whether we should just publish an informational document and press on; what to do about the rising peril of OSI...
OSI.
The International Standards Organization, it was our competitor for the definition of the world domination of Internet protocol. [laughter]
... (inaudible) sort of a Trotskyist ... (inaudible).
Let's ask a counter-question. How does a board of directors renew itself? The answer is from within, unless the stockholders throw you out, right? How does a private foundation's board of directors renew itself when there are no stockholders? The answer is sometimes they get very ingrown. But we looked at ourselves as a board of directors, and we said, "As long as we can, what we're going to do is control our membership by electing our successors," because at the time there was no commercial interest and we viewed ourselves as somewhat equivalent to a non-profit--you know, a private foundation. Now we noticed we weren't incorporated, we weren't a partnership from a legal point of view, we were nothing.
And that's an issue.
Which I want to come back to.
And your mission, what was your mission?
Sustain the growth of the Internet, to make the Internet a success.
And if you came to a decision, what would happen? How would it be implemented?
We ruled without power. We had no power whatsoever. At the time I took this committee over, I had no power, so ARPA--or the guy who ran this, Vince Surf, who withdrew to go to MCI for the first time (he went twice, the first time to do MCI mail)--bestowed the title on me. He said, "As my parting I hereby bless you, I give you a title, you are the Internet architect, go forth and do something. I give you no ... (inaudible), no money, no contractual--do it by force of persuasion." So we would write documents and we would ... (inaudible), and maybe I would go sit on somebody, and that was it. We always said, "If you don't want to follow it, go somewhere else."
   The reason people let us get away with this was that we had no power, so all we could do was claim we had moral high ground, whatever it was, and act rationally. If we didn't act rationally--in fact, this is why I was talking about when we threw out the leaders, because it was, in fact, lucky I wasn't chairing at the time--I would have been thrown out; they did something that caused people to think they'd lost their minds. They just threw them out, they just said, "We won't follow you anymore. We follow only by our own will--we have no obligation to follow, we reject you as leaders. We won't listen to you anymore, go away."
If you show up at Denny's this week for the meeting, sit at another table.
We killed off the ... (inaudible). This is another "we," but, you know--
Now you had issued documents. These documents, because they came from you, and you had earlier issued documents that everybody tended to adopt, would have power to them.
That's right. It's one of these things where as long as you maintain competence and quality and respect, it's likely they'll believe you the next time.
Now Jon Postel, how does he figure in? You'd write documents and send them over to him, and say, "Hey, we figured out something for you?"
He was the editor of the document series. He had the authority to call something a document in the series. We had a number, they're called RFCs, request for comments, and he still controls the number. You can't publish an RFC unless he gives you a number.
Or you could, but everybody would say, "Who are you?"
It's not an RFC unless Jon puts his stamp on it. Everybody recognizes he's the RFC editor. And now we've formalized certain classes of documents. For the RFCs that are standards, we've got committees and subcommittees, and Jon basically makes sure the machinery runs right, but if you want to rent something, that's called an informational RFC, which is to say it doesn't propose to be a standard, and it just means standing up and explaining something--Jon looks at it and says, "Ye-e-e-a-ah."
These RFCs might be, for instance, a new standard for e-mail. Here's how, if you want to communicate to another machine and send it e-mail, destined to a person attached to that machine, this is the sequence you go through.
RFC 781. I'm sorry, 821.
RFC 821, that's in base ten.
Yeah.
__: Are we having fun yet? [laughter]
Now, RFC 821, you say it's a request for comments, but it's not really a request for comments at all.
Well, it was in its earlier versions, but by the time we got done, we decided we'd had all the comments, but we put it in the same series. I should say once every three years, we reopened the question of whether we should have two document series, one called request for comments, and the other called standards. And finally we said, "Why have two? Everybody understands." And of course nobody did, and eventually we had to create a standards line.
So there is a standards line now, even though RFC 821--
Actually all documents that come out have an RFC number on them, but ones that are standards have a second number, and they say STD something on them.
So disregard the fact that this says it's a request for comment, that it's really a standards--
Librarians like single document series, because it's easier to figure out how to file them when they come--and don't start five document series, because then we have to have five folders, so just do one.
And these documents people don't lose, they send these all around because you've got to know what the RFC says if you're going in the ... (inaudible).
They're on-line, they're free, you can find them thirty-seven places on the Web. By definition, the rights of the author state that they are freely propagatable, anybody can pick them up and do anything with them, there is no license associated with any of them. They're on the Net, and that's what you get.
Now, at some point there was a prospect, I suppose, of Jon Postel managing domain names for the entire world, is that right?
That's right. Well no, nobody was ever so insane as to think that, but clearly when we started out, until we delegated and found the proper authorities, he owned the whole thing. I mean, we understood, for example, that we had to give the country codes off to the countries as quickly as we could find the proper authority in each country--try that.
Now, he's still the guy that masters the top-level domain, the very last thing on the right.
That's right.
And at some point he said, "The burden of this is too great to worry about Harvard and Finland. What I want to do is have a series of top-level domains for countries."
At the very beginning, we recognized that he wasn't going to think about Harvard and so forth, that he wanted countries top, that's right. And so there's an interesting question, and one I cannot answer, which is why for the U.S.--see, initially we thought we were creating two separate domains. We were creating the .com, and .edu, and we thought of them as international, and we created all the countries as well, so you could have country names. And we thought of them more as being for small businesses and individuals. You can have a domain like concord.ma.us, and you can actually have--there's a friend of mine, I forget what his address is, 322border.concord.ma.us, and it's his house. We have a domain like that, nothing wrong with that.
So now, let's say you're a business in the UK. You might want to go onto the Web and you say, "Well geez, maybe I should have a .com address," or you might say, "Maybe I'll have a .uk address." If I want the .com address, I've got to go talk to Network Solutions, but it's a long distance call to Virginia.
Well, actually you send an e-mail.
And you say, "Okay, can I have this .com?"
That's right, there's a little form you fill out on-line.
Now, if I want .uk, who do I talk to?
Well, I don't literally remember.
But it's somebody in the UK.
The United Kingdom is well enough organized that in fact we could find the proper authority to designate who would take responsibility for the naming within the UK, and they actually created within the UK "ac" for academic and com, and so you have ac.uk as a perfectly common prefix. So they've recreated the hierarchy within their own country, and I don't remember who runs it, but it's knowable. If I took a half-hour I could find it, and I could go read it on the Web.
And it could be a government organization, it could be somehow answerable to the British government, or it might be twelve engineers from Britain who--
Most of the European countries are much more formal in their approach to things like this, and as we approach the--some of ... (inaudible) academics, because at the time we were doing this, it was still very much an academic institution, and some of the ... (inaudible) government institutions.
So now this brings us somewhat to the present, where we have a collection of domain names, some ending with country codes. They tend to be two letters.
That's right. There is, in fact, coming out--I forget whether the ITU or the OSI or somebody, but there is, in fact, an institution that believes that their job is to assign two-letter abbreviations for every country for the purpose of things like mail and ... (inaudible).
So what they do is, if there's a new country born out of a civil war or something, they finally notify the ITU and they tell the UN what color flag they have and everything.
And then they get assigned a two-letter suffix, and then we put that in the table and start looking for a recognizable authoritative location within that country to manage the top level.
And that goes through Jon Postel again, because he has the table that goes into the unnamed computer that's the root of all--
That's right. Now, you understand at this point Jon gets a lot of advice. I mean, Jon just wouldn't casually go off and say, "Well, let's go root around in Bosnia for somebody." At this point the Internet has gotten pretty visible, so it's not as if Jon acts unilaterally or without consultation. But in a formal sense, yeah, he still is the embodiment of the Internet Assigned Numbers Authority.
Now, what would it take to change this system? Does Congress just pass a law? Can the President write an executive order that says, "I hereby say that Jon Postel no longer holds the keys to the top-level domain server?"
Well, you're getting to the point where we have to try to understand the root causes of the current disaster. And one of them, of course, is that we have no model of government. And if you let me, I'll talk just a little bit about the history of an attempt to put a government structure in place for the Internet, and second, if Congress passed such a law, would anybody listen to them?
I don't know. It wouldn't be an RFC.
Well, this is a serious issue, because you should understand that most companies are not in .com.uk, or .com.es. They're in .com, and the reason for that is that when we took domain names--this is not "we"--when "they" took domain names which were previously invented for use on mail and so forth, and put them inside the URLs that identify Web sites--
"They" being the people who started doing Web ... (inaudible).
Tim Burner's league.
... (inaudible) of the Web.
Yes. He was the one who decided that he was going to use domain names to name the computers that are the hosts for Web sites.
He could have come up with some totally different scheme.
... (inaudible), but he said, "They already use domain name servers out there, so I'm just going to use them, too, I'm going to piggyback the system." All of a sudden the names had to become even easier to understand, and so Netscape put a little clooge in their browser, so if you type in McDonald's, it looks at that and says, "I don't see no dots," and so it tries putting www on the front and .com on the back. Now the instant that the Netscape browser did that, nobody wants a Web host name that doesn't end in .com, because Netscape can't help you find it.
   And so all of a sudden .com is incredibly international. So if Congress passed a law, they'd say, "US ... (inaudible), we don't listen to them, go away. We refuse--" You know, the countries might rise up and say, "We refuse to believe that the US Congress has standing to speak--" He could nullify Jon Postel's contract, and in fact, NSF has declared their intention not to renew the funding of IANA. What is Jon Postel doing? Very simple. He's passing the hat internationally and trying to get international support so that he can generate multinational credibility to continue his role. And if he can do that, then what would it matter if Congress passed a law? It's an international issue.
And he's still the shadowy figure that tells NSI when they have gone astray.
Well, NSI of course doesn't know who they have to listen to either, right? Because they don't understand who they're really responsible to, and if NSF cancels their contract or refuses to renew, then by what right do they continue to do anything? Well, you know, possession is nine points of--and so they'll continue to do it until somebody tells them to stop.
And the "it" that they're doing is maintaining the table, the grand table of .com addresses, what name goes to what number.
Right, because particular .coms were all numbers to be made. So let me just talk a little bit about governance of the Internet, because this is important. The IAB wasn't any legal concept. It wasn't a corporation or a partnership or anything, it was just a bunch of guys that got together and had a beer.
And this is before you chaired.
That's right. And we looked at each other through the 80s and into the early part of the 90s when I wasn't chairing it, but I was still hanging around and said, "You know, someday we're going to get sued because we're going to make some judgment and somebody's not going to like it, and they're going to kill us. So we need an umbrella to hide under. We need some corporate structure, we need some structure to center us." So we were looking around trying to find a structure, and of the two fathers of the Internet--we argue about who's the father and who's the first cousin--Vince Surf and Bob Cahn, Bob Cahn had gone off to found a little non-profit called the Corporation for National Research Initiatives, where he was trying to use private sector funding to achieve large research results, and Vince Surf, who had gone off to MCI, came back and was working with Bob at CNRI.
   And so we, in this case a very small set of people who included me, went to Vin and said, "Could we use the corporate structure of CNRI as a corporate structure for us to meet under so we have the liability limitations that would arise from living in a corporate context?" And he said, "I'd rather we start a new corporation to do that than use mine." So after a certain amount of hesitation, he started something called the Internet Society, and the original motivation of the Internet Society was to provide a corporate structure to do things like buy liability insurance and give us a framework in which we could begin to sort of rationalize this group of people, so they weren't just twelve guys.
   So they set up the Internet Society with a board of directors, and it's a 501, C30--you know, you know exactly what it is, because it's a type of non-profit that the US government understands, which of course instantly made it US, which instantly made it suspect overseas, so already they've got one strike against them because it wasn't an international corporation. What's an international corporation? I don't know. But why don't you incorporate Zimbabwe, Libya, Liberia, anywhere.
   But it was a well-defined, non-profit US corporation. They had a board of directors, a process for election in which the members elected them, so now you joined the Internet Society, it was a membership organization, and so the members elected the board of directors. The board of directors had advice and consent over the nominations to the IAB, so now there was a structure. Now the IAB was an instituted component of this, and so now we could claim that we had a corporate structure and at least liability insurance to sit under.
   And there's this other thing, which was that this meeting that used to have one hundred now, the last time it met, had 2000 people, which is the Internet ... (inaudible) Task Force, which is where all of the working groups occur, and there are about 60 working groups, and they're organized into what we call areas, and there are area directors who are volunteers, and they constitute a body that's called the Internet Engineering Steering Group, which is chaired by somebody who--and then he's an ex-official in the IAB, so it's all gotten very organized.
And these are all still engineers ... (inaudible).
No, no. I'm sorry. This is a technical body. These are not academics or researchers, they're mostly from corporations at this point, but they are techies, because this group is trying to write technical standards, and in fact when they try to address economics and policy, they usually fall flat on their nose, which is why part of the trouble is the DNS.
   Nonetheless, the Internet Society was created as a corporation whose purpose was to provide a framework for governance. Now, if you read the back of the T-shirt, what happened was as soon as it was constituted was that the community didn't trust it and tried to kill it, because it had arisen and declared its intention to lead. And so there was this incredible stress from the very beginning, when actually I said, "Look, we've got to have some framework." And the community said, "But these guys are self-aggrandizing, and all they want to do is participate in the glory that we built," and then I said, "All they want is reliability ... (inaudible)," and we had this real struggle about motivation.
   But now you had the Internet Society in place as an institution, looking at this deteriorating mess in which nobody understands what NSR's rights are as the contractor, and NSF expires. Does NSF simply have the privilege of recompeting, at which point NSR's right to allocate .com goes away, or do they have a monopoly, a perpetual monopoly right to allocate .com. What's going to happen if ... (inaudible) Jon Postel, and into this structure, where no one knows what's going on, the Internet Society stands up and says, "Well, I'll give it one more shot. Let's try leading."
Well, this is really interesting, because now we have a big mess. We've been sort of vaguely referring to but not yet throwing off the blanket and actually looking at whatever this big mess is.
There are three dimensions to the mess, and we've only gotten to one so far.
Okay, now what's the first dimension of the mess, as you see it?
Lack of a clear government structure and no base of authority.
Okay, number two.
Total lack of alignment between this name space and any other legal structure for validating the use of names, like trademarks.
Okay, and number three.
The name structure is not suitable for the purpose the ... (inaudible) are trying to make, to meet the needs of the people. It's not properly organized to meet the needs of people. That's a technical problem. A human factor's technical problem.
So to put it one level higher then, maybe, you've got something of great value all of a sudden in these .com domains in particular. This is a great conical study of the problem. These .com domains are of great value. They are also scarce. Once you give out the .com to one place, you can't give it out to another. The whole point is to have a mapping of--
... accidental ... (inaudible), but from McDonald's point of view it's incredibly real, and they're going to fight to the death to defend McDonald's.
And people now want these domains, and if other people have them and they can't get them, they maybe even feel entitled to them, a legal entitlement--
Absolutely, and so the question is who do you sue?
Luckily not you, because you've created so many ... (inaudible). [laughter]
But the person who's being sued now is NSI, and NSI is trying to go public. And if you read their prospectus where you have to list all the pending lawsuits at the back, these are not insignificant, and essentially they're being sued over issues of people who believe they have rights to names, not being allowed to use them.
Now, they go public, they're for profit, is that right?
Right.
They're supposed to turn a profit, and they would turn a profit because they have a good nobody else can offer.
NSF indirectly created a monopoly. They didn't mean to.
A monopoly that now wants to go public and start transacting business for profit.
Full tilt.
And in fact they could, at any moment, yank the domain away from one person, and say, "You've had enough now, let's give it to somebody else"?
I don't know the rules. They have at this point pretty clear rules. I haven't bothered to read them. I told you, I haven't paid attention to them. They have rules, and when you go in and register a man, you function according to the rules. The rules are pretty clear. I think as long as you continue to pay a fee, it's yours.
Of course they could change those rules.
Yeah. And of course somebody could go into court.
And say, "You shouldn't change those rules, you're not allowed."
Or somebody could go into court and sue them and get an injunction or a court order saying that you gave McDonald's to the wrong person, and you'd sue them in their jurisdiction, but what happens if you're--well, never mind.
Are there any questions around here, by the way?
_: What's the community--I heard a couple of times, the community said this or that, the community did this or that.
We've always viewed that the designers of the Internet are an open, self-selecting set of people, and the Internet engineering test works meetings are open to anybody, anybody can come. We try to keep the registration fee down, we don't change the names, the meetings are open, we announce them on the Net, and that group of people--especially those who have come enough that they feel that they can open their mouths without getting inflamed--are in some sense a body of the whole. We have an open house every Thursday night, and the area directors, the guys who are responsible for the thirteen areas, put on T-shirts with bulls' eyes on the front, and they go sit up there and anybody can say anything they want. ... (inaudible) on-line, or it's just you have to be--
No, it's on-line. Being abusive over the Net is not very effective because of the time delays, but sometimes those meetings are incredibly abusive. I mean, there are things that get said there in front of seven hundred people that I would not repeat here, because they're really just needlessly offensive.
So how would we tune into one of these meetings?
The next ITF is in December, it's in Washington, D.C., on the M-bone, which means you have to have M-bone tools--
How many people have the M-bone tools?
That's the multi-cast over ... (inaudible) that runs on top of the Internet. With a head run, we could probably figure out how to get that running in here. Go talk to Scott Braedner, because Scott is an area director, so Scott could help you get attached to the M-bone.
Then you get special tools so that you can really get a feed going.
Oh yeah, you can get bi-directional feeding, and you can sit in a room like this and you can watch one of these sessions, and they're all multi-cast on video across the Internet.
And if something really incenses you, you can type something--
No, you talk, this is all on video.
You talk, and at some point they say, "Yes, the computer in San Diego, go ahead."
That's right. We actually have floor control--never mind, you don't want to go into that whole--
... (inaudible) anybody can participate.
Anybody can participate, yes. ... (inaudible) somebody will say, "Does anybody on the Net have something to say?" And there's a delay and then somebody will say--
Now, if you're McDonald's--
... (inaudible).
Why wouldn't you say, "Hey, I'm going to stack that meeting. I'm going to get a bunch of engineers, in fact I'll pay some people to go in and sit through the meeting, and they'll each..." It's like flooding your senator or congressman with a bunch of form letters that say, "Please, please."
That's why, among other things, that the T-shirt says, "We reject voting." We've always proceeded on the sense of the meeting rather than a vote, so there's nothing to stack.
And if there's a bunch of people at the meeting that all appear to be in cahoots and not like everyone else--
Everybody stands up and screams and swears at them, uses profanity and abuses them.
So this is a model of news group--
This is news groups acted out ... (inaudible).
This is the norm of the old Internet, news groups as they used to be before the era of Spam. And now we're suggesting that there might be an era of Spam applied even to these meetings. As people try to come in, make a concerted effort.
I mean, there are guidelines, there are documents that are written, that are guidelines for these meetings, and they say, "You leave your corporate allegiance at the door and always wear your badge, so somebody can come up to you and look at it and say, 'Oh.'" I mean, we don't worry about McDonald's, we worry about Sysco, which is a corporate player that dominates this entire--and in fact, I would point out that the person who runs the ITF, the chairman of the ITF, is an employee of Sysco, and he was elected by the group. They weren't scared.
And Sysco never sends him a memo that says, "Now, be really nice."
We don't know, but he is held--you know, like Ceasar's wife--he is held under such scrutiny, that he has to be above reproach, and he is. Nobody's ever complained that Sysco used their market share to twist that meeting. What they do is very public. They often come in and say, "We have an idea, and we just want to tell you something, which is that we're not going to submit this for ITF approval. We do not wish to relinquish change control over this document, so we'd like you to publish it as an Internet RFC. We are not asking you to publish it as a standard, it's supposed to have information with big text at the beginning that says, 'This document is from Sysco and is being distributed to the community as a public service,' and you won't implement it, it's open, but it's not an ITF standard, it's a Sysco--" And we said, "Okay." It's the same thing we did with NSF ... (inaudible).
We'll just put in our library. People can look at it--
That's right, that's what we said. We do not lead with any power. So if somebody else comes in and wants to publish a document, I think it's wrong to repress them. We put them in the document series and take them out.
So now as these problems that we've identified with the domain name ... (inaudible) have started to bubble, a big meeting was held to try to come to some kind of resolution, from which a memorandum of understanding emerged. Tell us about that.
Well, I can tell you sort of about it, but the important thing to understand is this is not technical, and therefore it did not come before the ITEF. The ITEF does not like to discuss policy issues, and when it tries, it fails. It's just not--they're a bunch of engineers. They're just not intellectually organized to deal with issues of law or policy or economics. So the Internet Society stood up and said, "We will lead, we will try convening this meeting, we will try to get people here who represent all of the issues that represent the users and the government agencies, and in fact we'll try to go international," which meant they would try to establish the liaison with the ITU, "and we'll have this big meeting and we'll see if we can't work out ... (inaudible)." It wasn't just a meeting, it was a committee that met. They formed this ad hoc committee which met for a long time, and tried to formulate--
In Geneva.
Well, all over the place, but the Geneva thing is really a reflection of the fact that one of the sponsors of this was the ITU, because they were again--
The ITU, International Telecommunications Union, associated with the United Nations.
Associated with the United Nations and located in Geneva.
And they worry, among other things, about radio frequencies and if Canada's might overlap with the United States, or satellite applications.
Right. Again, there's an issue of governance here, which is fascinating. The ITU is older than the United Nations, because it's been around since Spectrum was important. And so when the United Nations formed, the ITU actually voluntarily went and asked if it could insert itself under the United Nations to find a governance of shelter to sit under. It's an issue of how you find governance when you're not in a sovereign state, it's fascinating. So the ITU voluntarily joined itself to the United Nations, and I presume might decide to unjoin itself some day, I don't know.
So then they have the big committee meetings.
Right, they have these ad hoc committee meetings.
And they hash out a compromise.
Right. And they would tell you, representing their interests, that all the parties were been invited to the table, that this process was open, that once they got started they created this committee. But the process was open and anybody could have had a fair shot at being on this committee, so that all the interests were represented, and at the end they presented their results.
And they just sort of released them onto the Net and said, "This is what we think should be."
That's right.
I'll just read something real quick, and tell me what you think. "Stop the Internet coup. Oppose the GTLD MOU, which is to say the generic top-level domain memorandum of understanding."
Right, that's the thing that came out of this ad hoc subcommittee.
"Oppose the Internet Society and IANA. Your vital interests in the Internet are in serious jeopardy. A group of self-appointed autocrats have declared themselves rulers of the Internet, without regard to international law, the stability of the Internet, and the rights of you and your organization. Join with the Association for Interactive Media and the Internet Congress to stop this coup. Their weapon is the generic top-level domain memorandum of understanding. This document is disguised as an innocent standards agreement regarding domain names. Make no mistake. It actually decides permanent control over the Internet to six tightly controlled non-representative organizations. If you sign the GTLD MOU, you will give up all of your rights to have any say on the structure and management of the Internet forever. There are no provisions for election, representation or input from consumers, business and governments."
So part of what you're looking at here is an absolutely naked power struggle.
Now who are the good guys?
I am an Internet old boy. I am as tarred as anybody. I'm a friend of Jon Postel's, I've been around since the mid-70s. In this circumstance, I am barely trustworthy, and probably not. In fact, I did not sign the MOU.
You were asked to?
Oh yeah. And the reasons are rather subtle.
And people--we could sign the MOU, if we wanted?
Oh yeah, absolutely. And the reason actually that I didn't sign it, which turns out not to be relevant, is that for a period of time, the Computer Science and Telecommunications Board, which is part of the National Academy of Sciences down in Washington, D.C., which I chair, was asked to review this. And I figured as chairman of the board, I ought not to sign the MOU if we were being asked to review it. And I started phoning people around and asking questions, trying to understand what the situation was, and something very peculiar happened, and a person inside the government who is part of the sponsorship at the Internet Congress un-asked us to do the review. So this is very Machiavellian, and what you're looking at there is nothing but a power play.
And what are the interests, who are the stakeholders in this power play? Who are these organizations? What interests do they represent?
Well, I honestly can't answer that. The governance of the Internet is going to be a huge issue. So if you want to make the leap in your head between the way we manage the domain name system and the entire question of how the Internet is ... (inaudible), then obviously that has to be--look, there are a lots of people who associate their ego with making money, and there are a lot of people who associate their ego with having power. And sometimes you can do both, but it's clear that there are a lot of people who don't, in fact, have a lot of wealth. They're not aspiring to be Bill Gates, but they ... (inaudible). And somewhere at the root of this governance tree is probably a role which implies a lot of power.
   And so people whose egos are associated, they're going to fight over it, and there may be no stake except that. One of the players in this area, for example, is Tony Rokowsky. Are we off the record here?
Sure. [laughter]
Tony Rokowsky is absolutely over the top in his objection to the MOU, and your private ... (inaudible), you say, "What's going on?" And it appears that Tony has a wire loose in his head and hates every organization that he ever worked for, because he always leaves them dissatisfied, but there have been one or two important organizations that Tony Rokowsky worked for before he did this. The first was the ITU and the second was when he was executive director of the Internet Society, from which he was removed.
Another coup.
No, he was fired.
It's a real organization. ... (inaudible) you're fired, pack up your desk and out.
He was advised to seek other work, it was that kind of firing. So when Tony Rokowsky stands up and says, "I abhor an allegiance between the Internet Society and the ITU," he's complaining very bitterly about two organizations that he no longer works for. So what do we make of this? He says, "I have been inside both of them, and they are categorically incapable of doing any good, besides which they asked me to leave." So what are we to make of this?
Okay, so now we have a very interesting procedural account of the domain name problem, and how various solutions might be suggested, the parties to them, some sense of how they might be adopted or not--this kind of osmosis process, people either go for it or they don't, or you get enough signatures on this MOU, maybe people can say it's soup yet. In a little while I think we should turn to actual possible solutions.
   For instance, a good paper for this course would be solve the domain name problem [laughter]. Which is not to say to solve the procedural component. That would be nice too. That could be a different paper. But you could also write a paper that says the problem is we have these scarce resources, these .com names, or a sample problem. Everybody wants a piece of it, some people want the same name, they argue about pre-existing legal rights and duties as to who gets the name, and there's a disconnect between that and the actual assignment of the name.
   So we could solve it. We could solve it with a legal regime that somehow connects the two, we could solve it with a change in code. And we've heard a hint of some ideas about changes in code. Part of the complaint, problem number three that you mentioned, was this is not meant to be an airplane, it's a helicopter and you can't fly it around this way. So you can imagine one paper possibly for the source being, okay, we should stop calling it an airplane, let's make it a helicopter, and here's what the plates would look like. Here's how to somehow have the code itself have a domain name, reflect trademark sensitivities and geographic locations more readily. I mean, the code is just what we make it. It's just what we publish as RFCs and then a dot. So you can imagine all sorts of ideas that might play with that. And if we have time today, we should even broach some possible ideas on that mark.
   I want to shift now to phase two, which is to give Professor Nesson a chance to present an idea he has that presumes to address some of the procedural problems, and perhaps come up with another stakeholder in this, another organization in which we can place faith, and maybe be able to address the problem more thoroughly than in the kind of fascinating ad hoc process that we've just had described. So with that, maybe I should turn it over to you, if you're ready.
Thank you very much, Jon. I'm going to put down the ... (inaudible) just so people can see the slides slightly better. So Dave, what I want to do is build up to a series of questions, but it's going to take me a little bit of background in order to get there. I'm basically going to skip into the middle of the story, and ask you to assume that there's background to it and thought behind it. It didn't just pop out of thin air, but nonetheless I think that the story is coherent enough for you to get.
   The story starts on September 10 down at a hotel in Copley Square, when I had the opportunity to make a presentation to the top-level strategy board of the United States Postal Service, basically Alan Cain and his management team. I appeared there and presented the slides that you're about to see, which involve a build-up to a set of proposals that I made to them, so that I titled this Pushing the Envelope with ... (inaudible), looking toward some sort of combined Harvard USPS venture.
   I said I would undertake a strategic analysis of the company, and started off by identifying their most significant competitive edge, which is that they're the only company that actually has a human being come to the door of every residence and business six days a week. This was a room that had Cain and a whole slew of vice-presidents in it, and when I said a totally unleveraged asset, they all looked at each other and nodded.
   They're non-profit, like Harvard, which I think lends a certain element of character, and they have a big history. [laughter] It's a history actually of pushing the frontier. They basically have carried the mail right behind the frontier right from the beginning, they actually have a quite solid tech base, they have a resident telephone company, their own satellite network, they have huge database expertise with the management of the space that they do handle. They're currently very profitable. They have been profitable for the last three years. This last year was the most profitable in their history, by a long way.
   They have some concerns. Three years ago, Alan Cain introduced incentive management, and that produced a big entrepreneurial spirit within the post office. People are just now beginning to notice that it's become considerably more efficient than it used to be. The experience of going into the Cambridge post office right now is actually a good one.
   But in addition to making them extremely profitable, it also skewed their mission toward the big customers and the big customer feeders, because that's where they can make the money, and they're now feeling deeply the fact that their real moral roots are with universal service, and they're worried that somehow something is happening to them. This is, of course, in the midst of their losing market share, getting picked off by Fed Ex and UPS.
   They've got a huge image problem, going postal, snail mail, part of the vocabulary. Snail mail on the information highway, it makes it sound even worse. Politically that translates to a real vulnerability in the Congress, and any initiative that they show gets wiped out by anybody who's got a different interest, and they're completely missing the boat in Cyberspace.
   So this led to a few observations about the general context in which the whole thing is operating. The address space is totally screwed up and getting worse, there are no global directories, no national directories. And then I gave them a series of proposals. Talked a little bit about broader national context, of the effort to wire up schools and libraries and generally wire up what was going on about them.
   Proposals--first proposal was that they establish a USA postal Cyberschool. In a sense that's what we've been doing here. Since I last saw you, I've become the director of the Berkman Center for Internet and Society here, chairman of the next Internet and Society conference, and what we're involved in is taking the Harvard content up into Cyberspace, basically building a Cyberschool. We actually at this point are combining what we know with the business school and the law school particularly. I think we know something, how to do it. For the post office the objectives would be to help them make their work force much more flexible. It would be a flexibility designed to lead to leveraging their asset, and it would offer a real promise on the social mission of the post office, which is the largest employer.
   Other proposals--take the objective for the human touch of the post office and confront the problem, that America's biggest problem right now is that most people aren't connected up, and that when you try, you buy a box and a modem and an ISP, and it doesn't work, and you call one of them and you listen to Muzak for twenty minutes and they tell you it's another one, and finally you go postal. Something is needed, and the possibility--it's a radical thought when you first think about it--but the possibility of actually finding help in the huge work force of the post office is at least a thought I put in front of them.
   I had a bunch of suggestions, which I won't review here, to deal with their image and advertising. I suggested that they perform some immediately useful Cyber services, like assembling directories, particularly an international directory of non-profits, and actually facilitate them, helping to link up--that was the next place where I got a really big nod in the room. They all said, "Gee, we could do that and nobody could possibly object."
What if you did the NGOs?
The NGOs is the one they thought you couldn't object to. They were more worried about the national. And then the big proposal at the end--assume your addressing function. All right, so that was the end of the presentation I made to them, and it's proceeded rapidly since then. I just returned from Washington from a second meeting that they had set up, and they are definitely deeply interested in this. They see in this a possible remake of the postal service, complete with the new motto, "pushing the envelope." They see the possibility of really motivating their work force, of delivering a social mission, of getting ready for the 21st century, of taking advantage of the incredible gift they've been given by the strike, where their two biggest competitors, UPS and Fed Ex, have both broken the bond with their customers, whereas the post office can still say, "We're the ones you can trust."
   So they're ready to go. I won't pursue this further. I just want to put the question to you. Is it a good idea to get the post office into address space?
You understand that this is not topic which we've discussed before. We are literally having a conversation ... (inaudible). [laughter]
Unlike the discussion we just had.
This is not a ... (inaudible). So in some sense, my reaction is, "Hmmm, I want to think about that," but let's talk for a little while, just so we can sort it out. Let me be the loyal opposition to the idea and see how you respond, okay? The post office is still a quasi-governmental organization. That gives it a benefit, which is that it's semi-non-profit, but we're struggling with these sort of semi-privatized things. The government at the present time does not have a good reputation for actually being able to get anything done, and the post office clearly has a reputation of being slightly stodgy. Their previous ventures into this area, where they tried to do electronic mail in various forms, have totally failed.
   So right now I think they would not be trusted in Cyberspace, and the government has a very strange wire loose in its collective brain right now, which to overstate it is, "Well, we actually can't do anything right, so the only way to get anything done is to leave it to private sector, because if we try to do it, we'll screw it up." This has gotten to the point where I had somebody in the military explaining to a bunch of regulatory economists that they had to deal with security in the national infrastructure against war because they couldn't do it.
   So the extremes of this are just boggling. But you actually think that the post office could gather the respect either within the government as an arm qualified to do this or outside the government?
Here's our strategy. Our strategy is to open the Cyber postal schools in demo in Boston and the District of Columbia, to be up with them by January, built on the Harvard Business School platform, taken to its second generation, which is now current--we're meeting tomorrow at our technology group. The thought is that you demo it as a school, the interface, you build it and you show it to people. It's a matter of great interest in the D.C. community: the postal service would be doing this, with the build out to all the postal workers--Web TV to everyone.
   And through your demonstration of your ability to do that very quickly with an interface that knocks your socks off and that everybody can use, coupled with a media campaign--that denigrates it in some sense--but a communication strategy that's built entirely around that. The idea is to take an initiative that the public would respond to very positively, and do it within the domain of the politicians.
I mean, the government is desperate for a way out of this situation now, and it's clear that their concern is that they've created this monopolist whose interests may not be controllable and may not be entirely beneficial.
The idea that this power is in the hands of a for-profit institution is, to me, utterly mind-boggling.
Right, and I think that's a reflection of the fact bluntly that when you go back to the National Science Foundation at the time these contracts were put in place, you had people who really did not understand economic issues at all, and they just didn't understand what they were doing. They didn't understand the implication, they thought they were solving a technical problem by tech transfer.
   So to solve the monopoly, you'd have to go in one of three directions--either you have to regulate the hell out of it, which is essentially to say either create a regulatory board or pull it back into the government. But both of those today are very suspicious actions in Washington. And the other is to find a way of competitively allocating the names in .com so that in fact we have one name space, but we don't have a monopoly control over the process. Because of the shape of the name space, that seems very hard to do, too.
Just briefly, what's the solution that's contained within that MOU?
Well, it's vague, because they say create more top-level domains, and the one kind of competition is that--I think there's a deep technical flow on the MOU. They say create multiple top-level domains, and then one of the forms of competition is that somebody can run .com and somebody can run .bus. And so if you don't like .com, go register in .bus, but that's an incredibly flawed ... (inaudible) reasoning, but if you're McDonald's and there's .com and there's .bus, you're going to go defend your name in both of them. So I've made your problem twice as hard. I haven't made anybody's problem easier, because I have to fight two administrations for the privacy of your name.
   Now you say, "No, no, .com and .bus was the wrong idea. We really should have had .fastfood and .radiator and .so forth," but then we're going to spend five years fighting over what those categories should be, and the only way out that anybody has proposed is that they ought to align with the trademark categories. Because, in fact, there are substructures inside trademarks, so you can have--if you are McDonald's of Cleveland that does radiator repair, you can, in fact, have that name.
United Van Lines, United Air Lines.
Yeah, you can have that. So unless you really carefully design the new generation of top-level domains, which is what took a year the last time and look what it got us--we haven't solved the problem because of the uniqueness.
Now suppose I say to you we're not taking it over to redesign the top-level domain names. Initially the objective is to generate national directories, to make the existing structure completely interoperable so that your telephone number, your mail number and your street address are all tabled, and if somebody wants to send a postcard to Zittrain at law.harvard.edu, it would be delivered, or to his telephone number or an e-mail to either of the street addresses, so that people feel use.
Because you're saying there's no reason you couldn't put zittrain.law.harvard.edu on a postcard, drop it in a mailbox, and have it arrive at my doorstep.
Bingo. Now you run into this government versus private industry, because clearly there are private organizations that believe they're going to get out there and compete to build white pages and yellow pages services, and their white pages services is going to have your phone number and your e-mail address and your postal address if you want it to. So the government would be setting itself up, this post office would be setting itself up as a direct competitor to private industry, and generally what happens there is private industry goes "pfff" and the government falls over. So it seems to me that the post office has an uphill fight to expand their charter to get into an area where private industry feels that it has the right to be.
Of course ... (inaudible) mail itself is that way. I guess it's just sort of grandfathered in, right?
Well, constitution. How would you embrace private industry, rather than fight it? How would you say, "We want to generate a national directory that links telephone numbers to e-mail addresses? We understand that the telephone companies are the masters of the phone directories. We'd like to joint venture with you. We're non-profit entrepreneurs, you're profit entrepreneurs. We can do partnerships."
But there are profit entrepreneurs out there who want to build databases that have e-mail addresses and telephone numbers, so the answer is why is this a rational thing for you to do when the private industry can do it?
Doesn't exist--no national directories, no international directories.
There are people trying to build them on-line today, trying to get all the information they can. I can get a CD today which has, for every unlisted telephone number--because it's derived from a telephone database--the name, address and telephone number of 80 or 100 million people. And that's on a CD, I can buy it today. And obviously that's going to be on-line, it is on-line, and people are trying to figure out how to get it ... (inaudible) and how to deal with issues of privacy, "I don't want my phone number," there are tremendous issues to sort out about how to build it. But there are people actively organized in building these services on-line today. So what is the leverage that the post office has to overcome the natural predilection of the government to get out of an area where private industry appears to be going in?
Well, I guess I'd say it was appeal to the fact that there's a huge need for leadership, and the community feels it. That is, people sense the mess in this space, and the idea that an organization that was trustworthy and that built credibility in a careful way in order to make moves in this area would be, in fact, welcome and could gain strength as it went.
   For example, suppose that part of the Cyberschool demo in D.C. was after the postal workers were all hooked up and connected with Web TV--Web TV, at least as I understand it, basically you've got a monitor and you've got a keyboard, and there's like real long wires. You don't have a CPU or a hard drive. There's real long wires out to somewhere, and you don't know.
It's just a self-contained Web browser, which works out of a box. An impressive product.
Impressive, but there's nothing that stops the post office from having the hard drive for the community.
No, there's nothing that stops it, but again, there are people who are in that business commercially, so you have to ... (inaudible). For example, the cable companies are trying to do that.
Trying, but they're nowhere close.
But they're ahead of the post office.
I don't know.
__: ... (inaudible) has it's own browser, is that what you're saying?
No, not their own browser.
You imagine pop servers, I think is what's being proposed. That there'd be an e-mail server that's at pop.usmail.gov, and you'd want to--
__: ... (inaudible) ISP for the ... (inaudible), is that what you're saying?
No, what I heard him say right then--well, there's several different ideas flowing--what I heard him say right then was that in fact the post office could be the developer and maintainer of a Web site for the community. Today if you go into a lot of small--you go out to Concord, Massachusetts, there's a little tussle going on because there's a question of whether the town government should run a Web page or our cabling system should run a Web page or the library should run a Web page or one of the non-profits should come in and be allowed to run a Web page, or a profit-making organization come in and set up a Web page.
   And all I'm saying is that this is a space in which there's active contending right now. We don't have an answer which is universal or which has been societally accepted. What I'm saying is if the post office comes in there as a player, they are one of many. It's not an empty field into which they can walk, so they have to prove their right to be there, and the thing I'm struggling with is the federal government tries to stay out of state and local business, they're actually backing away, they're closing the agriculture--you know, the extension services are all closing. The government is pulling back up at all levels, and they don't like to get into areas today where private industry is, because private industry beats the hell out of them. The lobbyists and the special interest groups go down there and say, "You're going to stop doing that now."
__: But don't you feel it's true that the government would go for something that seemed to be overwhelmingly acceptable to people? I guess what I'm thinking of is there are lots of different kinds of phone books out there. There's the phone book that NYNEX puts out, there's these little local competitor companies that will do it, index by phone number. There's the CD Roms, which I've had with national directories, and they are just never up-to-date. I'm more likely to go pick up a NYNEX phone book because I trust the information in there is going to be accurate. And I think a lot of people might say, "Wow, the post office is handling this." Whatever things that they think about the post office, they're going to think the post office is pretty far-reaching and likely pretty accurate.
So if you're thinking about the positives--and I agree with what you said--first of all, you're articulating I think this general marketing strategy, which is you don't go to Congress and try to persuade them to let the post office do this. You first create tremendous popular appeal so that, in fact, the people are asking for it, so that there's actually grassroots support for the idea, which is the counterbalance to the special interests, and you have to do that. The question is whether you can do it quickly enough to take some territory.
   I think that there are obviously advantages to having the government in this area, which is that they would be more trusted not to abuse the information. And clearly when these for-profit on-line databases start collecting information about your name, your address and your telephone number, one of the questions that's asked is, "Well, just how many different ways is somebody going to be allowed to sort this and try to use it for demographics and--
__:... or use in post offices, or ... (inaudible) every citizen in the country?
Here's the initial plan. On October 25--well, actually let's continue the story from where we just were. Tomorrow night we do a satellite hook-up with Australia, where they're having the Australia Internet conference, and we're linking our conference to theirs. And in November we're linking to the Internet conference in China. And both of those represent the opening of communication channels, and there's nothing that stops us from keeping going and building a network--the synergian network, and once the pattern is established, it can run pretty clear.
   October 25, when they open this library over here, we have five reunion classes back for the weekend, and we're announcing the opening of our center, and the fact that we're going up with courses in Cyberspace from our center in an environment, the technological environment, which is basically the second generation of the business school environment, which in its first generation, in a single year, moved the business school to the absolute front in terms of technology in business schools.
   What we're saying on the 25th is that we're going to offer a series of Cyber courses--not many--this is prototype, this is demo. Arthur Miller doing privacy in Cyberspace, Terry Fisher doing intellectual property in Cyberspace, Larry Lessig doing architecture of the whole, and two more which I haven't quite got in place yet. I was talking just before we came to class with Anne Marie Slaughter, who is my choice for one of those slots.
   The idea is that we describe those courses with a single page, which has a course description and an up close and personal with the teacher, and we're offering from the time of the announcement to some time in January to sign up. We are probably going to charge something. The original thought was we weren't going to charge anything, but people have been pretty convincing that charging something declares people's seriousness.
   So we're basically targeting a Harvard global alumni audience for this, and the basic model of the courses, which will start in January and basically go through a semester--something like 12 weeks--will be the master teacher leads with some kind of narrative that builds up to the point where you have an issue that's focused and described. Once the issue is formed, it's put out for discussion in groups that have been organized in 12 within the environment, made up of the people who've registered according to some sort of strategy with respect to their preferences. Those groups can actually be flexible.
   And those groups discuss, for example, generating the best statement they can make of what they think the issue is, or beyond the issue what they think the solution out to be. That's filtered up to a series of TAs who work for the professor, and the TAs have the job of vetting the material that comes in, seeing what's common about it, meeting on a once a week basis physically with the professor, discussing things. Hopefully they're TAs who have taken the professor's course, they know the professor's stuff, they know the professor's material. And on the basis of discussion with the professor, they've got the job of communicating back out, and along with the communication back out goes the professor's basically next question. It's secratic message in Cyberspace, that's the basic idea.
__:Apart from the value of having a Harvard professor ask you questions for which we're all paying lots of our money to show how serious we are about our degrees from Harvard Law School, I'm trying to differentiate in my mind what the difference would be between the experience a user would get and if this is a model for the product that the post office would be describing to connect people as its new marketing strategy with the already existing methods of interaction from news groups or other forms of communications and chat rooms that are already in existence.
The basic difference--let me just see if I can--that's what I was hunting for, and I can find something here that will give you a look at it. It's not what I had thought what I would show you, and I don't know--I guess this is it. This is a basic picture of the business school model. It's a model that's built on a database. Most of the Web pages that we're familiar with are static pages. You go to them, there they are. If somebody else goes, they get the same page.
   The business school model is built on a database, so that when you go in, you get dynamically created Web pages that are tailored to you, and that makes a huge difference. That basically means that you can model the process of a school in the environment. If you imagine the metaphor of school, you can play it out in the way this system works. So what they basically did was Kim Clark came in down there, they developed a set of principles--let me see if I can give you some of those.
__:What does this have to do with the name server issue and how you resolve Internet, whether they do the business school or the Cyber school in DC--I'm completely lost as ... (inaudible).
I appreciate that it's kind of hard to get your hands on. It's an integrated strategy to put the post office--from one point of view--to put the post office in a position where they could do this and be fully supported by the American people in doing it.
__:Do you want to rescue the ... (inaudible) or do you want to solve a smaller problem, which is simply how do you set up this naming architecture? If that's the goal, then I don't see how the post office is suited to do that. It may be something they try to do to turn their ship around, but it's very counter-intuitive to me that the post office, as probably a paradigm of inefficiency and not at all ... (inaudible) should be the one to solve the problem. It's also probably a regulatory problem, and maybe they can set up e-mail addresses for everyone in the world if people on the Web can't do it, but basically what we need is some kind of policy decision or regulatory intervention to figure out what this naming architecture is. So that's the problem, isn't it? So how is the post office--are you just going to vest them with power, and they can exercise that along with setting up Cyber schools ... (inaudible).
No, I'd actually like to model the power, I'd actually like to model it. I think that we'd like to have a try of building dispute resolution system that would solve the problems when they arise. So for example, let's just imagine that we take the next step, so that we have a general audience, not just a Harvard audience. And suppose that you can organize groups of 12 scalable, so there's no limit. And the model that I've given you, there's a problem of scale, because the teaching fellows are a limit on how many groups that you can run, and you need to come up with another process architecture for running other groups, which I won't go into at this moment, but they're available.
   For purposes of running a jury project, however, a dispute resolution project, as far as I can tell the model is infinitely scalable, so that here's the model for dispute resolution. We've got a dispute, we're in an Internet environment. Dave thinks it should go one way, I think it should go another. We've created a model where you tell narrative in some structured advocacy fashion, so he gets to build his Web site to tell his position and I get to build mine to tell mine, and we get to expose it to an audience that's built already of a zillion--as many as we can put together and accommodate groups of 12 jurors who are interested enough in this problem so that they'll participate and decide, they'll consider.
   And what comes back to us, then, is a whole lot of data about what the public reaction is, what the judgment of the jury of the Internet is between the two of us. Now, as a dispute resolving method that may not be mandatory. We could leave it then to negotiation. We could supply further modes of negotiation, but we then have a metric that's got some credibility to it. We've got a fair process that we can both agree to, we both had a fair shot, we both went out to an Internet audience that was fairly representative and constituted, and we get a judgment back. I think that we can work that as a dispute resolution mechanism.
Alice, you've been waiting to get in for a while.
Alice:I think the discussion that you're bringing up is--there is some relationship to the domain thing, and I think the relationship is it seems to be begging government to get involved, and that's when I was listening to you speak earlier, when you had these 12 people that were doing it--and you know, when you've got a little bitty project and not very many people, 12 people can handle it. But as it becomes bigger and bigger and bigger, it's begging for someone to get in and take over and take leadership. And the fight I think is whether it should be government or whether it should be private.
   And from my point of view, as you say the lawsuits are being filed, will the courts write the law or will the government write the law, or will the Congress write the law, or will the state legislatures, which I think also somebody's talked about--the people are asking for it, and that's what I see often times, is good legislation comes from the states and then finally when enough states start speaking up, the federal government says, "Yeah, I think we better do something." And so they've set the tone. But the courts are going to write the laws if the government doesn't step in, it strikes me, unless private says, "We're going to do it," and starts stepping forward and being more open.
By way of full context, by the way, Alice, is a state legislature.
See, I think that the domain name is the absolute--you've picked an example here of a place where all of the aspects provide and make this as bad as possible. I mean, this is a worst case. And it's international, because .com is international, and that's one of the problems you have to deal with. You know, if the US Post Office steps up and says we're going to run .com, it will be viewed by the other countries as, again, another example of the US government trying to reassert its ... (inaudible) with the Internet, or so you're likely to be killed from overseas. That's the other issue you have to deal with, because .com today is very international.
   But of course trademark is not international, there is no international forum there, so we're struggling there to solve the problem of how do you get an international resolution. In the long run we're going to have to face something. I mean, I came across it just the other day in an entirely different guise. Somebody's who's running something, which is sort of like a radio station on the Internet, approached ASCAP and BMI and said they want ... (inaudible) typical license that ASCAP and BMI provide for the playing of music across the radio station, and they said, "By the way, I hope you realize, we cover the whole world." And ASCAP said, "Would you please unsay that last sentence, because we'll give you a license if you don't say, but if you say it, we fall apart," because of course ... (inaudible) and all these other players, "and we don't have an international basis, but ... (inaudible) so just please don't tell us it's international." And so they unsaid the sentence and they got the license they wanted.
   But this question of--again, you talk about governments, when you get above the sovereign state, we're struggling. Now, if you look at what the White House just did about electronic commerce, this position they've now taken which is being articulated by every magazine. They're going to all the other governments and saying, "Look, we believe that there is some basis for common ... (inaudible) of using the Internet." They're talking about duty free zone and minimize the non-tariff barriers.
   In fact, if you look outside the United States, we're seeing a fair amount of leadership out of the government, but when you're inside the United States, you approach anybody at the federal level now, you're getting a tremendous, "We can't do it, we're going to fail." And we know the epitome of this in this area, of course, is the tele ... (inaudible), which is the government's best attempt--and if you've ever read--have you read the tele ... (inaudible)?
Alice:No, I don't think so.
Well, in a mechanical sense it's an incompetent law, that is to say--
Alice:When was it passed?
1996. The successive paragraphs contradict each other. It's not that I'm saying it's ideological, something you wouldn't like, I'm saying ... (inaudible) it's incompetent. And that's because special interest groups succeeded in getting paragraph stuff ... (inaudible) government, and of course then delegated to the FCC rights, and the FCC exercised these rights, and they were challenged, and the circuit court ruled the federal government doesn't have the right to govern in this area, that it's state's right, and all the decisions have to be relegated to the 50 PUCs, and this is now going to be fought out in the Supreme Court.
   So right now when you go to the government and say, "Take leadership in this area," they're so demoralized, because they have ... (inaudible) exampled that they can't pass a good law. When they try to act it out, it's shot down in the courts. But their willingness to step up inside the United States and take leadership is from my perspective--I don't know, it depends what era you grew up, whether you think the government is intrinsically good or evil, but I've lived through both of those phases.
   I still think the government is capable of doing something, and I look at what they're doing outside the United States, where I think ... (inaudible) has actually been going around and getting some support, and I look at how demoralized they are inside, and it's really disheartening because they're basically saying, "Well, we have to let the private industry do it because if we do it we'll screw it up." I mean, ... (inaudible) giving a talk on the president's strategy for--I mean, electronic commerce begins by saying that sentence. He says, "Private industry has to work this out, because if we try to do it we'll screw it up."
   So the respect structure in which we're going forward here is really devastating, and I think the post office has a strike against it, because a, it's national, so it's going to have problems when it tries to go international, and b, it's a part of the government so it's going to be afflicted with this governmental malaise that, of course, we're the government so we can't do anything, and of course, private industry, vulture-like, is descending in this area and saying, "The government is weak, and so if we can just push this ideology that says the only method that works is the free market forces, we can get the government to stop doing a lot of stuff."
   And so anytime they see something they ... (inaudible), they send a lobbyist into the government now ... (inaudible). As a recently exposed observer in Washington, I've come away somewhat--well, bemused is the wrong polarity, but--I've induced a strange expression on your face.
No, listen, you describe the real world. The space was born with a pure heart, it had Shareware in its soul. And around 1993 it turned into com, and there's a mission for our enterprise. The mission is to reclaim a share of Cyberspace for public interest.
Internet 2, the Wrath of COM.
And I know some other people who are trying to reclaim it too, and that's a private foundation.
But I think that we've got a competitive advantage.
Now, I want to leave three minutes at the end so that we can hear a turbo explanation--I know you're capable of it--of your solution for the domain name problem ... (inaudible). But before we do that, Seth, you've had your hand up for the last--
I don't have a complete answer, so it's going to be a very flawed answer.
This is great, this is rev one for the draft of an answer.
That's right. But there was another hand too, there was somebody over here.
__:We don't have time.
You can ask the question, I just may not have time to answer it.
__:Well, I just wanted to ask more about standard ... (inaudible) in general, how much is it ... (inaudible) industry, how much is people like Microsoft or Sun really your competitors now, ... (inaudible). Like ... (inaudible), it seems like they've sort of ... (inaudible) the initiative ... (inaudible) in some areas.
Well, the ITEF lives at one level, which has to do with the core infrastructure of the Internet. And our assumption has been we do not control--we don't try to control the layer at which application to build. What's really important to me is that open conduits, open standard conduits between two computers, and if Microsoft wants to construct a wall such that if you run a Microsoft computer, the aps on your computer don't work with the aps on somebody else's computer, I think the marketplace--I'm happy if the marketplace ... (inaudible) that out. In fact, Microsoft is taking two shots at trying to basically put the ... (inaudible) teachers into their browser, and the answer is they've died both times.
If Microsoft comes up with something like Net meeting, which is a Microsoft program, that if somebody else is running a copy of the program, over the Internet you can chat--
But Netscape, I take it, just by dint of doing it, has given tremendous privilege to the whole commercial domain.
Sure, by .com. But that's a syntactic issue, and you could go to them and you could say, "Can you take it through your ... (inaudible), we're having a problem here." The issue--very briefly, you want the turbo answer--the issue is that we have two conflicting needs for this name space, both of which we didn't recognize at the beginning, or the point is that the ... (inaudible) wasn't designed ... (inaudible). And one is that the names need to be short and easy to remember and unambiguous so that you can put them on buses and soda cans and you can then say, "Like our product? Try blah blah blah." ... (inaudible) Seattle.com, and it's got to be zippy and it's got to be market-oriented, and they also have got to be marketing names.
   And the other problem is that you pretty much names that are non-unique, because there is more than one McDonald's.
And this is the scarcity problem, and you need to solve this problem.
That's right, we have to solve the scarcity problem, and so a radiator manufacturer in Cleveland needs to be somehow allowed to have a name that contains McDonald's. Okay, so to step away from the domain name system and say, "Well, what could we design?" Okay, now here's an idea that doesn't work, but it's lurking in there, a glimmer of something, imagine that I went to you and said, "You can invent any name you want and it's got to have two components, and the first component is, you might say, your brand name." And then it says dot, and then after it you can put any text string, subject to the limitation that the text string is not trademarked and not likely to be trademarked. So I want an and trademark as the second string, so it could be fastfood or radiators, but it's not a brand name.
   So McDonald's can come along and say, "I want to be McDonald's.fastfood." Now, it isn't top-level domain. You can't have a hierarchy to resolve them anymore, because if there 100 million corporations, there are going to be 100 million entries in this table, so it's flat, it's just a flat table. So now there's a tremendous engineering problem, but engineers love this, so push it back.
   Now the question is, well, there's many places where the ambiguity will still arise because there will become popular suffixes. So "city" would be a popular suffix. Every city would like to have a Web page. Today we have a lot of marketing with boston.com, but in the long run, if we settled on this, it would probably be boston.citypage or something. And so ... (inaudible) about Springfield, where there's a Springfield in every city in the United States. And the answer is lurking in there, you want an implicit qualifier.
   And to me the right organization for the implicit--and implicit means you can't see it--there ought to be an implicit qualifier which ought to match the hierarchy of trademark and trading of allocation, which is to say you can do it with the state attorney general and get a name with the new ... (inaudible) for a business, or you can get a trademark at the national level, and eventually you'll have to talk about multi-nationals being able to go someplace and protect McDonald's worldwide.
   And that the search process by which you go from the name back to the most likely hit, so that on the average you get one answer, ought to be one that's respectful of where the query is coming from, and now I'm getting to the point where I don't quite know how to do it, but if you look up boston.citypage in Massachusetts, you shouldn't get the page frame ... (inaudible) unless you do something extra special. So do something extra special, and you ought to be able to see all of them.
So part of your solution is to say that from every browser, you won't necessarily get the same result. From every browser likely to see a bus bearing Seattle.com around here, it's going to go to Seattle Best Coffee, but if you're in Seattle--
See there the ambiguity is not there, but it might be the case that you would end up at a different Web page for a different name in a different trademark jurisdiction, and there ought to be some way such that when you do one of these things, it's not quite like a normal browser in which you just get a page, because what you'd like is you'd like it to show you what it thinks is the right answer, but there ought to be something over in the sidebar that says, "You want to see them all?" So that it's clear that if you didn't get to the right place, there's an easy way--but you see, humans are great are resolving ambiguous names. Humans are really good, they have brains, ... (inaudible).
   So the point is it's really nice to give a hint to a human, which is, I think what you want. What you really want is a way to say, "By the way, if it wasn't, turn your brain on. Here's the list, figure it out."
Now also, part of this proposal does suggest, then, that there might be a sense from the browser of where the person is. You would tell Netscape, "By the way, I'm in Massachusetts, so when I say Springfield, I want Springfield, Mass."
That's right. And I think there's a truth, which is there's this illusion that Cyberspace is this space that is--and some of the previous courses you've had, like Jon... (inaudible), and they come in and talk about this sort of Cyberspace as if it's--I'm not sure Cyberspace exists. I think it's a very thin veneer on top of the real world. And there are an awful lot of things you want to do which, in fact, have to do with not where you are in Cyberspace, but because human beings live on the face of the earth, I think the idea that you're involvement in Cyberspace would be improved if the system had some rough idea of where you are in real space, is in fact a good idea and not a heresy. But there I am ideologically--well, there are other positions, so people would like to see Cyberspace as this thing which is completely divorced from real space and doesn't even have real laws. I think it's all nonsense, but I must be getting old.
So now we have a glimmer of an idea.
If somebody wants to work on a paper in that area, I'll talk to you further, because I think it's an idea that's worth developing.
Realize too, if you actually do solve the domain name problem, and you give it to this guy, there's a good chance that Jon Postel could get his hands on it, and then your idea actually becomes what everybody lives by.
Set it up so it can be multi-valued, and then these problems go away. You know, if it isn't a winner take all game for the ownership of the word McDonald's or Disney--I mean, they're very aggressive, so you might die anyway--but in general, if it's not a winner take all, it just diffuses a lot of the energy out of the system. And I think that we really need to get into a situation where it isn't a winner take all, because that's what's generating the excessive amount of heat right now.
__:If it's not a winner take all, isn't it just like a hyped up browser where you enter preferences saying, "I live in Massachusetts, and I'd like you to put non-profit corporations first," or whatever would be ... (inaudible) answers, and there aren't really--maybe it goes back to having numbers as your unique address. I mean, you're taking away the text.
I guess I'm not sure that the question, isn't it just like a browser, actually gets at the root of the question. What we're trying to design is a system that actually satisfies human need, and in fact I would imagine that this would be packaged inside a browser, but I think that what you need is a codified structure. And if you can agree on the right structure, then you can get everybody to--let me give you another example--because Web pages are completely unstructured, which means that it's really hard for computers to ... (inaudible)--humans again use their eyeballs and their brain and they can search them.
   But there's some friends of mine on the west coast who are saying, "Look, instead of this completely unstructured description that you get on the Web page, why don't we come up with a framework," and the thing they started with was a framework for describing a restaurant, and they said, "You can have any Web page you want, but in addition there's this template and we want you to fill this in, and it's got information on it--what is your principal--are you a Chinese restaurant, are you an Italian restaurant, how big are you, do you have a smoking section, where are you located, describe your menu approximately."
   And what this does is let people go into a database of restaurants and ask a structured query, which is is there a restaurant within five miles of here that describes itself as quiet and serves Chinese food and has a table that serves 12 ... (inaudible) tonight. And it might come back with three answers of what would the answers be ... (inaudible), so quick, one, two, three.
   But you need that intermediating structure so that the providers and the subscribers and the producers can meet each other a little more easily. And if it meets the needs of the market, then that intermediate structure can come to replace--when people try to build those intermediate structures and it doesn't meet the needs of either the providers or the subscribers, then the ... (inaudible) will kill you, as in the guy who tried to come with the intermediating structures to make it possible to search all over the purveyors of CDs to tell you what was actually the cheapest on the Net, and since that didn't, in fact, meet the needs of the subscribers, cause they do not want CDs commoditized, they killed the guy.
   But if you can find that intermediate structure--so what I'm saying is the name space that I'm talking about is some sort of intermediating structure, and if you're good at designing it, both the people who want to offer the names and the people who want to search them will find it satisfactory, and the people will go there. And the Internet ... (inaudible) by any sort of method other than that. And if the answer is it's embodied in the browser, so much the better, because it makes it easier to deploy. If I could do this with a browser and six ... (inaudible) of Java, it means I could roll it out in two minutes, I'd like that.
__:Dave, can I program by Netscape to default to .edu?
I don't think you can change that suffix, I don't think you can do it. Now, if you gave Netscape a year's warning and a reason why, maybe you could, but today they think the answer is all the purveyors of ... (inaudible) .com and so that's what they're doing. If you want the .edu, you have to type it in yourself.
Let me give a few administrative notes before we finally wrap up. First, there hasn't been as much opportunity today to actually have you all jump in. I'd be delighted if you have thoughts, while they're still hot, even, coming out of here. Put them up on our site at the Harvard Nucleo Net site. We might even be able to convince our guests to stick around in the ... (inaudible) area and jump in and respond to them, and even at least discuss it among ourselves. There should be at least a couple of sessions during the year where you're be incensed or inspired enough to want to kind of put some thoughts together, and they need not be manifestos. In fact, when they are, they're a lot harder to read and harder to respond to. So brief thoughts, kind of little nuggets would be great, if you care to share them.
   Next week is shaping up to be tax week, and we've got some tax on the Internet proposals we're going to discuss. I know there are some LRB people here, Legislative Research Bureau people. If you can convene, maybe talk to Molly right here, would be the right person to talk to, and we can help get an interesting battle plan in place for next week.

Other than that, thank you all, thanks to both our guests for a very provocative participation.


Back to index of transcripts