Carl Hewitt: Maybe Humpty-Dumpty is right. When he uses a name, he knows exactly what he means. All names are local, and can name anything you want. You're allowed to like the name or not, adopt or not.

Mark Day: This is a version of the web of trust applied to names. Maybe we need the PGP model for names.

Carl Hewitt: But its more than just naming people.

Q: Wouldn't devices be genericized, such as "Mark Day's cell phone"? Some devices are not associated with a person, like a shared printer.

Mark Day: Ah, but what to name.  One way is to name capabilities -- It's hard to get universal registration but maybe you really want the object. Like a Mobile Phone which does text paging, I may want to send a text message to it specifically.

Rohit Khare: You're arguing there is a global namespace of devices.

Mark Day: As devices change over time, a universal registry won't work. It doesn't matter that it's my cell phone. I care that it can receive short messages. What if I have 2 such devices? You could maybe create a universal registry now, but it would be brittle -- what if we did this 15 years ago?  I think that setting up such a namespace would be futile.

Mike Mealling:  Isn't this device naming a red herring? We want to name services, and don't care what device they're mapped to.

Mark Day: This sounds better to me. By and large, systems have been established that will route messages, but, what if a sender wants to explicitly send to a specific device?

Mike Mealling: But, in that case they'll have local context.

Q (Bob Morgan?): What about Bluetooth?

Mark Day: Look at this space, can you imagine yourselves in this room all with multiple bluetooth devices that are talking with each other configuring themselves to each other?

Mike Mealling: Perhaps we should talk about naming bluetooth services...

Henrik Rood: {mentions a site with many drafts on services in cars}. Why not look at 3rd generation mobile.  There you find
   use of links ... identification card ... Many issues have been dealt with by mobile vendors, and they do have a large namespace.

Mark Day: My feelings are based on looking at naming in areas away from telephony, but I didn't think there had been much progress with mobile systems. Should everything be a phone number?

Henrik Rood: The European mobile system is based on smart cards, and there are gateways to the Internet. The use of smart cards in Europe address many concerns.  For instance my smart card offers services.

Mark Day: Perhaps I overlooked the idea that telephone numbers will be unique. And that perhaps the end points are nothing but telephone numbers.

Henrik Rood: No not just telephone numbers, but the chip stores information for you.

Q: Service can be mobile.

Mark Day: That's portability of the endpoint.

Henrik Rood: But I can use Voice Dialing to find a place for you. Right now in the Netherlands, you have voice dialing for specific words, for example, you can pick up the phone and say "Coke" and it will dial for you.  This is a service that I have from the Telecom.

Mark Day: This is implemented centrally, right?

Carl Hewitt: Ten years from now, the unrecorder life won't be worth living. We will record entire lives on video and exchange video. Humpty Dumpty gets along just fine without names. Pointing is just fine. Names aren't always character strings!

Mark Day: Good point.

Rohit Khare: We have socially constructed expectations for devices. For example, is the cell or home phone primary? We assume a cell is more imperative than a fixed line, but for me that's not the case. My cell phone is for everyone to contact me, and my home phone is just for friends, and that's where I answer if it's important. The terms themselves are of relative value. I'm sure there are network effects on the notion of concept of devices.

Mike Mealling: There is an interesting interface between the IETF and telephony SDOs. There is some discussion of using DNS for telephony information.

David Rosenblum: Names seem to be used here for humans to name things, but how about names that have to generated quickly and have a short lifetime. What about names that machines must generate on the fly to talk to each other?

Walt Scacchi: Need guidelines for rolling your own naming schemes.

Mark Day: I'm concerned about this. The X window system allowed people to make their own window managers. The idea being if you build the capabilities, the people would come an build thier own GUI -- provide just enough functionality to roll your own.  But it turns out that people didn't want to build their own windowing managers. This was a little underconstrained for many users.

Walt Scacchi: Maybe that's the answer -- evolve towards a socially determined naming solution. Thus everything is a locally constructed name

Bob Morgan: If everything is context dependent, then what about the help desk scenario. Can you please tell me what context you are in? How do you come to a resolution?

Mark Day: Support is a very important issue. Don't want names to break if a resolver is down for one person, but not for another. This is a big challenge.

Rohit Khare: A closing thought. The battle in instant messaging is over ownership of the namespace.

Mark Day: I don't agree. Microsoft and AOL don't even agree on the business model. AOL wants to control the service, as opposed to the model where it's just out on the Web and anyone can just link to it.

Esther Dyson: It's a commercial dispute masquerading as a philosophical dispute.