Discussion during the presentation:

Slide: "Global Positioning System (1)"

Carl Ellison: How does differential GPS compare to military GPS?
Michael Gorlick: Similar - within 0.5m. But we can do much better than that with kinematic GPS.

Carl Ellison: Does that require line of sight [to differential station]?
Michael Gorlick: No, not to the differential station. Only to the satellite.

Slide: "Global Position System (2)"

Q: The receivers are small but not the whole unit, am I right?
A: No, the whole unit.

Slide: "Geospatial Addressing (2)"

Rohit Khare: The address resolution is not high enough today.

Slide: "Geospatial domain names"

Scott Michel: "uci.irvine.orangecounty.ca.us.geo" could simply be an absolute address, but you need an absolute reference.
Michael Gorlick: One man's relative is another man's absolute.

Post-presentation questions and answers:

Esther Dyson: Really neat. But why use the domain name system at all?
Michael Gorlick: I agree. Feel free to go ahead and fix that problem.
Esther Dyson: Why reuse the DNS, and get tied into all the politics?
Michael Gorlick: I used it for the same reason any implementor does -- because it's there
Esther Dyson: My favorite applications are luggage finders, child finders.  Why not a larger space, more general namespace than DNS? Why not many  kinds of executable resolution servers?
Michael Gorlick: Well, once you have it all fixed... :-)
Esther Dyson: With all due respect, route around the DNS
Michael Gorlick: route around the damage?
Esther Dyson: Route around this excellently working system :-)

Mark Day: What are you going to return when you *resolve* these names? a car ought to yield vectors rather than lat/long points.
Michael Gorlick: I'm a middleware/infrastructure guy -- these are just basics to a much more intriguing technology on top.

William Kornfeld: Looks suspicious. By calling it a domain server we assume you have a place you can send a message to -- there's nothing listening there.
Michael Gorlick: Unfortunately there is not enough time to talk about geospatial routing."

Larry Masinter:  I can see that "names vs. address' is fuzzy, but this is so much a locator that I'm uncomfortable. The mapping between the name and the physical space is an interesting thing -- the city of Irvine, e. g. -- you've been real free about talking about both things interchangeably. We want to map between names and locations and then route between them. The really serious problem is, where is the namespace of places? Dublin Core... a server could tell you where it's located, what it covers. We need to connect some kind of service in a name I designated to place names. Introducing time, 1) the span of a physical object -- car position changes over time 2) names themselves migrate in what they designate -- what were the boundaries of Irvine in 1945? (Null!). The mapping changes over time, too... the timespace of the intended mapping is a critical input, too.
Michael Gorlick: Yes, it is time dependent.

Rohit Khare: remarks that he refuses to talk the blame anymore. {Ed: Michael Gorlick used Rohit a lot as an excuse to shorten his answers.}