It may be hard to convince the DNS administrators for your site or ISP to add LOC records to their DNS data. There are two major categories of objection: that it's not worth the effort and that it may give out data they don't want to give out (like the location of expensive computers).
First of all, it's not that much effort. At a minimum, it's the effort involved in finding their latitude and longitude (something you may be able to do for them), plus the time involved in adding one line in the DNS zone file, putting a record in for the domain as a whole.
Of course, they may wonder why it's worth even that minimal effort. After all, they don't need the information, so why would anyone else? The answer to that is that it makes things possible. CAIDA already has a number of projects that need geographic data about the Internet, and DNS LOC is the best way to make that information available to any project that may need it.
They may find they want the information themselves, anyway. Not to find their own machines, of course, but to make tools like VisualRoute work for themselves, or their customers.
Even assuming that DNS LOC data are both totally honest and totally precise, it's still rather silly to assume that a thief would bother to carry a GPS around just to find expensive hardware, when your site's street address is already listed in the phone book!
However, if security is really a concern, the data need be neither totally honest or totally precise. A Seattle-based company might put in a LOC record roughly centered on the Space Needle:
seattle LOC 47 37 11 N 122 20 43 W 20m 1m 5000m 100m
The use of the (optional) precision fields indicate that the site is within 5km of the location shown; this allows the hypothetical attacker a fairly large area (78.5 square kilometers) to search within.
Of course, even the hypothetical GPS-aided attacker, unless they're the US Department of Defense, will also have to contend with "Selective Availability" (the GPS system's built-in "fuzz").