February 10

Newsletter Archive - February 10, 2005


Editor's Note
Microsoft Submits Patent on Simplifying Location Coordinates
Google Maps
ESRI Developer Network to be Unveiled
FINDbase Adds New Twist to Geotargeting

Editor's Note 

Today's issue of GIS Monitor will be the last that I edit. I'll be retuning to my consulting business and joining the team at Directions Magazine. GIS Monitor will continue with a new editor, Matteo Luccio, an MIT graduate, and former managing editor of GPS World Magazine.

It's been a wild ride since this publication debuted August 18, 2000. I've covered acquisitions, new technologies, lawsuits, exciting (and not so exciting) conferences, profiled interesting people, and generally spoken my mind on the technology and business worlds that relate to geography. When I started GIS Monitor, GIS was the term of choice, today this arena is popularly described as geospatial and includes location-based services, remote sensing, earth observation, GPS, fleet tracking, indoor locating, LiDAR, and a host of specialty geospatial technologies and business opportunities.

From a personal standpoint, the most rewarding part of writing and sharing the content of GIS Monitor was how it impacted, you the readers. Your often immediate reply e-mails that indicated something had "caught your eye" or "pushed your buttons" or "ticked you off" were gratifying and confirmed I was addressing things that mattered. And, I've been humbled by the number of corrections of fact, spelling and grammar you have shared over the years. They've made me a better journalist and writer.

Two special "thank yous" are in order as I depart. I need to once again acknowledge the contributions of Ralph Grabowki, whose model I shamelessly stole for this publication. His UpFront.Ezine continues to thrive after five full years, and I continue to read it every week. The other thank you goes to Roopinder Tara, co-founder of this publication and CEO of TenLinks. Or, said another way, he was willing to pay me to start this publication in the early days of the Internet economy. TenLinks, now focusing exclusively on CAD, is a must read for that world.

I wish all the readers and advertisers of GIS Monitor the best in their geospatial endeavors.


Microsoft Submits Patent on Simplifying Location Coordinates
It seems Microsoft has filed for a
patent to create a simplified representation of latitude/longitude pairs in base 30. The company calls it "compact text encoding of latitude/longitude coordinates." The idea, as I understand it, is to take a complicated floating number pair (with decimals and all) and communicate it in a simpler form based on the 10 digits and the 26 letters minus the vowels. (That last part is supposed to prevent real words, offensive words, from forming.) The reason for such a coding? Say the folks at Slashdot: it's to simplify addressing for things like location-based services. Reader Martin, who sent this on earlier this week, noted that it's sort of like the NAC Geographic Products Inc. (NACGEO) alternative coding of locations, that's been covered here in the past. Now, this is just a patent application; it's not clear a patent will be forthcoming.

I was still trying to digest the idea and to think through if this sounded like something that should be patented when I received a press release from NACGEO asking "Does Microsoft Infringe the Natural Area Coding System?" The title was odd since so far as I know, Microsoft does not say in the patent application though it may elsewhere, that it uses such an encoding. The release goes into significant detail about NACGEO's proprietary format and encoding technology. More interesting, however, is that back in 1994 the company did begin the process of applying for a patent on the encoding in the U.S. and internationally. It was withdrawn (the release uses the term "retreated") though it's not clear when that occurred. Instead of a patent, the company decided to copyright the format under international copyright law and treat it as a proprietary standard. The release also documents that NACGEO proposed its idea to Microsoft in 2002 and received a "no thanks" letter. The argument, which is almost lost in these details, is that save some small tweaks, the algorithm Microsoft offers is virtually identical to NACGEO's implementation.

From what I understand, a file format is not copyrightable. The data in it may be, but according to a graphics file FAQ, a format is considered an idea/system. Is a file format patentable? Yes, audio formats are patented, for example. Is a file format different from an encoding? Microsoft has applied for and been granted patents for the process of creating/unpacking XML encodings, a "specific software implementation." The encoding I hear about all the time is Geography Markup Language. I know that's not patented since it's an open standard. Discussion on Slashdot suggests that this may be the issue - whether such an encoding should be an open standard or not.

Other discussions suggest that such an encoding is parallel to the U.S. government's own Military Grid Reference System, discussed here recently. It essentially simplifies coordinate pairs, just in a different encoding than the one Microsoft suggests. Is that prior art? That is, does the existence of this encoding make Microsoft's technology discussed in the patent application "less" than unique and original? That's for the U.S. Patent Office to decide.

I find it interesting that NACGEO chose to have this discussion in the press. (Two other GIS publication editors contacted me on the matter after the release came out.) The release notes NACGEO's founder suffered "long time extreme hardship and numerous sleepless nights." It also acknowledges Microsoft's power. NACGEO suggests its achievements are now in "great danger."

This topic is complicated and includes serious mathematical and legal issues. I should think it would take technical and legal minds some time to digest the material before putting out a detailed statement. Normally, when there is a patent challenge (and recall there is currently no patent to challenge!) the challengee simply states: "We do not infringe" and years later a trial or settlement occurs. But then, NACGEO is not a large company with legal resources like Intergraph or Microsoft, so issuing a detailed release suggesting Microsoft's potential misdeed clearly seemed the best path forward.

Google Maps
To keep up with Microsoft's and Yahoo's searches and to perhaps take on MapQuest, Google this week rolled out the beta of
Google Maps. Reader Atanas passed on the link and I found a heated discussion of the new offering on Slashdot. An article from Ziff Davis (ZD) was one of the few in the tech press on Tuesday. That's what is called a quiet or soft launch.

A few limitations first: Browser support is for Internet Explorer and Mozilla thus far. Coverage is limited to the U.S., Puerto Rico, and a bit of Canada. Data comes from both NAVTEQ and TeleAtlas, so far as I saw, though ZD notes only Tele Atlas. For now, there's no way to save locations.

On to the technology! While I read on Slashdot the maps are "vector" it sure looks like they are delivered in grids blocks, one at a time. It's fast and the interface lets you "pan in real-time" across the map. I understand there's JavaScript going on to enable that. A sliding "thermometer" allows a change of scale. Jumping between very different scales was the only time I found myself waiting any appreciable length of time for a redraw. Late word today from Telcontar - its technology is under the hood. A full technical dissection is available here (thanks to reader Jeff), along with Slashdot's follow-up discussion.

The maps are pretty - there are double line streets with the main streets highlighted in yellow. (That's Davis Square, above, the center of my universe.) Notice how nicely the labels are placed. Any guesses what's behind that?


Geocoding is fast and I appreciate that instead of just a dot or star noting the location, a flag with text identifies the address on the map. (That's my college dorm.) If the location is wrong you can immediately tell if you keyed in a typo. The flag include a "Directions from Here" and "to Here" set of links. That's pretty intuitive: if I'm interested in this location I might want to leave it for somewhere else, or make it my destination. On quick look, the directions seemed fine. A click on any single step in the directions creates a callout of that part of the path. Unfortunately, the callout is not live so you can "pan around" with it. (In Cadcorp SIS that's the "roamer.")

Searching for a local business was easy and free form. "Coffee near 4 main street Medford ma" yielded a map with 10 upside down read "teardrops" pointing to the establishments found. Details and links are provided in a list on the right side of the page. There's not yet a way to add missing businesses, though Google is working on that as part of its Google Local program.

I found only one thing confusing. For each type of search on the Maps "tab" (Go to a location, find a business, get directions) there are two input boxes. I believe these are to offer several examples, which is fine, but not practical for actual use. If you put an address in one and other in the other, the system will not provide directions from the first to the second. That was my first guess at how to use them. To get directions you put both start and end point in a single box with "to" in between.

A "what" box and a "where" box which can be found on the Local Search "tab." You put what you want to find (coffee) in one and where to look (San Francisco or 90210, etc) in the other. One code that will not work: an area code.

On the Directions "tab" you can key in a starting and ending point each in its own box to find a route.

On the whole, the interface is slick and friendly. And, the maps are pretty and on quick look, as accurate as any others. So, why should you care about Google's offering if you are a geospatial professional? How about because a Google "search appliance" will be part of ESRI's version 2 of geodata.gov? Consider how a tool like this might enhance not just business searches but geodata searches.

ESRI Developer Network to be Unveiled
I received a heads up on the soon to be announced ESRI Developer Network (
EDN). EDN (pronounced "Eden") is modeled after Microsoft's Microsoft Developer Network (MSDN) and provides unification of a several resources for ESRI third-party developers. The program is offered per developer, that is, per person. Each subscriber receives developer software (ArcGIS Server, ArcIMS, ArcSDE, ArcGIS Engine Developers Kit, ArcWeb Services (100,000 transactions, [corrected 2/14/05 erroneously said 200,000 in original version]) and a license to run it until the subscription runs out (one year). The software is to be used only for development, not deployment, and can be used only by the named developer. It can however be installed on more than one machine for use by that developer.

Subscribers also gain access to the EDN website currently listed as "in beta." It offers samples, downloads, and opportunities to share with other third party and ESRI staff developers. It's the main source of developer support, though additional phone support is available at an extra cost form the Developer Support Group (DSG). Subscribers can purchase discounted training along with the subscription (but not afterward).

ESRI has not yet announced pricing.

EDN has been in the works for a while and appears it will be formally launched at the ESRI Business Partner Conference which begins on Saturday.

FINDbase Adds New Twist to Geotargeting
Back to the Future

One of the first issues of GIS Monitor highlighted Quova, a company offering a then relatively new type of service to locate Web surfers. The idea was to use it to help target ads to specific geographies. Since then geotargeting has been used to block out access to entertainment (baseball games, gambling) and track illicit money transactions.

This week a company new to me, FINDbase, launched Geocation, a solution "to authenticate the geographic location of Internet users." Geocation, a term the company made up, is defined as "the action of using two FINDbase technologies, Geocate and GNT, to detect the physical location of an Internet user in real-time." The combination, says the company, overcomes major shortcomings of other widely used geolocation tools including triangulation and IP mapping.

The Old Way

Triangulation in this context refers to guessing at location based on the speed of signal to reach the target hardware. Known locations and the time to them are mapped. Then, when a new location is to be mapped, its "times" are compared to the known ones to find its "neighborhood." Signals on the Internet take different paths and can get slowed down by different factors, so this is not an "exact science" by any means. The company compares it to using GPS, but of course, it's a loose comparison since GPS works on a "straight line" path and those through the Internet are not only circuitous, but generally not repeatable.

IP mapping, according to the website refers simply to looking up an IP address in a "proprietary database," a sort of IP phone book. How do you write the phone book? One way is to send out "requests" to IP addresses. Some of those requests, which are sent several times a day or more, look like denial of service attacks and are "blocked" by ISPs. Another is to simply call someone up and ask where they are! This is sort of a "brute force" method. And, there are complicating factors. Sometimes the location returned is that of the Internet Service Provider (ISP), not the actual user. So, for example, most AOL users appear to be in one location. That means that if you want to be sneaky you can simply use an ISP in the area from which you want to be associated. Another complicating factor: fewer and fewer IP addresses are static. That is, over a few hours someone may have several different addresses assigned on the fly. Quova and Digital Envoy, the companies I profiled four years ago, both use IP mapping, according to FINDbase.

The company highlights that its service supports both IP4 and IP6 addressing. The "old" Internet used IP4, which provided about 4,294,967,296 (2 to the 32 power) unique addresses. That number was reasonably mappable in a "phone book" noted above. IP6 has 2 to the 128 power (34 followed by 37 zeros) unique addresses, far more than current technology can handle, says the company.

The New Way

Once you get beyond all the "issues" with existing offerings, FINDbase describes its offerings: Geocate and GNT. Geocate is an artificial intelligence (AI) solution that learns more and more about the network as it follows more and more traffic. The idea is that if the patterns of use of different IP addresses are known, it's easier to see "misuse." I compare it to when the credit card company flags "odd purchases" that don't fit your profile. (When some fellow in Texas started buying computers with my card, I got a nice call from the credit card company.)

GNT stands for Geographical Network Tracking, and is, from what I can see, a "souped up" triangulation solution, also built on AI. It uses time, but unlike GPS, measures both time from the user and time to the user. The company contrasts its offering with the National Security Agency's NGT, which uses another flavor of triangulation to determine location. Don't get confused!

The company makes it clear that "the best" results come from using the two technologies together. Still, each option is available separately. When I asked why, the reply was simple: for different types of questions, one offering may be far better. Some questions are quite simple: Is the user inside or outside Nevada? For others, say crime tracking, a more detailed location may be needed: "On which side of town is the alleged assailant?" Both Geocate and GNT are available as a service or as an installed product.

» Back to our GIS Monitor FEB 2005 Issue