In the course of a few e-mails, including questions and comments from several people, I received permission to share Dr. Yvan Leclerc's comments. He is Senior Computer Scientist at SRI and project director for the .geo initiative.

*****

Here are some details of the dot-geo concepts that address your concerns and, I hope, demonstrate that your worries are unfounded.

SRI International is firmly committed to working with international standards-setting organizations such as OGC, FGDC, ISO, W3C, Web3D, etc. and to developing the dot-geo implementation through these bodies. We have presented the dot-geo concept at a number of OGC member meetings, as well as in other related avenues such as the Digital Earth working group, the first JTC1/SC24/WG8 meeting, SIGGRAPH 2000, Global Grids 2000 conference, VSMM99 conference, the Web3D 2000 conference, and we are presenting it at the GIScience 2000 conference this week.

As a fully-paid Technical Committee member of the OGC, SRI submitted proposals in May 2000 under the Geospatial Fusion Services (GFS) and Web Mapping Testbed II (WMT2) RFPs including a test version of the dot-geo infrastructure (amongst many other elements, including GeoVRML and geoparsing), and proposals for integrating this with OGC's proposed services. SRI, a non-profit company, requested funds from the OGC to cover part of the expenses, but was denied due to insufficient funds.

David Scott and Martin Reddy (both members of the SRI dot-geo team) attended the opening GFS/WMT2 testbed meetings and we have continued to participate to the extent possible given the limited resources we had available internally for this project. Nonetheless, we are still fully committed to the interoperability of dot-geo and the geospatial services being defined by OGC and other bodies in the geospatial community.

When ICANN announced to the world in August 2000 that they were ready to accept proposals for a new TLD, we saw this as an opportunity to make georeferenced information available on a global scale, thereby adding value to the efforts of us all. ICANN's schedule provided very little time to prepare a extremely complex proposal, and very little time for outreach to organizations such as OGC. We spoke to a few members of the OGC staff to let them know we were in the process of submitting a proposal to ICANN and got invited to present our proposal at the next Technical Committee meeting in October, which we did.

We are, as always, open to any dialog that can enhance the usefulness of dot-geo to the geospatial and general Internet community, and ensure that it will work seamlessly with other geospatial services.

To this end, let me respond to some of the questions raised by you and the GIS Monitor article. (We are preparing a white paper to put all of these issues in context, but I felt that it was important to respond to your questions directly.)

- Formats and standard

SRI International is not enforcing any proprietary standards on the community, or intending to develop these ourselves. We believe this is the job of the standards communities, and in our proposal we explicitly state that organizations such as OGC and ISO should provide the foundation standards for dot-geo. SRI has a history of working with international and open standards. For instance, initiating and contributing to the open 3D GeoVRML format (http://www.geovrml.org/1.0/), based upon the SEDRIS spatial reference model and coordinate conversions API. GeoVRML 1.0 is currently being considered for ISO standardization.

- Dot-geo as a discovery platform

Dot-geo is an index for georeferenced data currently available over the Internet, akin to an Internet search engine, but with the ability to index by location. In a sense, it is modeled on the FGDC clearinghouse and Alexandria Digital Library efforts: a database of metadata with URLs pointing to the full metadata and data, indexed by keywords, location, time, and other attributes. The form of the full metadata and data are left entirely up to the data providers and end-users. However, there will be mime-types associated with each metadata/data URL so that end-users can limit their search to only the type of data they can handle.

Dot-geo is not a GIS, or a replacement for maps or atlases, instead dot-geo is a platform that will enrich these with current georeferenced information available on the Internet. There will be data registered in dot-geo which is not suitable in a traditional GIS (recordings of an outdoor concert, 360 degree panoramic images taken in Yosemite valley, temperature and rainfall measurements made by school children as part of a grade-school project). We believe that making such georeferenced data available on a global scale will create market forces that will accelerate the development of interfaces for registration, discovery, and display of geospatial information that are far more advanced than any available today.

Thus, dot-geo is complementary to ongoing OGC and other efforts. In fact we feel that it will provide further leverage for OGC-defined services and are very interested in continuing our relationship with OGC and other bodies (such as the FGDC) to maximize this potential.

- Validation of data for interoperability

dot-geo allows for optional validation by third-party validation agencies. Thus, a user registering data can optionally choose to have an outside validation agency certify that their geodata and/or data meet certain stated specifications, and users discovering data can optionally choose to discover only that data certified by those agencies he/she trusts. SRI encourage agencies such as the USGS or OGC to either sponsor or accredit such agencies, and let market forces decide which agencies are most useful for different applications.

- Dot-geo governance

As we stated in our proposal, we want to take advantage of input from all interested parties in dot-geo. But interested parties go far beyond the geospatial community per se. For example, the UN Commission on Trade and Development (UNCTAD), UNESCO, tourist agencies around the world, businesses and organizations that register in dot-geo, educational institutions that could take advantage of dot-geo, end-users, IETF, GeoRegistrars, GeoRegistries, ICANN, countries and regions who believe they are not being well represented by specialty sites, and so on. We believe that our proposed Forum and Advisory Council, combined with SRI's commitment and record for achieving consensus whenever possible (as recently demonstrated in our creation of the GeoVRML Working Group and GeoVRML 1.0 specification, which is now being considered for ISO standardization, and as the InterNic for over 20 years), is the right approach.

- Competition and level of performance in dot-geo:

What makes dot-geo different is that it's designed to work on a large scale, with millions of users discovering data simultaneously, and with 100 million to 100 billion geodata records or more stored across all servers. Imagine a user panning and zooming over a map of the world, from the globe down to street level looking for relevant information (or doing this in 3D). At peak speeds, this might require tens of queries per second to the index, multiplied by a million users.

Dividing the world into cells of different sizes and assigning a virtual server to each can provide this level of service. Using the DNS for delegation allows the service to grow from a few physical servers to as many servers as there are cells. In addition, if a physical server is delegated to a single cell, and demand increases beyond its capacity, than that server can be replicated and placed behind an appropriate load balancer, redirecting traffic as necessary. This allows any region to have as many physical servers as necessary to meet required performance levels.

This example illustrates one of the advantages of our approach: by dividing the world into cells of different sizes, replicated physical servers (when needed) are only responsible for a small fraction of the data. This is quite different from today's search engines in which load balancing is done by having the entire database replicated on many servers. This approach is simply not scalable to the massive database anticipated for dot-geo.

Meeting specified service levels will be a contractual obligation of GeoRegistries (the organizations that own and maintain the servers), which will be enhanced through the competition we propose between GeoRegistry providers.

- Competition and pricing in dot-geo

Another reason for having extensive competition amongst GeoRegistries and GeoRegistrars is to provide the market forces to keep the fees as low as possible to the end-user. Furthermore, competition will ensure that no single GeoRegistry or GeoRegistrar can dominate in one region.

- Registration vs. web-crawling

It has been suggested that indexing georeferenced data can be accomplished by adding a location tag to HTML pages. We welcome standards for doing this. However, this would only cover a fraction of georeferenced data available on the Web, and would miss out on multimedia files pertaining to a site (audio, video, photos, panoramic images), aerial/satellite photography, maps, USGS Digital Ortho Quads and Digital Elevation Models, 3D models in various formats, and much more.

There is currently a DNS record that can be used to specify the location of an IP address. This is very useful and could allow for automatic registration of IP addresses in dot-geo. But note that a single IP address can contain information about thousands of places (e.g. a bank's web site listing the location of all of its ATM machines). So, while useful for georeferencing the location of the computer that serves one or more web sites, that is not enough information to georeference the contents of the site.

- The need for the dot-geo infrastructure and TLD

Various organizations have estimated that between 50% and 80% of data on the Internet has a georeferenced component. This puts georeferenced information in a different category altogether, one that does indeed require a TLD and the massively scalable architecture of dot-geo.

- The role of JVTeam (NeuStar and Melbourne IT) in dot-geo

JVTeam has agreed with SRI to provide the domain-name registry services for dot-geo. They will not be providing GeoRegistry or GeoRegistrar services, nor will they have any special access to this data. We chose JVTeam because they will provide a high-performing, highly stable Domain Name Registry and Resolution service for dot-geo, thereby enhancing the stability of dot-geo and the Internet.

 
Warning: main(http://sparc.profsurv.com/gismonitor/feedback.php) [function.main]: failed to open stream: HTTP request failed! HTTP/1.1 404 Not Found in /home/sites/www.gismonitor.com/web/articles/comment/110200geo1.php on line 229

Warning: main() [function.include]: Failed opening 'http://sparc.profsurv.com/gismonitor/feedback.php' for inclusion (include_path='.:/opt/lampp/lib/php') in /home/sites/www.gismonitor.com/web/articles/comment/110200geo1.php on line 229