Directions Magazine (DM): How is this year's event shaping up compared to previous ones?
Paul Ramsey (PR): Every year, there is one "international" conference for open source geospatial, where the developers from every project make an effort to come together. In 2003, the "Mapserver Users Meeting" in Minneapolis drew 100. In 2004, the "Open Source GIS" conference in Ottawa drew 200. In 2005, the "Open Source GIS" conference in Minneapolis drew 350. In 2006, the FOSS4G conference in Lausanne, Switzerland drew 530. There's a trend there, if you squint your eyes and look really close.
DM: I understand there was a unique way to select content for this year's conference. Can you tell us about that?
PR: We knew we were going to have more presentation abstracts submitted than we would have room for, and that there would have to be a process to select the program. The default process would be something where the committee would sit down in a room, everyone would select their program, the results would be summed up, and the program chosen using the average scores.
As I was considering this, I thought, "What makes the program committee so special?" Really, all we want is a program that appeals to the maximum number of attendees, so the best population to sample to create that kind of program is the attendees themselves. So rather than a "closed source" selection process, where the committee huddles, we went for an "open source" selection process, where everyone in the open source community and attendance pool could contribute.
We put all the submissions into an online form, and let anybody on the Internet provide scores on the submissions, with a simple email system to verify identity. It worked really well! We got over 12,000 "opinions" in the database, from 250 different people, and it really made clear what presentations were of the most interest, to the extent that we're also going to use the scores to decide which talks to put in the larger rooms.
When the call for presentations was coming up, I was talking to some folks about what topics they were going to submit, and it was "Obscure This," and "Advanced That," so I told them that they should really do introductory topics. Because so many people come to FOSS4G to figure out the open source world, it is their first taste. And so, to avoid being a hypocrite, I submitted an abstract from an old talk I did a couple years ago titled "A Survey of Open Source GIS," and it ended up as the top vote getter. A lot of the other top talks had a similar "introductory" or "overview" feel. There is a huge thirst for basic knowledge about "How This Stuff Works."
DM: What events/technologies/products from the last 12 months have helped push open source geospatial forward? How?
PR: From my point-of-view it is FDO, the Autodesk data abstraction library that they quietly open sourced along side MapGuide last year. FDO stands for "Feature Data Objects," it is a software layer that allows applications to use multiple databases and file formats for both reading and writing without needing to know what the underlying format is. FDO has added support for a bunch of new data sources in the last 12 months. There is a new PostGIS native FDO driver. There is, curiously, an open source Oracle FDO driver, even though Autodesk delivered an Oracle driver as part of their release. There is a multi-format driver that wraps the open source OGR format conversion library. The net result is a huge win for Autodesk, they have a much wider format support than they did a year ago, and they had to invest relatively little to receive it. It's a validation of the open source model, everybody wins.
Tyler Mitchell (TM): On the events side, there is continuing momentum creating new regional user groups under the OSGeo banner. The OSGeo official Local Chapters list is good, but the ones that are currently forming or planning to form soon are incredible. This year has seen several new groups get started, with several more likely to form over the upcoming months. Some plan to use FOSS4G as a place to launch their new chapters. Helping bring together this international set of communities is a great way to push open source forward.
DM: What new open source projects or products will be introduced at the conference that readers should be aware of?
PR: FeatureServer and TileCache, both out of MetaCarta Labs, are fabulous new projects that are well worth a look. TileCache provides a back-end for Google Maps style Web mapping - it can stand in front of an ArcIMS server and provide a tiled view of the data for super fast Web mapping. FeatureServer is a demonstration of a new form of "RESTful" access to spatial data. REST models Web services to work the same way as more "normal" Web content, and as a result can provide much simpler and more flexible systems designs. It's brand-new technology, but the REST approach to Web services is receiving a lot of interest these days, and FeatureServer is an honest-to-goodness implementation, not just a notion.
DM: In years past I noted the success of open source Web mapping and the slow uptake on the desktop. Is that changing? What evidence is there?
PR: It is changing, there is no doubt. The evidence is the vitality of the four different desktop GIS projects: uDig, QGIS, gvSIG and OpenJUMP. Each is managing to hold together a diverse development community and continue to move forward, and generating more and more users. The key is not the "high end" GIS of analysis and cartography, though that will come, it's the low end ArcView 3.X level of application. The first project to produce a stable and complete ArcView 3 replacement will gobble up a huge user share, and become the default application for building the "high end" analysis and cartography functionality.
But the demand for the simple things ArcView 3 does - data capture, data editing, querying, selection sets, sorting, some scripting - is huge, and the new ESRI desktop alternatives are too expensive for the kinds of folks just scraping by with ArcView 3 now. They want a new option, it has to do all the things ArcView 3 does now, and they want it to interoperate with the new technology available - spatial databases like PostGIS and Oracle, Web services, and so on. Watch for it, it'll be a tsunami when it happens.
DM: The business model for incorporating open source solutions is not new to smaller GIS consulting firms that may have previously offered custom software development and other solutions from proprietary vendors. Yet, we do not hear many firms offering services specifically for these open source platforms. Do you see it differently?
PR: Yes, the services firms are increasingly adding the open source tools to their skill sets. You are not hearing about it because services firms don't advertise, particularly - product firms do. Also, because companies are not saying "we are an open source company," they are telling clients "oh yeah, we can do that for you with this other open source thing instead." It's not being marketed as a replacement, but as an enhancement to their existing skill base. I am writing this response from the airport, because I'm traveling off to give a PostGIS training course, not to a government bureau, but to a consulting firm.
The open sourcing of FDO and MapGuide has also driven "open source awareness" into some companies that might previously have thought of themselves as "Autodesk companies." They are finding they can build solutions with hybrid stacks of Autodesk offerings, open source databases, open source scripting, and so on, with a lot more flexibility.
TM: OSGeo has been gathering an international list of service providers that will be unveiled shortly. Our tool will allow you to search by country, technology and/or language. The organizations range in size from single consultants to dozens of employees in a firm. We currently have about 80 organizations listed. We expect this to be a powerful tool for open source adopters who need to find support, but also for advocates who need to show that development support services are actually available.
DM: Open source technology appears widely adopted in many corporations, or at least is not perceived as a drawback to adopting proprietary solutions. Do you have any information, anecdotal or quantitative, on the adoption rate by government agencies or businesses of open source geospatial technology?
PR: I wish I did! There is a steady rise in download and traffic statistics at the PostGIS Web site that I can track over time. And there is our local experience as a company, with changing attitudes, from "open what?" to "open source!" But rates are hard to determine. One thing is clear, when people are making a green-fields decision on what technology to use when building a new enterprise, open source wins out over and over and over. The costs of building a proprietary system from scratch just don't make sense these days.
The main resistance to open source comes from organizations that are already in the proprietary camp: they have a sunk investment that is very difficult to walk away from, so they keep on shouldering the incremental costs of keeping it running, in over-priced maintenance, upgrades and new products from the incumbent vendor. In the short term, it's cheaper than changing. In the long term, they are getting hosed.
DM: What do you see as OSGeo's responsibility for marketing open source technology?
TM: OSGeo aims to draw more attention to open source technology as a viable option, in particular where solution providers may not have heard of it. We do this formally by giving talks at conferences and exhibiting alongside the closed source vendors. We also do it informally by bringing people together under the OSGeo banner. Those involved in various committees, local groups, end users, etc. have unique opportunities to help address common questions and concerns. This serves as support for those within organizations making a case for open source or those who need more assurance that it is viable. Because most projects do not have a company with a large marketing budget behind it, OSGeo fills that role for them by representing them publicly and helping to direct questions or new users back to the project itself.