Abstract
For the last four years, the City of Philadelphia has been working to bring GIS technology to the level of neighborhood planners. While successful in generating enthusiasm for the applicability of GIS for this purpose, use of the technology in the neighborhoods is still minimal. A 1995 paper by these authors documented the project with particular focus on the limits of existing mapping techniques and symbology for mapping urban neighborhood environments. This paper documents the progress of the City's continued efforts to give its neighborhood planners access to its GIS resources and the impact on that effort of new technologies. Our theme continuing from the awareness of the difficulties of mapping urban environments, is that although the City may now be in a much better position to distribute its GIS data through less expensive, easier to use interfaces, it could be exacerbating the difficulties of effectively mapping urban neighorhoods.

Background
The City of Philadelphia, with the foresight and support of the Rendell administration, has been on the forefront of GIS since the technology became viable for a municipal environment. In 1994, recognizing both the appropriateness of moving neighborhood planning back to the neighborhoods and the applicability of GIS for this purpose, the Office of Housing and Community Development (OHCD) funded a pilot project to bring GIS to the neighborhoods. The project provided equipment, software, data and training to a 6 of the City's 25 Community Development Corporations (CDCs). CDCs are inner city neighborhood organizations with a goal of neighborhood revitalization. They emerged in the 1970's as participants of the funding and support generated by the "War on Poverty." Both authors were drawn into the activities surrounding the GIS pilot; Casey as the head of GIS for the City, and Pederson as the consultant under contract with OHCD to provide training, support and data to the CDCs.

After working with the participant CDCs on almost a daily basis we came to well understand the issues emerging from the GIS pilot. There were a number of practical and logistical problems ranging from bad addresses to problems involving the transfer of data between incompatible operating systems (DEC VMS and Windows). However, while addresses can be corrected and data transfer paths can be kluged, during this pilot an unforeseen category of problems emerged, much broader than the program itself, for which there seemed to be no significant recognition. The maps we started to see as a product of the pilot could not be compared side-by-side or collectively. Each attached significance to color differently and used different classification schemes and symbology. It placed a tremendous burden on the map-reader when he or she attempted to compare the maps. In addition, the maps seemed very limited in their ability to portray the qualitative aspects of a neighborhood environment. Because we were witnesses to maps from neighborhoods scattered over the City and had developed site context based on our repeated visits, we were in a position to notice that the maps did not meaningfully convey the very distinct physical and social disparities in the neighborhoods. In 1995 we wrote a paper for the ESRI Users Conference that focused on the limitations of traditional mapping standards, techniques and symbology as applied to mapping neighborhood environments.

In our research for the 1995 paper we discovered that while the problem of mapping the elements needed to portray neighborhood environments had been recognized, there were very few suggestions of means to resolve it. Our paper proposed a three tiered approach that included standardization, structured classification, and the development of appropriate symbology. However, as we acknowledged in the 1995 paper, "the answer for the CDCs is, obviously, not a simple solution that we can profile in this paper and implement through our roles as promoters and supporters of the GIS project."

Current Status
Four years have passed by since the inception of the GIS pilot. The vision, which was that by now scores of neighborhood planners and interested citizens would be sitting at PC's in the CDC offices using GIS to both query about information regarding the particulars of their environments and to perform "what if" scenarios to assist with strategic planning, has not come to pass. If bringing that vision to reality were the only measure of the program's success, it failed. For all distribution of PC's and software, the cleaning and organizing of the data, and the hours of training and handholding, there is not widespread use of GIS at the level of the CDCs. A recent survey showed that only three CDCs use GIS at all and those three sites use only the system's most basic functions primarily for presentation purposes, instead of for analysis and planning.

Everyone concerned, OHCD, other city agencies watching the process, the CDC's themselves and the authors realized that one obstacle in reaching this vision far overshadowed all of the others--the two-forked problem of lack of skills necessary to use a GIS and the turnover of those with the aptitude or training. CDCs have extremely limited budgets and their staff's do not come highly trained. People with GIS skills, especially good conceptual and analytical skills, can easily find higher paying jobs. Our problem was that we underestimated the gap between the skill level needed to navigate a Windows based GIS interface (ArcView in this case) and the skill level we would find in the CDCs. Too much hinged on the ability of the group's designated technology enabler.

However, while the specific vision of "public participation GIS" described above was not realized, the project was not a failure. On the contrary, the work that went into that pilot, the personal contacts and the "bell ringing" about the applicability of GIS to neighborhood planning brought, across the board, increased awareness of the potential of this technology. The best witness to this is the fact that OHCD is still funding neighborhood access to GIS at a rate of about 60 thousand dollars each year.

The City, OHCD, and others involved responded to the problems of the GIS pilot with new strategies. Instead of continuing to fund individual CDCs, OHCD funded the Philadelphia Association of CDCs (PACDC), the umbrella organization for the CDCs to provide a "center" for GIS activity where CDCs could find continuing technical support for neighborhood mapping without having to employ the skills themselves. They could walk in to PACDCs office and emerge with a map. PACDC over the last two years has created over 300 GIS generated maps responsive to the requests of CDCs. The City made GIS data available to numerous non-profit consultants to provide maps in support of funding requests. The City also collaborated with the resources of local Universities to provide GIS based services.

Probably one of the most remarkable collaborations to further the goal of dispersed participation in GIS was Philadelphia's application for a grant from the National Spatial Data Infrastructure (NSDI) Competitive Cooperative Agreement Program. In 1996 the City partnered with ESRI (developers of ArcView and Arc/Info), ADR (a local GIS consulting firm), PACDC, and the University of Pennsylvania for a grant to design an interface for Internet access to GIS for neighborhood planning (reference number 96001). One of the things being promoted by NSDI through this grant program was standards. Having already realized the need for standards for effective neighborhood planning, we reasoned that if there was wide spread Internet access to data presented using the standards on the Internet it would lead to a familiarity with the symbology--a shared language--which would encourage its adoption. The grant was turned down with for some very specific reasons--among them was, "It is not clear if there is a willingness to continue the project or build upon it after completion of the design." As it turns out, the willingness was there. The City, OHCD, PACDC, and the University of Pennsylvania have continued their efforts to provide neighborhood access to GIS which may be beginning to come to fruition.

New Technologies
Accompanying the realization that wholesale access to GIS was not sufficient to bring GIS to the neighborhoods in a useful manner, have been continuos changes in the underlying technologies. The most important of which are: 1) that GIS software manufacturers began to provide the capability of linking GIS systems to Internet technology; and, 2) that the same GIS software manufacturers now provide open development environments between GIS and standard database interface tools such as Powerbuilder and Visual Basic. As a result, the tools available for dissemination of GIS technology have drastically altered.

In 1994 the goals were to move portions of the GIS data and related city records to stand alone PCs located in distributed offices for the staffs and interested citizens to "have at it." Now, the technology exists to maintain and store the data on centralized databases and provide Internet access through an interface that requires very little training and skill to operate, eliminating the problem of finding and keeping appropriate personnel. The user-centered task driven interface can provide users with a series of coverages that can be turned on and off, tools to manipulate the extent and scale of the data and tools for analysis. In this year's proposal for funding from OHCD, PACDC specifically describes taking advantage of this technology. "PACDC will reincorporate and centralize processing and fundamental analytical functions before it is distributed to CDCs. The focus will be to develop information "packages" that can be easily disseminated and interpreted immediately no further processing involved. A user will remotely submit a query whereby the interface will access the data sets in real time, dynamically aggregate and calculate the data to a neighborhood level and display a community profile in tabular and graphical formats."

Besides ease of use, there are other advantages to the new technologies. The cost to the user is reduced to the cost of a PC and a Web browser which can usually be acquired at no cost. With the data sets stored centrally and the processing being done on the server, the need for hefty PC's to accommodate the large files and complex processing is eliminated. A network (or bare bones) PC would be ample.

Another advantage is that it will become de facto the means by which the use of standards in neighborhood mapping, which we have been championing for some time now, will be enforced. The themes, symbology and classifications will all be pre-set and unalterable. As we projected in our NSDI grant application, the presence of these standards as part of an interface allowing access to data needed for neighborhood planning will make all users, interested in that same data, familiar with the "language" and encourage its use.

 In addition, the City has responded grandly to the new technology. An unprecedented application in 1997 from the Mayor's Office of Information Services for 1.7 million dollars to build a server to provide Citywide access to GIS including Internet access was funded. This server, which should be installed in early 1999, is being designed with the capacity to accommodate the needs of neighborhood access.

The Continuing Problem
Because the technology is in place to more easily disseminate GIS data, there has been much discussion and activity that touch on "public participation GIS" in the City of Philadelphia. This is true in the context of neighborhood planning but also in many other contexts such as the title community, those interested in zoning information, tax assessment patterns, permitting, code violations, tax delinquency, property value, utility maintenance data, and many others. The obstacles to publishing this data caused by the limitations of the technology may have faded, but, in Philadelphia, as in many other places, the issue of distributing the data evokes numerous strong opinions regarding legal implications, political ramifications and the appropriateness of charging for the data. Although Philadelphia is not as far along on its distribution policy as some other cities, in time, these issues will be resolved as were the technical issues before them. However, in the context of neighborhood planning, the difficulties that we found hard to resolve in mapping neighborhood environments in general become even more complex.

The problem at issue in the 1995 paper, namely that existing mapping technique and symbology are inadequate to map the qualitative aspects of neighborhood environments has not gone away. In addition, the "packaged" GIS intended to ease the impediment, i.e., the skill level required to operate a stand alone GIS, potentially limits the available data, the tools for presentation and analysis, and the features that can be manipulated. The components that are included in the interface between the user and the City's data, which is in reality just a very sophisticated Web page, become the only components available. A great deal of research and architecting will have to go into the development of the interface for it to be truly effective.

If one looks at the City of Oakland Web page (www.oaknet.com/government/ceda/ceda.html) and navigates to the GIS data sections, there is a wealth of information available there. Philadelphia would love to have its corresponding data sets available on its Web site. The data is useful for a variety of purposes such as for visitors to the City or checking the surroundings of a specific address. The Web site states, "Check ownership by address or Assessor Parcel Number (APN). Search for and view parcels which meet your development criteria. Select neighboring parcels, and print a report of same. Determine zoning, lot size, distance to fire stations, hospitals, transportation, etc." However, it is not particularly useful for neighborhood planning. Even if it included information that neighborhood planners usually seek such as tax delinquency or participation in local or federal assistance programs it would still generate maps that were essentially flat. Features that make a neighborhood unique such as cultural characteristics and architecture as well as places that have community value are not classified. An interface that included even more of the data currently available in City files along with Census data and standard survey data would still omit many of the factors that create unique neighborhood environments.

A GIS package that was prepared for use throughout the City could present the data in a biased fashion and unintentionally overlook neighborhood assets and resources. A good example in Philadelphia is an area in the Germantown section of the City that is very much impoverished and abandoned. The architecture of those abandoned properties is rare--they are beautiful stone structures on large treed plots (at least by Philadelphia standards). The City's databases simply categorize them as tax delinquent vacants. This fact harkens to the issue of insufficient symbology but, in addition, points to the limitation we see if elements neighborhood planners want to map do not correspond to the mappable features presented in the interface. Enough diverse, mappable features would have to be included in the interface for it to address the true needs of neighborhood planners.

A GIS neighborhood planning package for use in surrounding suburban environment would likely lend itself more easily to standardization in addition to being easier to design. The factors that would need to be taken into consideration would be more homogeneous. Suburbs generally do not have the type of drastically diverse environments that are serviced by the City's CDCs. They can be categorized more easily by the cost of the housing and proximity to service centers, shopping, recreational facilities, public transportation and jobs. Classification categories and mapped features would apply, more or less, throughout. In a dense and varied urban environment like Philadelphia the features that make a neighborhood valued to its residents are widely diversified. In South Philly it might be proximity to a vast open-air market that specialized in hard-to-find Italian food stuffs or corner cheese steak stands. In Center City it might be easy access to restaurants and cultural institutions. In the Russian section of the Northeast it might be the fact that shop keepers speak Russian. In lower Kensington there is growing concentration of Palestinians who have built an environment where they are surrounded by those who share similar cultural and religious practices. There are also areas in the City that are quite similar to a suburban environment.

This is not to say that an appropriate and useful interface cannot be designed for an environment like Philadelphia. In Philadelphia where neighborhoods are generally in much greater need of planning and, as was stated above, expensive skills sets are not affordable, such an application would probably have greater utility than in suburban environments. Our point is that in the same way that mapping technique and symbology need to be built to effectively convey the condition of urban environments, the component sets in a GIS interface still need considerable attention and examination before they could be effectively employed for neighborhood planning in a environment such as Philadelphia's.

In addition to the component set, the manner of presentation needs to be carefully considered. If map themes are limited to "hard wired" depictions, it will have a direct impact on the portrayal of various factors. Information coded as cross hatches could be used in conjunction with information coded with solid colors, but the possibilities become more limited when the need is to depict solid color-coded themes with other solid color-coded themes. The order of the themes, what is displayed on top of what, would also be preset and limit ways to view the data.

The stringent limits to maneuverability of features that this type of interface would impose would mean that experiments or "what ifs" would be severely curtailed. In Philadelphia, where we are losing population, a planner could want to show the effect of moving a few active residents in an otherwise vacant area to a viable cluster, creating a new neighborhood and freeing a large unutilized area for redevelopment. Or, a neighborhood might want to show the impact of a new business on the support services the business would need to take advantage of which could have a positive impact on community businesses. Or, a neighborhood might want to contend to be the site for a new City sponsored mural (a growing phenomenon in the City) and demonstrate to the selection committee the path a tourist bus might take. Unless permitted by the designers of the interface, these would be impossible to map.

Call for Attention
We see the limitations described above, not as a dead end, but as a call for attention, for more planning and research. If the City with PACDC and the local universities can create an interface that will get more City data to the neighborhoods with greater ease and lower cost that will be a major accomplishment. We will both strongly support that effort. If we continue to keep in mind that the ultimate goal is the creation of an interface that is effective at neighborhood planning, a much more effective tool will eventually emerge. This will be a highly iterative process taking place between those with design and programming skills and those with knowledge of the neighborhoods. It is possible that even the limits of these newer technologies will frustrate the process and we will have to wait for new technical capabilities to meet these needs fully.

In the mean time it is important to guard against forgetting that until the inflexibility of GIS packages described above have been addressed, the GIS interfaces being deployed will impose a bias on the way in which the information is disseminated to the CDCs. And, in reverse, it will also impact the ways the CDCs can present their story to politicians and potential funding sources.