ࡱ> ` jbjb (|c|ci F F F t V+V+V+,:,\ ':,:,,,,...9999999,;R >9F .....9?3 ,,9?3?3?3. 8,F ,9?3 t. .9?3?3rN8T. F 8, ƿV+928890':82>?3>8?3F   dr$  r$Political E-Identity: Campaign Funding Data and Beyond David Wolber University of San Francisco 2130 Fulton Street San Francisco, CA., 94117 (415) 422-6451 wolber@usfca.edu  ABSTRACT This paper describes a joint university-government effort to develop software that helps journalists, voters, and watchdog organizations visualize campaign funding data in San Francisco. The paper also presents broader plans for constructing comprehensive electronic identities for politicians, and describes how emerging trends in on-line information systems can be leveraged. The newest version of the software is at http://www.whosfundingwhom.org. Categories and Subject Descriptors D.3.3 [Programming Languages]: Language Contructs and Features abstract data types, polymorphism, control structures. This is just an example, please use the correct category and subject descriptors for your submission. The ACM Computing Classification Scheme:  HYPERLINK "http://www.acm.org/class/1998/" http://www.acm.org/class/1998/ General Terms Design, Human Factors, Standardization Keywords Keywords are your own designated keywords. INTRODUCTION There has been little progress in campaign finance reform and public disclosure since Roosevelt spearheaded the Publicity Act of 1910. Computers have helped-- San Francisco and New York initiated the first required on-line filing systems in 1993, so that there are now megabytes of data existing in thousands of databases across the country. Unfortunately, there is a lack of sufficient software for viewing that data: existing software leaves journalists and voters to perform the virtual equivalent of rummaging through file cabinets to discover the web of financial relationships that control our elections and country. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Conference04, Month 12, 2004, City, State, Country. Copyright 2004 ACM 1-58113-000-0/00/0004$5.00. The inaccessibility leads to: 1) Voters being ill-informed concerning who is funding candidates, 2) Journalists and investigators spending weeks uncovering information that could be at their finger tips, 3) Campaign ethics commission administrators investigating only the most egregious of filing violations. The end-result is less transparent campaigns and more corporate influence handcuffing our leaders. The lack of sufficient visualization software can be attributed to the lack of funds for such projects in local and state governments, as well as the high cost of developing software. Most city Information Technology departments are overwhelmed with the challenge of making government services available on-line. There is also a natural disincentive for our leaders to push for making their campaign funding records readily available. The Transparency in Government project at the University of San Francisco (USF) is an example of how universities can help fill this void through collaborative city-university efforts. Computer science students provide a pool of relatively inexpensive software developers who are highly motivated to work on real-world projects. Working directly with the San Francisco City Ethics Commission directors and staff, USF students have created a software tool for viewing campaign finance data that is filed in San Francisco. The software periodically downloads the raw data from the Ethics Commission and builds graphs and web forms that make it easy for users to follow campaign funding trails. PROJECT ORIGINS Through its Ethics commission ( HYPERLINK "http://www.sfgov.org/ethics" www.sfgov.org/ethics), San Francisco is one of the leaders in on-line filingthey enacted the first mandatory electronic filing bill and built the worlds first on-line campaign finance database [Alexander]. With the Ethics commission site a user can, with a bit of work, find the campaign funding data he is looking for. In terms of campaign finance data, the commission, like most municipalities, has focused on the input-side-- replacing the old paper forms with on-line equivalents and thus complying with legislation calling for the use of on-line filing systems. Less attention has been given to outputallowing the public to visualize the data. Currently, a user can search the resulting database to retrieve forms and summaries. Though it is more than what most local municipalities provide, the site is not ideal for analyzing datait does not allow a user to quickly view the key officers and candidates, navigate the money trails inherent to campaigns, or provide a graph view of multiple entities and funding relationships. As a simple example, when the user views the committees that have funded Mayor Newsom, the user cannot then click on one of the committees to see who has funded them. [cut or move following p] Furthermore, only data involving committees and individuals who file in San Francisco is provided-- one cannot view data concerning lobbyists or consultant filings, or data for those who file at the State level or other jurisdictions. Aware of the potential for this on-line data, as well as the lack of resources available to the Ethics Commission, Commissioner Joe Lynn envisioned a university-city collaboration. He approached representatives from the Leo. T. McCarthy Center for Public Service and the Common Good at USF. The center is a focal point of USFs service learning initiative, and the universitys mission of educating hearts and minds. His idea was met with enthusiasm from the Centers namesake, Leo McCarthy, as well as its board member, former mayor of San Francisco Art Agnos. The missing link was the connection to technology. Centers like the McCarthy Center, and public service efforts in general, have traditionally been concerned with humanitiessending students to work on political initiatives or inner-city soup-kitchens. Fortunately, USF had emphasized service learning in the sciences as well, perhaps most prominently in the computer science department. The McCarthy center had already helped fund the departments Community Connections effort (www.usfca.edu/cc), which regularly sends students into inner city computer centers to provide information technology services and annually sends a group to South America to build and maintain computer labs at needy schools. The project is an example of how city governments and local universities can partner to serve the public. City governments are short on funds and city information technology departments are typically overrun and debilitated by the skyrocketing cost of developing and maintaining software. Computer science departments have skilled software developers who can benefit from the experience that a real-world project provides. Students gain a sense of how they can apply their powerful skills to help people, and they learn more because theyre motivated. The partnership with the City Ethics Commission is an experiment in a software development project with similar public service goals. The goal of the Transparency in Government project is to build software that addresses some of the limitations of the current system. The current version of the software, found at whosfundingwho.org, takes raw data supplied from the Ethics commission, builds a relational database from it, then displays the data in forms that allows a user to easily view campaign activity and funding trails. We have also implemented an on-line lobbyist filing system and plan to do the same for political consultants. The plan is to then provide a single associative view of campaign activity involving committees, lobbyists, consultants, and individuals. SYSTEM OVERVIEW The key contribution of whosfundingwho.org is to help users do what Deepthroat suggested to Woodward and Bernstein in the Watergate investigation: follow the money. The system provides two methods of navigating such trails: a table-based method that displays the data for a particular entity and allows quick navigation to the page for another entity, and a graph view that provides a birds-eye view of multiple entities and chained relationships. Key Offices Page Search The initial page is a search screen that allows a user to type in one or more letters of an entitys name, and receive a list of entities that match. The user can also choose to only view individuals, committees, or organizations (other than committees). Figure 1 is a screenshot of a user searching for all entities whose name contains the string Peskin. In this case five different entities are displayed including the individual Aaron Peskin and the committee Aaron Peskin for Supervisor. Table-Based Navigation When the user selects one of the entities from the search screen, data concerning that entity is displayed in a tabular format, as in Figure 2: In this example, the Incoming Funds for Aaron Peskin for Supervisor are displayed. The user can change the transaction type to view Outgoing Funds or All. There are also subcategories within incoming and outgoing funds so that the user can view only contributions, expenditures, or loans.The data is sorted by amount. Each row in the table displays a funding relationship, which is a summary of all transactions between two entities. Sometimes a relationship consists of just a single transaction, e.g., the first row shows a single transaction between Aaron Peskin for Supervisor and Minhall Inc.. Sometimes two entities have shared more than one transaction. For instance, the table above shows that there are two transactions between James Rueben and Aaron Peskin for Supervisor. Instead of displaying a date for that funding relationship, the system displays a link which allows the user to view all of the individual transactions (and their respective dates). Unlike many current systems, including the one at the San Francisco Ethics site, the user can navigate a money trail by selecting one of the entities listed. When a link is followed, the system displays transactions of all types Graph-View While a table provides detailed information concerning a funding relationship, it shows funding trails at only one degree of separation. The graph view available at whosfundingwho.org provides a birds-eye view of a group of entities and funding relationships (See Figure 3): The graph shown in Figure 3 was invoked by selecting Graph View from the Aaron Peskin for Supervisor tabular view page. Both incoming and outgoing funding relationships are shown and color coded. The graph allows the user to view multiple-degree relationship chains, e.g., Peskins committee gave to David Binder Research who also received money from Yes on C. The user can click on a node to expand it and follow the money trail. Eventually, such a graph could be designed so that relationships other than just money flow could be shown. For instance, each individual could point to an organization which would show its money flow, or a committee might be connected to the lobbyists it has hired. The key to implementing such views lies in integrating data from various databases, as will be discussed later. Issues with the Current System Eliminating Unwanted Aliases Electronic filing greatly improved the tracking of campaign funding. The old paper system made it difficult to track the flow of campaign cash . . . The candidates often did their best to keep the public in the dark. Former Governor Mario Cuomos reports regularly included handwritten entries, some illegible. [Governor George] Pataki filed printed reports, but used extremely small print and alphabetized his list of contributors for a time by first name. Marc Humbert[x] Electronic filing has not eliminated all barriers to openness, however, as most data is still input by typing it into unstructured text fields. This allows filers to wittingly or unwittingly input the same data differently at different times so that even an extra space or comma can lead to problems. For example, in San Franciscos campaign finance data 2004, Haight Street Mortgage appears as: Haight Street Mortgage Haight Street Mortgage Co Inc Haight Street Mortgage Co., Inc. Haight Street Mortgage, Inc. A human can easily see that the text strings all refer to the same entity, but a computer program cannot make that assumption, so data will be displayed incorrectly. This is especially problematic in an associative viewing system where data is not always listed in alphabetic order. For instance, if the user viewed the contributions page for Aaron Peskin for Supervisor he would see Haight Street Mortgage Co., Inc. as a contributor. If he then clicks on the link for Haight Street Mortgage Co., Inc., Aaron Peskin for Supervisor is reported as the lone receiver. Only if he knew of the additional spellings would he learn that Haight Street Mortgage contributed to many other entities. It should not be assumed that such misinformation is due to wrong-doing. In this case, Haight Street Mortagage had no part in the problem-- various other entities filed received payment forms and typed in the mortgage companys name in slightly different ways. The root of the problem is that many on-line input forms are just paper forms directly transferred to the computer. With such PDF-like forms, the user can only enter text in boxes and cannot choose from a list of existing entities. This is an example of a more general problem in human-computer interactionprogrammers modeling the on-line world too closely to the paper world and not taking advantage of what is possible virtually. Web applications, as opposed to PDF-forms, can connect directly to the live database and allow the user to choose from existing entities. Such applications can also notify the user, as he types, that an entity with a similar name exists. Such simple facilities can reduce the count of unwanted aliases significantly. Data can also be processed after the user input phase to find unwanted aliases. Such processing should be a joint effort between software and humanthe computer can flag similar names and ask the administrator to make the final call on whether the similar names refer to the same entity. Typically, this work would be completed by a government administrator such as the employees of the San Francisco Ethics Commission. Free-text input and the resulting unstructured data also reduces the users ability to query the data in interesting ways. For instance, if a user enters text concerning the type of organization he belongs to, it is likely the database will end up with thousands upon thousands of organization types. As with the example involving names above, users will enter different text to mean the same thing. Instead, where appropriate, the filer should be given a set of choices. The resulting database can then be queried to correctly find all entities of a given type. First Page Copyright Notice Please leave 3.81 cm (1.5") of blank text box at the bottom of the left column of the first page for the copyright notice. Subsequent Pages For pages other than the first page, start at the top of the page, and continue in double-column format. The two columns on the last page should be as close to equal length as possible. Table  SEQ Table \* ARABIC 1. Table captions should be placed above the table GraphicsTopIn-betweenBottomTablesEndLastFirstFiguresGoodSimilarVery well References and Citations Footnotes should be Times New Roman 9-point, and justified to the full width of the column. Use the standard Communications of the ACM format for references that is, a numbered list at the end of the article, ordered alphabetically by first author, and referenced by numbers in brackets [1]. See the examples of citations at the end of this document. Within this template file, use the style named references for the text of your citation. The references are also in 9 pt., but that section (see Section 7) is ragged right. References should be published materials accessible to the public. Internal technical reports may be cited only if they are easily accessible (i.e. you can give the address to obtain the report within your citation) and may be obtained by any reader. Proprietary information may not be cited. Private communications should be acknowledged, not referenced (e.g., [Robertson, personal communication]). Page Numbering, Headers and Footers Do not include headers, footers or page numbers in your submission. These will be added when the publications are assembled. The Future of Political E-Identity One goal of e-democracy is to use information and communication technologies to better inform citizenry on how to vote. Journalists and other interested citizens can now forage for information at sites put out by the politicians themselves, at non-partisan sites that attempt to provide objective information about our (potential) leaders, and finally to blogs and other participatory sites that provide a forum for public discourse. Key data points of a politicians public record include campaign finance information, information about the lobbyists or consultants hired by the politician, the politicians voting record and stand on issues, the politicians appointees and appointers, the organizations the politician has awarded government contracts to, and the politicians employment records. Much of this information exists on the web, but it is scattered. Compiling a politicians public record takes a single investigator days, weeks, or even months to collect it from the various web pages on which it resides. And because the information is not structuredit is free textsoftware cannot perform such a task in an automated manner. There are two strategies for tackling the problem: one is to induce organizations to publish information in a standardized XML format. If data were published in this manner, instead of as free text reports on web pages, then automated software could collect, process, and display it in various ways. Such a strategy is an example of a general movement in information systems to a more semantic web []. The second strategy involves collaborative tools that can harness the time and effort of the many politically interested individuals now roaming the web. The idea here is that, until the day when all information sources publish data in a standardized xml format, there will be a need for humans to read web pages and extract the data that is pertinent to a politicians public record. Whereas a single individual cannot perform the research necessary to extract pertinent information from many web sources, the collective could. Such public collaboration has gained wide acclaim with the success of such efforts as wikipedia.org, an on-line encyclopedia that allows anyone to create and edit content. But can a collaborative strategy be used to collect and organize political information? And is it possible to harness the power of the many while also staying objective? 2. Current Status of E-Politics Campaign finance data is the most popular form of electronic information concerning politicians. Such information can now be found at the federal, state, and local levels. At the federal level, opensecrets.org has been rated the top site but the list includes the Federal Elections Committee (FEC) site [] and x. At the state level, California and Washington were top rated. These sites display the amount and source of a politicians contributions, including those from organizations, individuals, PACs and even xxx. Figure 1 is a snapshot from the opensecrets.org site. All data found on the site is objective in the sense that it was filed by politicians and committees as required by campaign disclosure law. Such sites provide summaries and listings and to some degree allow the user to follow the money trail, i.e. to explore association chains amongst entities. opensecrets.org excels at summarizing data in various waysone can view a politicans contributions categorized by organization type, by business sector by state, by Individual/PAC, by business/labor. The site also organizes the politicians by office so the user can easily compare two candidates in a race for a particular office. The FEC site is stronger than opensecrets.org in terms of allowing the user to follow associative chains. For instance, when the names of a politicians contributors are displayed, they are displayed as a link so that the user can easily follow the money trail. Neither opensecrets or fec.org provides information or links to information other than campaign finance data. Whosfundingwho.org is a site that focuses on a local jurisdiction, the city of San Francisco. The site shows both campaign funding and expenditures, thus allowing the user to see how a politician/committee is spending its funding. The site also plans to integrate lobbyist and political consultant data within a single user interface. Whosfundingwho.org provides associative linking, like the FEC site, and it also provides a graph view which gives a birds-eye view of funding relationships involving multiple parties (see Figure 2). 3. Semantic Web Tim Berners-Lee, one of the founders of the web of today, is the leader in the movement towards the creation of a semantic web [], one that consists of structured data and not just web pages. These ideas are important both in terms of information sights providing web service access to their data, and providing input forms that enable the creation of structured unambiguous data. 3.1. Web Services Of the campaign finance sites discussed above, only the FEC site publishes data in a structured format that can be read by computersthe others provide only human-readable web page reports in HTML format. And the data published by the FEC site is in the form of a single downloadable file, not a web service, so the only way to use it is to download it in its entirety. Web services are distributed programs that allow client software to query another computer for particular data. Unlike normal web page requests, which return HTML, web service requests return data in a machine-readable XML format. In essence, web services provide the data without all the presentation formatting of HTML, which is hard for a computer program to ignore. [example of some xml] If the FEC site provided web service access to its information, a client program could request, for example, all the contributors of a particular politician. The results would be returned in a structured manner, without presentation formatting, so that the client program could process it, combine it with other data, and display it in any way it wants. Such web services are the key to the interoperability problems involving databases owned by various sources and in various formats. If interested parties can agree on a standard xml format for queries like return all contributors of politician X, then client programs can query different web services, e.g., the federal campaign finance service, the local campaign finance service, the lobbyist service, and a board of director service. Programs can then process the various data and display it in a comprehensive way. A user could then go to one site, enter a politicians name to view all information about the politician, and be able to follow associative trails, e.g., view information about the politicians lobbyists. 3.2. Structured Input The push for mandatory electronic filing did improve the state of campaign information. Consider this statement: The old paper system made it difficult to track the flow of campaign cash . . . The candidates often did their best to keep the public in the dark. Former Governor Mario Cuomos reports regularly included handwritten entries, some illegible. [Governor George] Pataki filed printed reports, but used extremely small print and alphabetized his list of contributors for a time by first name.[Marc Humbert] But if input is based primarily on typing free text into unstructured text fields, the problems with paper filings can persist in the electronic world, as filers can wittingly or unwittingly refer to the same entity differently at different times. For example, in San Franciscos campaign finance data of 2004, Haight Street Mortgage appears as: Haight Street Mortgage Haight Street Mortgage Co Inc Haight Street Mortgage Co., Inc. Haight Street Mortgage, Inc. A human can easily see that the text strings all refer to the same entity, but a computer program cannot make that assumption, so data is displayed incorrectly. This is especially problematic in an associative viewing system where data is not always listed in alphabetic order. For instance, if the user viewed the contributions page for Aaron Peskin for Supervisor he would see Haight Street Mortgage Co., Inc. as a contributor. If he then clicks on the link for Haight Street Mortgage Co., Inc., Aaron Peskin for Supervisor is reported as the lone receiver. Only if he knew of the additional spellings would he learn that Haight Street Mortgage contributed to many other entities. It should not be assumed that such misinformation is due to wrong-doing or purposeful. In this case, Haight Street Mortgage was not involved in the various filings-- various other entities filed received payment forms and typed in the mortgage companys name in slightly different ways. Part of the problem is that many on-line input forms are just paper forms directly transferred to the computer. With such simple PDF forms, the user can only enter text in boxes and cannot choose from a list of existing entities. This is an example of a more general problem in human-computer interactionprogrammers modeling the on-line world too closely to the paper world and not taking advantage of what is possible virtually. Of course input forms can connect directly to the live database and allow the user to choose from existing entities. Such applications can also notify the user, during the text entry process, that an entity with a similar name exists. Such simple facilities can reduce the count of unwanted aliases significantly. Data can also be processed after the user input phase to find unwanted aliases. Such processing should be a joint effort between software and humanthe computer can flag similar names and ask the administrator to make the final call on whether the similar names refer to the same entity. Free-text input and the resulting unstructured data also reduces the users ability to query the data in interesting ways. For instance, if a user enters text concerning the type of organization he belongs to, it is likely the database will end up with thousands upon thousands of organization types. As with the example involving names above, users will enter different text to mean the same thing. Instead, where appropriate, the filer should be given a set of choices. The resulting database can then be queried to correctly find all entities of a given type. Allowing a user to choose from existing entities can provide significant improvement, but that in and of itself does not eliminate the potential for ambiguity. Semantic web proponents also consider issues of identity through the use of Uniform Resource Identifiers (URI). URIs are like URLs in that they are globally unique identifiers, but unlike URLs they do not necessarily map to a file on some server. What URIs can provide is help in uniquely identifying real world objects like people. Instead of referring to David Wolber in a web page to refer to the author of this paper, one could refer to cs.usfca.edu/wolber which is the unique uri for David Wolber, the USF professor. Of course filers cannot be expected to know the URIs of individuals. What is needed is a global name service which client programs can access, and which provides a mapping between basic information (last name, first name, etc.) and a URI. When a filer enters a name, this service would be queried to return a list of matching known individuals. The user would then be allowed to choose from the list. Beneath the surface, the software would identify the individual using the URI from the service. [note could have a political name directory (instead of waiting for a global one). Also note relationship to dns] 4. Public Participation The previous section discussed one strategy towards building comprehensive political information sites: inducing information sites to publish their data in web service form, which would allow other software to access data from various places in order to provide comprehensive views of a politicians public record. But such an automated strategy is dependent on many organizations agreeing to publish data in this way, then agreeing on a standard, and finally actually implementing the changes. Such a process may take years and it is doubtful that all pertinent information will be made available in xml form. The key problem, remember, is that most data is in web page report form, which is understandable by humans but not software. A human can read it and understand it, and if given the time could research thousands of pages on the web and come up with a cohesive collection of data. For instance, Josh On researched the web pages of hundreds of companies to compile a comprehensive list of the board of directors in corporate America. He then made this information publicly accessible at the website theyrule.org. Unfortunately, humans dont have unlimited time to perform such work. theyrule.org, for instance, has only 2004 data and is not kept current. Can software perform such a task? Well, as discussed, software is best able to read structured datadata in pre-defined tables and fields. Natural language processing has progressed, but is not near the point in which a software program could crawl the web to discover the websites of all Fortune 500 websites and extract the board of directors information from them.[especially as those sites change or restrict crawling] One solution would be to require corporations to file a list of their board of directors, and then have the government make that information available in a machine processable XML format. Short of automation, there is also the strategy of engaging the masses. There are thousands of intelligent people roaming the Internet along with emerging collaborative tools which can tap their enormous potential. One prime example is Wikipedia (www.wikipedia.org), an on-line encyclopedia that is constructed and edited by all that choose to contribute. Another is de.li.co.us, an on-line bookmarking system that allows users to categorize web pages and together create a folksonomy of popular web pages. A third example is Slashdot [] which allows users to submit and rate technology stories. The basic idea in all of these systems is to tap the power of the technorati on the Internet by letting everyone become not just consumers of information but publishers of information (prosumers). The power of such collaboration has barely been tapped in the political world. politicalfriendster.com [] is one of the earliest political sites allowing the public at large to enter data. It allows any user to add a new person or organization to the database and add associations between entities, e.g., George H. Bush is senior advisor to the Carlyle Group and a close friend of Prince Bandar Bin Abdul Aziz. Such relationships probably cannot be found in any existing database, only in news articles and documentaries. By entering such relationships into Political Friendster, a user brings them into the structured data world and potentially to any software that focuses on displaying political relationships (political friendster currently does not publish its data in xml format, however, so one can only view the associations through that website). The key issue with participatory political sites is the sensitivity of political information and the potential for erroneous information to be input. With Wikipedia, erroneous information might steer a 5th grader in the wrong direction, while erroneous political information could effect an election. But this doesnt mean participatory sites for entering structured data should be disallowed. Political blogs have even worse potential for erroneous information as users can enter any unstructured text, though one could argue that the associations in PoliticalFriendster seem more objective than blogs because of their form. The issue is to determine an appropriate level of site moderation and whether the moderation should be administered by a trained individual, the users themselves, or a combination thereof. Slashdot, the technical news site, takes the latter approach. A team of individuals from the organization chooses which stories to publish from all those input by the public. Users then comment on stories. Because a story can have hundreds or even thousands of comments, a reputation mechanism is provided whereby users rate each others comments. Those that have had highly rated comments in the past are given high reputations and their comments are listed more prominently. Because users themselves help moderate the site, a significant burden is removed from the site maintainersSlashdot would have to hire thousands of employees to keep up with the work. As non-partisan political information sites are often maintained by non-profit organizations, it is likely that funding will be scarce and public moderation will be a necessity. The reader is referred to [http://www.firstmonday.org/issues/issue9_7/masum/] for a discussion of reputation and moderation schemes. ACKNOWLEDGMENTS Our thanks to ACM SIGCHI for allowing us to modify templates they had developed. REFERENCES Bowman, B., Debray, S. K., and Peterson, L. L. Reasoning about naming systems. ACM Trans. Program. Lang. Syst., 15, 5 (Nov. 1993), 795-825. Ding, W., and Marchionini, G. A Study on Video Browsing Strategies. Technical Report UMIACS-TR-97-40, University of Maryland, College Park, MD, 1997. Frhlich, B. and Plate, J. The cubic mouse: a new device for three-dimensional iput. In Proceedings of the SIGCHI conference on Human factors in computing systems (CHI 00) (The Hague, The Netherlands, April 1-6, 2000). ACM Press, New York, NY, 2000, 526-531. Lamport, L. LaTeX Users Guide and Document Reference Manual. Addison-Wesley, Reading, MA, 1986. Sannella, M. J. Constraint Satisfaction and Debugging for Interactive User Interfaces. Ph.D. Thesis, University of Washington, Seattle, WA, 1994. Columns on Last Page Should Be Made As Close As Possible to Equal Length     PAGE  9   - 5 z4:<=>~p~]$jhh?B* Uphhh?5B* \phhh?B* phjhh?B* Uph h?] h?CJh?6CJ] h?CJ]h?0J OJQJ^Jjh?OJQJU^Jh?OJQJ^J h?6] h?5\ h?5CJh`wlh?@ h?@h?$89F  - 6 a n z=1+Dgd?x=1H&#$+D/0$gd?xxxgd?$a$gd?<izy,w-*""% %0%&'gd?xgd?=1H&#$+D/0$gd?=1H&#$+D/0$gd?>LRS["w"(( ++a.b.22273444^6t6u666666U8u888BBBBBBC0CMMOO^Q׹ײץץיייייבבׁyqhh?6h-@h?6 h?5\h?mHnHujh?U hE %h? h?] h?6] h?0J" h4h? h%h?h%h?6 h|h?hQjh?6h?jhh?B* Uphhh?0J 5\hh?0J +'' '()))**z-{-3.b.m./9122263734444]6^gd? 7$8$H$^gd?xgd?gd?]6^6u6666699::D<E<=='?(?\AxAABBBCC $$If`a$$`x`x^gd?gd?CC)C0C1C8CW?W@WXYYYR[d[\H^^^_bbgd?x^Q~Q>WYYYR[c[bb!cdddfpuus}}~/8_aZOsϋ{ƌnjЌ5fijlmoprsuv|}~ h?0Jjh?0JUjh?Uh?6]y( h?6]h Gph?H*h Gph?6h Gph?]h Gph?6] h Gph? h?5CJ hh?h? h?56b c!cddff*fHfifff:i[j lDmdnp0rHsuuu*x+zza|}~;gd?;„Ć#) iklnoq  & Fh^h` & Fx`xgd?qrtu~&`#$ &P/ =!8"8#$%) 0&P:p?/ =!8"8#$%+ 0&P/ =!8"8#$% P + 0&P/ =!"#$T% P - 0&P:p?/ =!"#$T% + 0&P/ =!8"8#$% P # 0&P/ =!8"8#$%DyK www.sfgov.org/ethicsyK 8http://www.sfgov.org/ethics$$If!vh555~5#v#v~#v#v:V l6 tc0,55~554$$If!vh555~5#v#v~#v#v:V l6 tc0,55~554$$If!vh555~5#v#v~#v#v:V lU tc0,55~554#F@F Normal $Pa$CJ_HmH sH tH V@V Heading 1$$ & F(@&a$ 5CJKH8@8 " Heading 2  & F@&B@!B Heading 3  & F@& 56CJ8@18 Heading 4  & F@&^@^ Heading 5*$ & F(@&^`a$6CJT@T Heading 6 & F<@&6CJOJQJkHL@L Heading 7 & F<@& OJQJkHP@P Heading 8 & F<@&6OJQJkHT @T Heading 9 & F<@&6CJOJQJkHDA@D Default Paragraph FontVi@V  Table Normal :V 44 la (k@(No List P&@P Footnote ReferenceCJH*OJQJkH>O> Author$a$CJOJQJkHPOP Paper-Title $xa$5CJ$OJQJkHFO"F Affiliations$a$ CJOJQJJ@2J  Footnote Textp^`pCJOB Bulletp & Fp>Th^`p4 @R4 Footer  !,Ob, E-Mail<LOrL Abstract$ & Fx@& a$5CJ;@ List Number 3p & F8>Th.^8`POP Captions$H#$+Dp/0$a$5CJ>O> References$ & Fa$.)@. Page NumberHC@H Body Text Indenth`hRYR  Document Map-D M OJQJ^JB"@B Caption$a$5\^JaJtH TB@T Body Text#/H&#$+D@/0$CJ6U@6 Hyperlink >*B*ph4@4 Header ! !>O!> B< Char5CJKH_HmH sH tH 9?Q@Q  " M z  ֮ #zzzzzzz9 4{'0;=DOP?Q@QL\ fsGp\l:]89F-6anzy , w -* 0 !! !"###$$z'{'3(b(m()9+,,,6-7-....]0^0u000003344D6E677'9(9\;x;;<<<===)=0=1=8=<=A=G=H=P=U=]=g=h=i===Q?Q@QRSSSRUdUVHX^XY\\ ]!]^^``*`H`i```:c[d fDgdhj0lHmooo*r+ttavwx;z}~Ā#) iknqt0000000000ɀ000000 00a0a0a0a0a0a0a0a0aɀ0aa 00ɀ0p0u0p000p0 0`0p0Ж0Ж 0 0  0   0   0  0! 0  0"0"0"0"0"0"0"0" 0  0b( 0b(00b( 0  0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0  0\; 0  0;0;0;80; 0; 0; 0; 0; 0; 0; 0; 0; 0; 0; 0; 0; 0; 0; 0; 0  0i=0i=0i= 0  0#A 00A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A0A 00 0 0 0 0 0 0000\>00000@0@0@0@0\~00D0P !"###$$z'{'b(m(),6-7-....]0^0u000003344D6E677'9(9\; ]!]^^``*`H`i```:c[d fDgdh@0$*@0$@0o&*@0$@0(@0(@0(@0(@0(@0(@0(*@0$@0m. @0@0X@0X@0Xh@0Xh@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X^~0Q0@0X@0X@0Xh@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0X@0XL, >^QHLSz']6CGCgCGb;qIKMNOPQRTUVJ=R<<<XX t !Ob$.뜡`qyb$/7=*1b$Ϲ:1P?yw?b$_R?|t*o4W_U$@ 0(  B S  ?u>n<A%*t~h"n"""""##$$%%%&&&))**..90?0^0d0u0{000000262U2[22222O3U33333KCPC4O?OHQSQ^^__``*`0`H`N`i`o`aab b\bbbbbbccckk5l9llmm!mxm|mooww xxxxxxxyyy/z8z{{t|~|}}/~8~). ƆȆІ҆)0234fg hiz(%%3(6(<<LLM!MNN@QVQIXPX^^jjkk:o>o(v*vi::::::::::::::::::::4  =M2Ή!jo hh^h`OJQJo(@.@.@..@...@ ....@ .....@ ......@ .......@ ........88^8`o(. ^`hH.  L ^ `LhH.   ^ `hH. xx^x`hH. HLH^H`LhH. ^`hH. ^`hH. L^`LhH.hh^h`CJOJQJo([]!jo=M        8E0===)=0=1=8=<=A=G=H=P=U=]=g=h=#AAa0@`R @UnknownGTimes New Roman5Symbol3 Arial3Times;Helvetica5 Tahoma5 Miriam"Ah\drfPkF*Bm8Dq9!24dІ#o3qHP?ZProceedings Template - WORDEnd User Computing ServicesUSF ITS     Oh+'0 $ @ L X dpx'Proceedings Template - WORD8u`End User Computing Services8u`NormalUSF ITS4F Microsoft Word 11.1@| @`j*@@ Bm ՜.+,D՜.+,H hp|  'ACM8І Proceedings Template - WORD Title, 8@ _PID_HLINKS'A /Dhttp://www.sfgov.org/ethicsyqhttp://www.acm.org/class/1998/  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWYZ[\]^_abcdefghijklmnopqrstuvwxyz{|}~Root Entry F#Data X1Table`>WordDocument(SummaryInformation(DocumentSummaryInformation8CompObjX FMicrosoft Word DocumentNB6WWord.Document.8