Facilitators from the afternoon workgroups reported from their sessions. Then the floor was open for questions from participants about the workgroups or about general topics regarding the workshop and NICS. (2IMP)
Participant Needs Workgroup Summary (2IMQ)
- We need to clarify what we mean by users at the local level, does it mean local government or residents. We have talked about it as they were one time of user, but we should distinguish as we proceed (2IMR)
- We really haven’t done a good enough job of market analysis for NICS – if you go to the local level, and ask what are your data needs - they wouldn’t be able to articulate them, because they haven’t been exposed to such as system (technology creates demand). Suggestion: do a survey, to get a sense of what such a system as this could be used. (2IMS)
- At the fed there are lots of questions about how to get agencies to share data. OMB requires impact evaluations – which pushes federal agencies towards sharing. As well, GASB34 requirements – new requirements implicitly require the type of data through the type of system NICS CoP is suggesting. If we articulate this connection, we can effectively make our case to federal agencies why a NICS is necessary for business of federal agency and program management. (2IMT)
- GASB and OMB requirements can answer the question “what is in it for the Fed”. The management requirements demand a NICS – or we can argue that for our business proposal. (2IMU)
- Part of NICS responsibly is/should be to advocate ‘captive’ data sets that aren’t quite good enough yet, and detail the type of quality improvements that data needs and why. (2IMV)
- There are often a of locally collected data sets (tax assessor, local property tax), with a varying degree of digitalization. We should put on our agenda the task of compiling a list of data sets that would be helpful for community-level folks, as well as useful to feds (NNIP may be able to already do this. They could list what each of their partner cities use, and generally detail what is available in cities) (2IMW)
- How do we create the top-down and bottom-up feedback loops to address quality standards? Is this built in the technology or in the manner in which data is allowed in the system? (2IMX)
- We need to tell federal users – that there is a language barrier – local users don’t need greater detail necessarily, but smaller geographies. e.g. – Business registry; local-level users don’t need more categories, but do need smaller geographies (2IMY)
- For institutional users – the ability to create custom tabulations (2IMZ)
- NNIP cities have models at the state of organizations with confidentiality agreements, and are using it. (2IN0)
Governance and Finance Workgroup Summary (2IN1)
- We need to think through any and all additional functions to add to a NICS (2IN2)
- Become aware of and use the federal data reference model (designed by the Federal Enterprise Architecture Program) to ensure that their architecture is useable at the local levels. (2IN3)
- We ought to think about how we broaden our concept of stat literacy, and create tools for use in public policy decisions that engender and enrich statistical literacy (2IN4)
- We need critics – particularly informed practitioners regarding privacy and confidentiality issues to get in this discussion. (2IN5)
Governance structure… (2IN6)
- The CoP structure can work in this phase as we put together a business model. As we move into ‘the prototype phase’ we need to think of a different way we are governing ourselves. (2IN7)
- Let’s look at the three options in the GAO report on indicators 1. public, 2. pub/private, 3. private (2IN8)
- Examine types of efforts by NICS in terms of constituents who would provide financial support—i.e. divide services and topical areas as fundable by people of different geographies, issues (2IN9)
- Think through licensing tools as a way to becoming self-sustaining, as NICS moves into its 3rd/4th year (2INA)
- We are creating markets as we built this tool – so we should expect to find new revenue sources later that we don’t see now (2INB)
- We need a workgroup focused on emerging issues to develop a research agenda for NICS– “If we want to lead, we need to know what’s coming down the line.” (2INC)
- We need to develop a model that ensures sustainability – one model might be a training exchange. e.g. the Bonneville Power Authority model. (2IND)
Systems Architecture Workgroup Summary (2INE)
Issues in discussion generally fell into three categories (2INF)
1. Where are we headed? (2IOQ)
- There is a voiced concern that we are building a single system (2INH)
- Is technology an issue? Does current technology match our ambition? (2INI)
- Is a stove pipe/the stove pipe model good or bad? Can we get around them? Can we reshape the market? How do you know when a stove pipe isn’t the right answer (2INJ)
- The divide between public and private roles, is generally of some concern. Do we let the market decide what that division is and what each sector’s role is? Do we shape those roles through supply? (2INK)
1. What is hard to do? (2IOR)
- “Our users want data, rather than finished applications, as well as the ability to integrate data. If you give them the applications then it is easy to integrate data in the future. If we don’t supply applications, people value-add to their version of the data and then it is really hard to integrate with future updates to those data sets” (HUD representative) (2INM)
- XML? Web services? (2INN)
- One size doesn’t fit all. We need to remind ourselves that. Different interfaces for different users. While there will a considerable degree of customization to appropriately meet the need of all the different types of users we envision for a NICS, to save duplication of effort, we will building reusable components becomes crucial. (2INO)
- We should list a catalog of examples of reusable components/places we can save money and time, pitch reasons – this list will help in writing a business plan (2INP)
3. What can systems architecture do? (2IOS)
- Cross referencing tools (master address files), we see these a lot in community statistical systems (2INR)
- Privacy issues. The data LED uses has been around for a long time, but the confidentiality issues has kept the data from being released. They use a statistical innovation to get around this confidentiality issue. Their use of adding noise, is a set of computations that should be considered a ‘reusable component’ (2INS)
- Orthophotos – at this point, there are a lot of orthos, and places to obtain orthophotos out there. Different map projections, different coordinates system remains an issue. (SAVI in Indianapolis has solved this problem with a middleware solution). Every GIS can do it, but they tend to use their orthos in their own absolute universe, rather than a relative one to the larger map. So this is an issue NICS will need to address. (2INT)
- Area allocation tools– to redefine census tracts to more appropriately reflect neighborhoods and areas that are more sensible to the planner, researcher or resident. (2INU)
- Normalization of data, geographies– its hard to get the right denominator for the right geography. That stuff gets hard to do because every one decides their own appropriate base for analysis/ (2INV)
- GIS integration –Often when you combine maps from different data sets, you get conflation--areas of maps that lay over each other and upon integration they duplicate records for actual parcels, residents, households, etc. (2INW)
General Discussions (2IOT)
- We’re not talking about replacing the jumbled market that does exist around statistical services, but we are looking to make it work better and possibly replace existing institutions that exist but are not capable for solving issues in this marketplace (2INY)
- Need more “market research” to get a sense of the need for what NICS might develop (2INZ)
- GASB34 – accounting requirement can drive demand (2IOV)
- Users needs – need to be articulated, and exposed though future demonstration of what a NICS can/could do. (2IO0)
- It is wise to acknowledge the current market for these services and to leverage against existing activities (Paul Bugg) (2IO1)
RoderickHarrison, Participant Needs (2IO2)
- Use NICS to think about how to build common tools to deliver data. This is the issue when 6 different agencies develop a system then you have 6 different tools. This issue is a common discussion between the participant need and systems architecture groups. (2IO3)
- GASB discussion seems to be the way to get the federal agencies the incentive to discuss data sharing. (2IO4)
- We had great concrete examples from our group (Participant Needs)– We need to catalog these. For each of these examples, we should develop specific next steps to find out more information. Who do we talk to? What is happening, and what are next steps for those specific projects? (2IO5)
- For mutli-state sharing agreements, can NICS reduce the work load a program like LED’s by having all the data sharing agreements done at once, before they get there or because of confidentiality requirements is a general data sharing agreement or MOU not possible? (2IO6)
David Seidman (2IO7)
- Data intermediaries will be key factors in this effort – cities use of intermediaries make them a built-in constituency. Intermediaries may have to be licensed in order for confidential data provided to them to remain confidential (2IO8)
- With the tools NICS proposes to build – can we define software standards? Are we able to do that? Do we have to use someone else’s platform? This is a policy and political issue. (2IO9)
Dennis Culhane (2IOA)
- I can be considered a representative of Philadelphia. The Fed should give strong and good advice about disclosure risks for certain ‘cell sizes’. HIPPA says you can report smaller than 20,000. that’s useless for community health planning. Most federal agencies state “all you need is a statistician to tell you that you haven’t violated a disclosure guideline” and that isn’t clear enough, because in implementation HIPPA policy differs from that guideline (2IOB)
- While I was impressed by the commitment, we can’t find much use for the federal data because of the geography issue. County/census track is too big. We are trying to move to a raster-based representation of reality. We are look at 100’x100’ area, with raster rendering, so we can build up to larger geographies (2IOC)
- Can we have confidentiality experts – legal, technical, etc – come and speak to the group? (2IOE)
- Pari – let’s organize a forum on IP and Confidentiality (2IOF)
- There are very good people at census on confidentiality, you can speak to them to get a speaker. CynthiaTaeuber is an expert at this issue. There is a group at Census that meets once a month. Since private data users have little or no restrictions and as people and firms are matching records, the issue of confidentiality is exploding. (CavanCapps) (2IOG)
N.B. Andrew Reamer notes the Census AQS is undergoing system testing. If you want, you can become an alpha/beta tester of the alpha version. Contact Lindsay Clark at the Urban Markets Initiative, The Brookings Institution. (2IOH)
David McMillen (2IOI)
- Be sure differentiate between confidentiality and privacy (2IOJ)
- Census got in trouble b/c they provided locations of arabs in small area estimates – while it didn’t violate any rules, it received criticism because there was a sense that action was not what our statistical system was built to do (2IOK)
- Be wary – you need to bring the Ira Scwartzes (sp?) and Bob Gellmans (sp?) and Center for Democracy and Technology to bring their sensibilities to your discussion to think abstractly about the weight that you (NICS) will bear on its shoulders or the extra weight it will transfer to participants in NICS (2IOL)
- We should follow up on suggestion to write up a document of use cases. (2ION)
- Pari –use march meeting as a place to build a use case document. (2IOO)
end of notes{nid 2IOP} (2IOX)