Notes from March 5 meeting at Socrata, Chat with Candace Faber about Hack to End Homelessness and what rattled out of head on the way home: Please note: these opinions are my own and do not reflect anything one way or another for people I spoke to.
What questions / projects to focus on?
--Big global? Small things with potential for big impact? Long duration or ad-hoc that might or might occur more than once, perhaps with iterative modifications?
--What are topical open data resources and what will open data tell us about?
--How can info obtainable through open data channels be applied to improve services, time to get different services, efficiency of processes? How can these improvements be achieved even if hack against homelessness folks would not necessarily be permitted access to organizational data or be able to do matching at client level?
--As a policy analysis / problem framing exercise, what all different privacy provisions apply to different streams of client / customer data
Some big issues:
--privacy / confidentiality: barriers to data sharing.
--client / customer reluctance and options not to share
--fuzzy matches across data sources.
--Think OLAP cube with dimensions by which analysis values can be sliced and diced. That is don't think customer level data
How do we know it works?
--How can info obtainable through open data channels be applied to improve services, time to get different services, efficiency of processes? How can these improvements be achieved even if hack against homelessness folks would not necessarily be permitted access to client level or organizational data?
Does the problem need coding or just people who think like coders?
--story of person who realized that allowing people to complete form online and then print a copy to carry with them visiting different agencies would be valuable: anecdotal comment that dramatically shortened time between enrollment in homelessness program and finding housing.
--how to define ad-hoc analysis variables such as the time interval in the example above?
What might be valuable to service providers whetehr or not there is open data to support:
--tickler lists so staff with direct contact can review with client progress in dealing with multiple agencies and service needs
--OLAP cubes. Did I mention OLAP cubes, analytic datasets for slicing, dicing, summarizing.
--development of administrative processes to use and manage volunteers with specific skills such as enhancing organizational websites
Example analysis question where open data might generate helpful insights:
--Assume a homelessness / social services agency interacts with transportation system and there are data available about transportation delays daily, hourly... How over periods of weeks or months do things like travel times over key parts of the transportation network, or say overcrowing on key bus routes affect clients' ability to move past milestones associated with their programs?
--Would just thinking through this model yield insights that could be used even if no code ever gets written?
--What would be some dollar figures which with right dimensions added could be monitored as other parts of service delivery system work or get adjusted?