Web conference notes, 2020.11.12 (Joint Working Group City Services)
Joint Working Group - City Services
- Every other week call at 9am PT / 12pm ET / 6pm CET
Meeting ID: 671 331 832 - https://zoom.us/j/671331832
One tap mobile: +19294362866,,671331832# US
Dial by phone: +1 929 436 2866 (US) (Find your local number)
Add your own name, link, and organization during call
- Michael Schnuerle, OMF
- Wiliam Henderson, Ride Report
- Henri Bourdeau, BlueSystems
- Bill Dirks, Ride Report
- Stephen Zoepf, E&A
- Dirk de Kok, Spin
- Joseph Yawn, ARC
- Jascha
- Joan
- Sean
- 20 attendees
Release preparation meeting #2
-
Metrics
- State of feature - Michael/Sean
- k-value discussion - Joan
-
Geography-Driven Events
- Recent updates - William
- City discussion on disaggregated data
- Jurisdictions if time - Brian
-
Release 1.1.0 Status
- 1 week to finalize release
- Will cut features instead of extend release timeline, per the Working Group Steering Committees' Consensus
- Next week's meeting will be a release review
Notes and or transcript of the call with presentation, document, GitHub links and calls to action.
Metrics
- K-value presentation applies to geospatial data, so only aligns to Metrics based on geographic slices of data (eg smaller than a whole city).
- K-value can affect policy outcomes, could be too high to make good decisions
- K-value should be variable based on use case needs and audience accessing the data.
- LADOT has built a k-value analysis mapping tool called Flow that can show how much data is affected by k-values sliced by time and geography. Shows that k-value is not consistently good for all use cases, especially when trip counts dropped during Covid.
- Recommendation to have the OMF Privacy Committee look at how to get expert recommendations on k-values and other studies on MDS reidentification.
- Another option is differential privacy, which is high on the complexity scale but good for anonymization.
- William points out the triangle of Complexity, Utility, and Privacy, where you can pick 2.
- Add note to spec: Cities/Providers/Third Parties using metrics need to consult experts on privacy risks and access
- Rec Note: Data from metrics should be considered sensitive just like all of authenticated MDS, and not be published or shared publicly without addressing privacy first.
- Rec: A diagram that shows how Metrics are more privacy sensitive than say trips in MDS since they just show endpoints and no trip lines. More sensitive than open data or final reports.
- Should k-value be a parameter that is passed into the API, to allow 0 for detailed analysis/research or 10 if you want to be extra careful? No, not yet. Need to set a minimum of some sort with guidance.
- Noted that k-value is not the only solution. Eg, some tiny geographies and timeframes could be defined and if someone is taking lots of trips to and from the same locations, that could still be discovered.
- This topic needs more discussions.
GDE - leave comments over in PR about changes, especially merger or start/end locations into just geography change
Jurisdictions - needs a presentation - got cut twice now.
Cities and aggregate data - Michael shared that the Strategy Committee is convening OMF city members to talk about concerns about aggregate data raised by the Board, in order to get consensus and address issues before we get into the release approval phase. May affect release timeline, will know more in a week or so. William offered to make any language changes needed for GDEs.
