This document explores the concepts behind how DDOD (Demand-Driven Open Data) can be used in conjunction with FOIA (Freedom of Information Act) requests. It describes how DDOD and FOIA can leverage each other's strengths to help overcome their inherent challenges.
DDOD is an initiative by the U.S. Department of Health and Human Services (HHS) started in November 2014 as part of its IDEA Lab program. The goal is to leverage the vast data assets throughout HHS’s agencies (CMS, FDA, NIH, CDC, NCHS, AHRQ and others) to create additional economic and public health value.
DDOD provides a systematic, ongoing and transparent mechanism for anybody to tell HHS and its agencies what data would be valuable to them. It's the Lean Startup approach to open data. With this initiative HHS can move from measuring Open Data in terms of number of datasets released to value in terms of use cases enabled.
DDOD website: http://ddod.us
1. Demand-Driven Open Data
More info:
Contact
http://ddod.us
David.Portnoy@HHS.gov | tw:@DPortnoy
DDOD for FOIA Organizations
(Freedom of Information Act)
Originally presented in March 2015 to FOIA directors meeting for all agencies under HHS
Concepts explored here do not necessarily
represent the views of HHS.
9/2015
2. Abstract
Industry and researchers can get valuable data through DDOD (Demand-Driven
Open Data) and FOIA (Freedom of Information Act). Both paths have their
advantages and limitations. A subset of FOIA requests can also be worked as
DDOD "use cases". Such requests involve structured, machine-readable
datasets that continue to be generated and are not for restricted data. This paper
explains how combining the two methods by working this subset of requests in
parallel significantly improves the effectiveness both DDOD and FOIA.
This paper was originally presented at the FOIA directors meeting for all HHS
(U.S. Dept of Health and Human Services) agencies in March 2015.
3. Demand-Driven Open Data (DDOD) is a
framework of tools and methods that…
Provide external data users✽ with a
systematic, ongoing and transparent way to
tell HHS what data they need
...To be managed, measured and executed in
terms of use cases, enabling allocation of limited
resources based on value
What is DDOD?
✽ Including industry, researchers, nonprofits, media and other government organizations
More effective open data initiatives
More engaged user community
More economic value & discoveries delivered
4. Add
What does DDOD deliver to the user community?
Implementation of a use case could fall into one of 3 categories
Time to execute
Cost/Effort
Improve
Catalog
Facilitate deployment of
● New datasets
● New APIs
For existing datasets
● Add needed fields
● Improve data quality
● Add / improve metadata
● Add / improve API
If datasets already exist in legacy systems, make
them more available and discoverable
● Publicize availability
● Index to HealthData.gov and Data.gov
● Instructions for efficient FOIA handling
Current
State
5. Processes for administration of use cases, such as
• Encouraging responsiveness, transparency and documentation
• Ensuring use cases and resulting datasets are indexed in HealthData.gov
Specialized tools for administering use cases
• Workflow engine, communications method, knowledge base
• Data processing, storage, hosting, versioning
Proactive outreach to industry and academia for a thriving
community
What does DDOD provide to data owners?
1
2
3
6. “The Freedom of Information Act (FOIA) is a law that
gives you the right to access information from the
federal government. It is often described as the law that
keeps citizens in the know about their government.”
What is FOIA?
Source: FOIA.gov
7. U.S. Department of Health & Human Services
What does FOIA look like within HHS?
Source: FOIA.gov
Agency Received Processed Pending
ACF 1,864 1,255 763
CMS 26,361 25,027 4,717
OIG 843 815 34
CDC 1,141 1,028 691
FDA 10,224 10,191 2,896
HRSA 355 352 92
IHS 152 160 20
NIH 1,169 1,156 84
SAMHSA 193 238 51
OS 1,531 1,605 314
OASH 437 455 120
ACL 13 11 2
Total 44,283 42,293 9,784
8. Some FOIA requests look quite similar to DDOD use cases
FOIA requests that have all of the following attributes could be potential DDOD use cases
1. The data is still being generated or refreshed on a regular basis.
(Not a single pull of historical information.)
2. The data can be delivered in a machine readable, structured format, such as CSV,
JSON, XML or Excel files.
(Not freeform text-only, PDF, or scanned images.)
3. The data has widespread usefulness for multiple organizations.
(Excluding complaints and investigations, which are often specific to an individual
or organization, as is often the case in lawsuits.)
9. Although only a fraction of the 45,000 FOIA requests
received so far by HHS have attributes that make them
applicable to DDOD, there are still plenty that could become
high value use cases.
10. But both FOIA and DDOD have their own challenges
DDOD’s challenges:
● DDOD is a relatively new program that doesn’t enjoy the awareness, recognition and credibility
that FOIA has developed over the years
● Proactive outreach to the user community is a big effort and expense for DDOD
● The toolset for managing use cases is early in its lifecycle and is still being developed
● While the DDOD workflow is documented and actively in use, it’s still being fine-tuned
FOIA’s challenges:
● Growing backlog, due to high demand, complexity of some requests and limited resources
● Process for spotting similar requests and posting the response to an electronic “reading room” is
subjective and time consuming
● Difficult to maintain consistency in both format and actual data delivered. Requests may be
subject to interpretation. Each request may be fulfilled by a different employee or contractor
● There’s no mechanism for automatically refreshing results for data that continues to be generated
11. Joining forces on applicable requests is mutually beneficial,
because it can help address many of these challenges.
12. DDOD can benefit FOIA by...
1. Reducing number of future requests
More data indexed and made publically available via HealthData.gov (See diagram)
Data that’s automatically being refreshed doesn’t need to be requested multiple times
Rework stemming from inconsistent format and data values between consecutive extracts on
refreshed data
Better documentation of data provenance and usage via use case methodology
Data User
runs search on
HealthData.gov
Data User creates
/ updates use case
DDOD Admin
engages Data Owner
on use case
• keywords, subject
• data dictionary
HealthData.gov
Data
Dictionary
Dataset
Inventory
EDI*
Use
Case
DDOD Admin ensures changes to
EDI get propagated to HD.gov
* Enterprise Data Inventory (EDI), which is a catalog
of HHS “Strategically Relevant Data Assets”
Data
Dictionary
Dataset
Inventory
DDOD Admin enters
use case on HD.gov with
links to specifications
Data Owner adds entry
to EDI, including
metadata
DDOD Admin
curates entry &
ensures SLAs
DDOD Admin
creates repository
for use case
Process of adding a new DDOD use case
13. DDOD can benefit FOIA by… (cont)
2. Fulfilling the electronic “reading room” requirement for requests made multiple times by publishing the
data using HealthData.gov infrastructure. Results end up benefiting more users.
3. Instituting best practices for open data projects in terms of machine readability, standards,
documentation and access
4. Helping address the backlog, specifically for the subset of requests applicable to DDOD. These are
also the requests that are beneficial to a wider audience.
Much of the current backlog is unrelated to machine readable datasets that serve the broader user
community. FOIA is straining under the weight of one-time requests that benefit only the few. For
example, a single lawsuit-related request to CDC required processing of 90,000 documents.
14. FOIA can benefit DDOD by...
1. Since FOIA is legislated and well-known, it can bring in more use cases into DDOD.
By identifying specific requests that are applicable to DDOD
By promoting DDOD on FOIA websites and materials
2. DDOD can leverage work being done anyway to fulfill FOIA requests
Data owner already needs to be involved and resources -- whether employees or contractors
-- need to be allocated to do the extract
Once a data extract is performed, it might be trivial to configure periodic runs of the same
code and perform automated pushes to HealthData.gov
3. Association with FOIA lends additional awareness, recognition, credibility to DDOD
15. Requester
receives
documents
Requester
can sue
Data owner
produces
documents
Submit for
fulfillment to
data owner
Requester
can appeal
We start with a simplified overview of the FOIA workflow to determine how DDOD requests
could be initiated and executed in parallel
Agree to
fulfill
Requests that are partially or incorrectly fulfilled
can also go through the same appeals process
Agency decides
whether and what
to release
Agency obtains
clarification
(if needed)
Request routed
to ultimate
agency
Respond within 20
business days
(Can extend for unusual
circumstances: Research,
volume, consultation)
Route within 10
business days
Decline
request
Agree to
fulfill
Decline
request
Requester
submits request
to FOIA
16. In DDOD workflow, use case execution relies on three types of participants: Data User,
DDOD Admin, and Data Owner. The Data User initiates the request
All implementation decisions ultimately are made by the Data Owner
The DDOD Admin only facilitates the process when needed
17. Although it’s a legislation-driven process, FOIA organizations may voluntarily notify DDOD
on incoming requests. For such requests, the DDOD workflow works in parallel to the FOIA
workflow.Data
User
Data
Owner
DDOD
Admin
DocumentImplementEngage Discuss
DocumentEngageClarify
DocumentImplementDiscussClarify
FOIA
Org
Workflow for FOIA - DDOD partnership
Typical
FOIA request