ML & Algorithmic Risk - Deloitte Playbook

Under Goal 3: Machine Learning 
<http://stratml.us/carmel/iso/CDOPBwStyle.xml#_6ffd2334-31cd-11ea-842c-c5da1d83ea00> 
- Promote machine learning in government, Deloitte's "playbook" for CDOs 
posits five vectors of progress:

    Objective 3.1: Data Science
    <http://stratml.us/carmel/iso/CDOPBwStyle.xml#_6ffd258c-31cd-11ea-842c-c5da1d83ea00>
    - AUTOMATE DATA SCIENCE

    Objective 3.2: Training Data
    <http://stratml.us/carmel/iso/CDOPBwStyle.xml#_6ffd274e-31cd-11ea-842c-c5da1d83ea00>
    - REDUCE THE NEED FOR TRAINING DATA

    Objective 3.3: Acceleration
    <http://stratml.us/carmel/iso/CDOPBwStyle.xml#_6ffd2b86-31cd-11ea-842c-c5da1d83ea00>
    - EVOLVE TECHNOLOGY FOR ACCELERATED LEARNING

    Objective 3.4: Transparency
    <http://stratml.us/carmel/iso/CDOPBwStyle.xml#_6ffd2dfc-31cd-11ea-842c-c5da1d83ea00>
    - MAKE RESULTS TRANSPARENT

    Objective 3.5: Deployment
    <http://stratml.us/carmel/iso/CDOPBwStyle.xml#_6ffd30d6-31cd-11ea-842c-c5da1d83ea00>
    - DEPLOY LOCALLY

Under Goal 4: Algorithmic Risks - Manage algorithmic risks, the playbook 
proposes:

    Objective 4.2.1: Strategy
    <http://stratml.us/carmel/iso/CDOPBwStyle.xml#_6ffd4094-31cd-11ea-842c-c5da1d83ea00>
    - Create an algorithmic risk management strategy and governance
    structure to manage technical and cultural risks.

Such strategies should be rendered in open, standard, machine-readable 
format.

    Objective 4.2.3: Monitoring & Testing
    <http://stratml.us/carmel/iso/CDOPBwStyle.xml#_6ffd458a-31cd-11ea-842c-c5da1d83ea00>
    - Establish processes for assessing and overseeing algorithm data
    inputs, workings, and outputs.

Such reports should be shared with stakeholders in open, standard, 
machine-readable format.

It will be interesting to see how this group might contribute to 
progress along the lines set forth in Deloitte's "playbook".

BTW, usage of that term -- as well as "policy," "roadmap" and 
"framework" -- is among my pet peeves.  They are poor, immature 
<https://en.wikipedia.org/wiki/Capability_Maturity_Model#Levels>, 
artificially ignorant 
<https://www.linkedin.com/pulse/artificial-ignorance-owen-ambur/> 
substitutes for actual performance plans.

Owen