Ethics and Identity overhaul

[schanzer]

• [ ] Consider "my data identity" as a first lesson in the DS pathway. Leverage June Ahn's work, along with "Dear Data" and Kathi's opening activity in her DS course at Brown
• [ ] Rewrite the existing "Ethics, Privacy and Bias" lesson to use case-studies and (a possibly-simplified) Data Science Ethos Cycle from ADSA
• [ ] #1335
• [ ] Update Datasheet spec for compatibility with these intiatives, if possible
• [x] #1561

[flannery-denny]

Dig into the ethos cycle

• which portions map onto the datacycle
• which portions map onto the readme
• ethical datacycle worksheet - points out all of the points in the datacycle where you're making an ethical decision
• new lesson plan that returns to the datacycle with an ethical lens (by january)
• dear data anchors data in the real world
• what could you infer about someone based on their gps data (ask kathi for more details)
• emmanuel can send some of june ahn's papers

[flannery-denny]

[flannery-denny]

Framing to think about from the ASDA publication

"The tool operationalizes what we understand to be the responsibility of data scientists: to support flourishing and justice. This means supporting the well-being of communities whose data is engaged or whose lives are affected by data science work as well as promoting collaboration and expanding the boundaries of the research community itself. We also recognize that sound workflow is essential to the ability of data scientists to realize this responsibility (Markham Citation2006, Markham, Tiidenberg, and Herman Citation2018)."

"The lifecycle approach presents an effective means of engaging questions about human flourishing and social justice throughout the process of data science work, actively taking into consideration and inviting the feedback loops and iterations that are part of real-world projects. As such, the tool aims to operationalize ethics, instead of merely suggesting ethical principles—a need that researchers of computing ethics have identified (Morley et al. Citation2020; Ayling and Chapman Citation2021). It is our aspiration that the Ethos Lifecycle will contribute constructively to the development of a data science ethos that recognizes a responsible workflow in which practitioners engage with the human and social dimensions of data science."

"One specific advantage of embedded instruction is that it helps students realize what they can do to influence the social consequences of their technical products (Verbeek Citation2006; Christiansen Citation2014; Joyce et al. Citation2018; Markham, Tiidenberg, and Herman Citation2018; Grosz et al. Citation2019; Baumer et al. Citation2022). In the neighboring field of computer science, educators who have experimented with teaching “responsible computing” report that students learn more effectively when they are introduced to ethical issues at the beginning of their study of computer science and are given opportunities to deepen and build on their knowledge throughout their studies (Mentis and Ricks 2021)."

"The lifecycle is motivated by the normative project of supporting the visions and experiences of those who are disempowered in today’s technological societies"

"Rather than conceiving of ethics as being first about avoiding harm, the world-making perspective on ethics and justice begins by asking what world is built by data science, for whom and by whom (for an example of an approach to data ethics that centers avoiding harm, see Janeja Citation2019).

What distinguishes the Ethos Lifecycle are the combination of the commitment to justice, the orientation to data science as world-making, and a primary focus on supporting data science novices as much as seasoned practitioners in the process of learning about ethical contexts and developing a responsible ethos (in contrast to an evaluative or accounting mode).

Co-production says that our ideas about what the world is—as a matter of fact—and what the world ought to be—as a matter of social values—are formed together (Jasanoff Citation2004). Instead of privileging either the role of social forces or technological forces as explanations for the way that the world is, the co-productionist approach considers the symmetrical interplay between social and technical forces in weaving the fabric of the world.

[flannery-denny]

ASDA LENSES from Paper

Positionality

• What assumptions do I bring to the work based on my personal life experiences?
• What are the opportunities and limits of my own expertise?
• What identities are represented among my data subjects, partners, and stakeholders, and how are these identities linked to privilege and disadvantage?

Power

• Who has influence over the data science work and who does not have influence even while being affected by it?
• Who gets to decide if the work is valid or good and who may dispute that conclusion?
• Whose power is increased through the work and whose power is diminished?
• How does the transformed field of power reorient identities, relationships, and life chances?

Sociotechnical Systems

• What social constructs or assumptions are embedded in my data and presented as fact?
• What actions (and whose actions) are enabled by data science tools and what actions are proscribed or hindered by them?
• Who or what is held responsible when something goes wrong?
• How does my data science project seek to contribute to the modification of an existing sociotechnical system?

Narratives

• What narratives motivated and informed the creation of this data?
• What underlying beliefs inform the idea that data science approaches are appropriate for the question or problem?
• What kind of change is the work implicitly or explicitly intended to bring about?

[flannery-denny]

[flannery-denny]

ASDA STAGES

1. QUESTION: This stage of the data science lifecycle, Question and Problem Identification, defines the research question to be addressed, ensuring the feasibility and scientific rigor of the project. A research question can emerge from multiple insights: a review of existing literature, direct observation, discussions, and interactions with stakeholder(s): communities, researchers, activists, government officials, corporate managers, etc.
2. DATA DISCOVERY: In the Data Discovery stage, researchers identify potential data sources, exclude irrelevant, analytically unfit, and ethically questionable data (Data Screening), then transform and integrate these data into a usable dataset (Data Cleaning) to support the next stage: Exploratory Data Analysis. FD - We do this stage of the cycle for our students - perhaps we could more explicitly engage them with the READMEs? But Joy usually had 4-5 students find their own Datasets and did data cleaning of those dataset with the full class, so maybe we should have an additional form for classes that go that far. Joy had them enter fields that could be filtered out 9999 or "na"
3. ANALYSIS: In the Exploratory Data Analysis stage, researchers investigate specific variables of interest in a dataset. This stage aims to validate whether the research question corresponds with the data collected during Data Discovery.
4. MODELING: In the Use of Analytical Tools (Modeling) stage, the researcher selects and implements analytical tools based on the research question, the intended utilization of the data to support the research hypothesis, and the assumptions required for a particular statistical method. FD - Perhaps we want to incorporate the word "modeling"?
5. INTERPRETING: In the Interpreting, Drawing Conclusions, and Making Predictions stage, researchers distill the results of the Use of Analytical Tools (Modeling) stage, but also may consider the results of this stage in the context of many other stages of the lifecycle, and may involve explaining how a result came about or making predictions about how the research question might evolve.
6. SHARING: The Communication, Dissemination, and Decision Making step involves communicating the results to the research community through conference presentations, journal articles, social media, and other communications venues. This stage also includes communicating results to decision makers, managers, the general public, and other stakeholders – an important element of the research process.

WE COULD ADD THE FOLLOWING TO OUR THREATS TO VALIDITY SECTION

Positionality

• RESEARCHER AND STAKEHOLDER How might your identity, experiences, and personal situation impact your understanding of the project? What about those of project stakeholders?
• CULTURE, RACE, AND GENDER What cultural, race, and gender assumptions might you and/or your stakeholders bring to the research question and project development?
• KNOWLEDGE AND SKILLS What knowledge and skills do you bring to the definition of the research question? Are there others (stakeholders, researchers, etc.) with relevant knowledge and skills that could help define the research question?
• ETHICS Does the project plan incorporate regular checks, discussions, and documentation about the ethical dimensions of the project?

Sociotechnical Systems - FD - I don't think these are particularly relevant to BS students

• INFRASTRUCTURE What human and non-human infrastructure does the research project require? How are these choices made?
• TRANSFORMATIONS What risks, benefits, and responsibilities do your choices of human and non-human infrastructure present? Is this the first time some of these social and technical systems have interacted? What are the implications?

Power

• STAKEHOLDERS FD: I don't think these questions serve BS students Who are the apparent stakeholders for this project? Is it clear why and how these are project stakeholders? Are there individuals or groups who could be considered stakeholders who are currently not?
• PLANNING FD: I don't think these questions serve BS students To what degree were stakeholders included in developing the research question(s) and the research project, and why?
• RESPONSIBILITIES Who is responsible for framing the research question and developing, funding, publishing, and reviewing the research, including its ethical dimensions? How does this delegation of responsibility impact the research and its ethical dimensions?

Narratives

• DOMINANT AND ALTERNATIVE NARRATIVES What is the dominant narrative in which the research question is embedded? How does this impact the research question? Are there alternative narratives you could consider? How might these alternative narratives impact the research question?
• SOCIAL JUSTICE AND PUBLIC GOOD Does your research question include a social justice or public good component? Why or why not? How could this social justice or public good framing impact the research project in the short and long term?

[flannery-denny]

@schanzer please share links to the June Ahn articles you'd like me to integrate into my thinking. (FYI - I had a phone call with Kathi today about her DS course.)

[schanzer]

Data Science Education -- Identity Book Chapter - Google Docs.pdf @flannery-denny this is not yet published, so please do not share

[flannery-denny]

From Book Chapter: framework for how a learner’s emerging occupational identity (Callahan, Ito, Campbell Rea, & Wortman, 2019) as a data scientist could be understood through five components:

1. one’s self-positioning: Imagining my Possible Self in a Field
2. competency beliefs: Perceptions of One’s Knowledge and Skills
3. social capitalRelationships that One Perceives as Beneficial
4. structural opportunity: Access to Experiences that Learners are Afforded
5. navigational understanding: : Knowledge to Take Advantage of Opportunities and Engage with Organizations or Institutions

[flannery-denny]

From Kathi: Data about you that search engines gather and draw on Data that devices (cars) collect about you Matrix of domination and four domains of power Notes on responsible computing learning goals for the course (couple of years old, not updated to include power)

[flannery-denny]

Recommendations:

• [ ] Add a sharing section to the data cycle... the point is to communicate to the world about what we've learned. (FYI - The section of our data cycle called Analyze Data is called Modelling in the ASDA cycle and Joy thinks that makes more sense.)
• [ ] Add Additional Resources links to the NYT Privacy Project article library - perhaps fill out this form permission to link pdfs of the two kathis' class uses - they're from 2019.
• [ ] Add a workbook page that looks at some of the information that Google gathers and infers about users. Include the following from kathi's assignment
• Identify 3 pieces of information that are inferred about you by Google and two possible ways that Google inferred that information (your answer would look like “Y likely comes from a search I made and X piece of information, X likely comes from A and B, …”).
• Pick two pieces of ad-matching information (either the same or different from what you used in the last question). For each, describe one possible use which is beneficial or benign, and one which you believe is problematic, unethical, or harmful (your answer could look like: Google knows I like hybrid alternative vehicles; one benefit is X and one harm is Y).
• Reference at least two of the following criteria somewhere in your response -- Intent/purpose (academic research vs. targeted ads) -- If the data is public, does that mean it can be used in any way? -- Timing (eg. data set from a decade ago) -- Attributes collected (e.g. those protected by law from discrimination such as religion, race, age, or disability) -- Inference potential: What might this information help companies infer about you?
• [ ] add a workbook page that has students brainstorm about a device they use. Include the following from Kathi's assignment:
• Identify an object you use in your daily life that could be collecting data from you to improve its features. -- Name the object (e.g., “my car”). -- What data does the object collect? -- What does the object use that data for? -- What’s a potential misuse of that data, either by the company that gathers the data or by a third-party that could plausibly/legally gain access to that data?
• Perhaps connect to the data that cars collect article she uses with these questions -- What’s something in the article that you found surprising or interesting? -- What’s one example of collected car data that you think the driver should own or control? -- What’s one example of collected car data that you think the car manufacturer should own or control?
• [ ] Discuss which of these we want our materials to address and then audit to see which we are already addressing: What principles of socially-responsible computing do we care about?
• Representations of real-world entities may abstract away details that are relevant for determining the social impact of a system
• A seemingly objective formula can mask social consequences.
• Search can be used either to expand or limit the information one sees. And when it’s used to expand, data voids can become an issue.
• Predictions or decisions made on biased data can perpetuate that bias
• Collected data can be used for positive or negative consequences
• Humans and computers can interpret the same data differently
• System data may be compromised by either intentional attacks or negligence
• Developers determine who can access information and how that access occurs
• Something about the affects of automation
• Technology consumes physical, human, and social resources
• Technology affects users’ interests, preferences, and habit
• Something about relationships between entities that get controlled via technology
• Technical solutions are not always the appropriate solutions
• Something about transparency

[flannery-denny]

@schanzer I think I'm ready to start drafting some of these pages. I'm not sure which lesson to attach them to. Wondering if I should

• make a new ethics branch or stay on master?
• store new pages in the authoring folder of an existing lesson for now or make a new ethics lesson folder (with an empty lesson plan for now) as a place to store pages so that they can get made and we can think about what to do with them later. Thoughts?

[schanzer]

@flannery-denny I think an ethics branch is a great idea!

I'd also like to have you and I meet with the guy from ADSA, so you can share your thoughts and plans with him. It might also be a good opportunity for him to hear about the potential overlaps or mis-matches between what they've developed and what we need for K12. Any objection to me sending an email and CCing you?

[flannery-denny]

@schanzer Is it free to add another user to our bootstrapworld.org email ? If it's not a big expense, I would love to have an account called student@bootstrapworld.org with which to play around with https://myadcenter.google.com/ to make some screenshots for workbook pages.

These are the screenshots from Kathi's assignment and I find them pretty hard to read.

[schanzer]

@flannery-denny sorry for the delay on this! I've create an alias: test_student@bootstrapworld.org. It just redirects to your inbox, but should work for signing up for a service if you want to avoid your named address.

[schanzer]

@flannery-denny just tagging the need for us to send a writeup of options for collaborating with ADSA.