What's Art Got to Do with it? (weeknote: 2024/09/27)
On my proposal for DAMA Days Conference, Data Maturity, Rules As Code, Ontologies, Knowledge Graphs, and Colonialism & Data and Truth and Reconciliation
I really need to find a way to better integrate these weekly notes into my week. I think I’ve missed the whole summer now. There’s a few drafts that never made it - got too bogged down trying to tell specific stories that it overtook the wanting to share some quick hits from the week and less structured thoughts. To be pondered, in the meantime, I don’t mind a bit of nagging if anyone wants to keep me honest ;)
Analogue Tools for Digital Change
The title of a talk I pitched that was accepted for DAMA NCR-RCN - DAMA Days Canada - Ottawa Conference (dama-ncr-rcn.ca). I still have to write it, of course…
The pitch was something like:
In this talk, Heidi will share insights gained from the interwoven story of growing up as an deeply imaginative daydreamer and highly sensitive odd duck, their non-linear path to a career in data in the Government of Canada, and how that led them to using arts and crafts to teach data.
Since 2023, Heidi has been iterating on a workshop now titled “Paint by Numbers: exploring creativity with data”. Although data can be used to break people into the sum of their parts, but data can also be used in slow, playful, beautiful, and personally meaningful ways. Learners are invited to recognize the data all around them, to play and explore using tools and knowledge they already have.
Each iteration of the workshop has been as unique as the learners have been. It ranged from guided exercises to draw a map reflecting how they imagined the world, to adding individual threads to a collaboratively woven tapestry collectively, to unstructured studio time where the learners could share ideas and experiment with different mediums to create an artwork.
The final message to the learners is Heidi’s hope that they all continue to bring their whole selves to help shape not only a digitally enabled and data driven future, but an inclusive and equitable one.
Something that is missing from the original pitch that only really came to me this morning was how to concretely make the connection between the process of art to the process of “doing” data. Since the theme of this conference is “Data as the Foundation to AI”, I’ve been trying to puzzle out the best way to articulate that linkage. I guess I was processing in my sleep again, I woke Zander up this morning to vomit out the thoughts into words so I could capture it. Because I don’t think in words, I often can only make my ideas “real” by saying it out loud and in writing. This actually really strongly parallels the connection I see between art and data.
To get there, I’ll need to take a step back and tell you what I’ve been up to since last I wrote.
Data Maturity Assessment
I inherited a lot of research on data talent and literacy that have been done over the last 3 years. The summer has been spent trying to figure out how to translate that into prototypes for testing. I also inherited some products that need some updates and iteration.
One of these was the Government of Canada Data Competency Framework (DDN3‑J03) - CSPS (csps-efpc.gc.ca), born from a few years of collaborative development in an interdepartmental working group that was published as a job aid to support departments. In reviewing it, I found that there was an internal conflict within the document between the individual employee and the organizations they work in - something that becomes harder to resolve when I tried to use this framework in a prior job as we tried to find the balance between where the individual interests and career goals and organization’s need to transform for the “age of data and AI”.
I think it’s not a reasonable expectation that individual employees overfunction in order to correct the shortcomings of an organization, or leadership. I say this as an individual who has been overfunctioning for years to my own detriment, with multiple burnouts under my belt. Data literacy is frequently cited as a barrier for building data culture/data maturity, including when we’ve asked in surveys and workshops.
While I think it is certainly a concern, it’s actually a sign of deeper structural problems and there’s danger in overemphasizing improving data literacy in employees and upskilling without ever changing anything else to enable successful integration of the learning and upskilling into their work. You can’t force people to change and people won’t change unless they have to.
I think that’s not really a life changing statement. When the pandemic started, the first few days, weeks, and month certainly saw public servants completely locked out from systems and remote access. And then in a matter of weeks and months, we all got back online, for those of us who worked in the digital transformation space before the pandemic, the rate of new software being rolled out was completely baffling. If it was possible all along… why didn’t it happen until now? We took an emergency driven shortcut to the future that had been advocated for years to no avail. Suddenly all the excuses for why Disabled folks can’t work remotely disappeared as IT processes that use to prevent adoption of chat and videoconferencing that use to make regional employees feel excluded at every single meeting was gone as we got MS Teams in a matter of weeks. Hard to put the cat back in the bag once you show people what’s possible…
So what’s the point here? Let see… Oh yeah, I was getting to how reviewing the competency framework to iterate on it led me to propose a completely different project.
There’s a lot of competencies that are a bit jammed into the individual competency that should not be assessed at that level, let alone ever punishing individuals for. For instance, “1.2.10 Assesses data for bias, representation, accuracy, and validity. Identifies and implements steps to resolve issues if needed.” So I have worked on a lot of projects where I wore a lot of hats, but very rarely are those hats simultaneously, nor continuous. Unless we’re talking about a project that has a timeline of a few weeks or a few months, the scale of government means we’re always passing the baton to one another. If I got a dataset, I can certainly assess a particular dataset for bias, representation, accuracy, validity and try and salvage the output the best I can, but if the data is garbage, it’s garbage. I can’t change the design of the collection tools. I can’t fix how the enabling legislation was written. I can’t demand that decision makers change the business rules they developed and sign off on. That’s if I even know who to call or the provenance of the data.
So let’s say I take a ton of training on data bias. I come back to work and I say, hey, the way we’ve been collecting this data for the last 20 years is biased. Is the response from my management going to be, “Oh you’re so right. We need to scrap it and start from scratch!” In my experience, they probably know and have been fighting for that change the last 20 years and are equally angry about it, they shrug because it’s what we have and the decision is above our pay grade… or best case scenario, “congratulation, you’ve just volunteer to lead a multi-year project where you coordinate hundreds of constantly churning stakeholders who disagree with you about the scope, scale, and priority of the problem to make that change happen.”
To be fair, it’s my favourite response to get, but that’s because I’m like… an autist whose special interest is data/policy/people and doing your special interest as a job is a very good recipe for working 12+ hour days because you genuinely love, believe it, and have a lot of fun doing it and would probably do it for free which is well and good until you stop taking care of yourself and eating properly and start crying daily at work. It’s not for everyone is what I’m saying. The organization probably needs systems in place so that entire isn’t dependent on the shoulder of an 18 to 29 year old that has been struggling for months to teach uninterested colleagues how to cover for them after barely getting approval from their doctor to return to work part time is what I’m saying.
So yeah, I’m working on an data maturity model. The UK Government already published an open source one under a creative commons license that allows non-commerical use, adaption, and publishing: Data Maturity Assessment for Government (publishing.service.gov.uk).
If I had one wish, I’d really like to stop answering questions about why it’s important to have non-proprietary and public sector context specific tools that we’ll never lose access to.
A shadow financial commitment gets its wings every time someone uses propriety software and content. Something something values and ethics stewardship of public monies…. Something something guidance on contracted services….
Rules As Code
I’m working my way back to why learning art is relevant to learning data.
I’m obsessed with ontology at the moment. And particularly, the intersection of ontology, law, knowledge graphing, and AI. Here’s an OECD Working Paper on Rules as Code that’s a pretty good survey of the subject. There’s a team working on Public Sector Experimentation that I’ve become quite good pals with and that’s led to some very fruitful conversations and partnerships on projects. See post from Josh on exploring generating knowledge graphs by pulling RDF triples from unstructured text.
For the Data Maturity Model work, I’m thinking through 2 guiding questions:
Does this [list of features] fully describe a [GovernmentOrganization] of this [Maturity Level] when it comes to this [Topic]? (Rules as Code)
If we ask about [features] as a question, what is the most effective way to find the answer? / who can actually answer? (Evaluating Administrative burden)
What this approach is forcing me to do is try and articulate not only the explicit and implicit rules we tend to work with, but also the “tacit knowledge” that we might understand intuitively or based on years of experiences, but have difficulty articulating. It’s not that intuition or experience is bad or can be replaced, but that squishiness simply does not work when you’re trying to teach a computer something. You have to establish a set of assumptions, relationships, and roles in order to produce an accurately functioning program. Generative AI attempted to shortcut that by asking a program to do it based on statistical reasoning, producing answers that are statistically probable without having to defining the object level relationships. This is fine for many situations that are common, but does not work when situations are novel, exceptional, or rare - ie. making new policies, precedent setting legal cases, eliminating discrimination, etc.
Do not ask me how much time I accidentally spent reading about the nature of knowledge vs information vs data this week… but it’s been very fun. Here’s the one I found the most helpful: Knowledge – Explicit, implicit and tacit: Philosophical aspects by Martin Davies.
Below is my attempt to articulate the relationships and flow between knowledge, data, information and a “generic government organization” and their employees.
The syntax is generally based on RDF 1.1 Primer (w3.org) - all statements about resources are written as “Subject”, “Predicate”, and “Object”. We’ve established our common vocabulary to be Schema.org to minimize the data cleaning as much as possible since there’s a community contribution component to this project. I still have to check the graph for conformity, but this took a deceptively long time given the small amount of nodes to write.
Oh! And when I say write, I mean that (for the draft at least), I’m writing it in Obsidian (markdown files) plus the community extension Juggl. A lot more to say about Obsidian, because I’m building a “second brain” using it that will eventually make its way to my GitHub. Josh was telling me that he had build a vJosh to answer the most frequently asked of him questions using markdown files of the second brain variety. Might help with cutting down time in writing emails to the same questions over and over again…
Now the connection:
Art has 3 components:
knowledge
materials
expression
In order to code, we need to perform an “abstraction” - the attempt to make sense of problems by distilling it down to most smallest element, then putting it back together to convey meaning. In the case of coding, we need to communicate to a computer what an object is, how to find it, and what to do with it once it is found. So while we might be wanting to sort a row of data from biggest to smallest, we communicate that by translating the command into a language it understands. In this example, there would be 3 parts the computer needs to know:
you want to perform the function on an entire row
the row you want to perform the function on is named “X”
the action you want to perform on the entire row named X is a sort
You can also add the order of sorting, but there’s usually a default sorting order if it’s not specified.
When you want to express to someone else how the events of the week made you feel, you can’t literally implant the feeling or the time you spent into their brain. So you might take your feelings and experiences (it might play like a movie behind your eyes, be felt as a tight feeling in your tummy when you hear certain feelings, or feel like you’re back in that moment, or a million other ways), you could turn them into words.
“I had a bad week.” Language in itself is a code. You have the subject, “I”/ “myself”, the predicate “had” describing the relationship between “I” and an object, and “week” describing a thing of time that was “had” by the “I”. The week also has a modifier, “bad”, to indicate the type that the “week” that was “had” by “I”. That knowledge of your week only exists to you inside your head, until it is made tangible through expression (writing it down or saying it out loud). Until that point, it existed to you alone and could not be shared in its raw form with another person.
You might also choose to send some emojis to a friend to communicate frustration - for example, an angry face:
In my case, I do art.
A few weeks back, I needed to capture the feels of frustration, anger, and helplessness I felt relating to an event that happened in my workplace and about we continuously get derailed and distracted from the mission of good government for the pointlessness of infighting and interpersonal conflict. I chose to use markers. Instead of focusing on creating a recognizable object, I instead used jagged lines scribbled down to show how “strong” the feelings were, they were deeply entangled because these feelings felt jumbled, they were in a bold red colour because it felt like it would stand out. As went through the process, the red felt too strong of a colour, so I went to more gentle colours like brown or blue. The feelings were still expansive and looping, but there was more order and precision.
Then as I took a step back to look at what was drawn, I started to see the little spots of white that appeared between the entanglement of lines. I started to colour in the spots, one by one. It took a long time and a lot of careful application to patch the gaps left by the rush to scribble. As I filled them in one by one, I started to explore larger patches, working outward to larger patches still. Once I was at the edge, there was room to move around. Once I was free to look around I could look to see the big picture and realize I could have been colouring in the open space to start, but I chose to start in the tangliest of places. But it took so much longer to be amongst the tangles than fill in the open blanks.
To button this up: coding is a form of expression. One that is intended to communicate the complexity of the world to a machine that cannot grasp the human experience - when we are still grappling with how to communicate the human experience to ourselves.
The distillation of knowledge down to the barest truths so as to be able to express them in one way or another, so as to be seen and understood; it’s an instinct that is so deeply human.
National Day for Truth and Reconciliation
On Monday September 30, I will be out of office in observance of the National Day for Truth and Reconciliation. Although it is a statutory holiday, it is not just a day off, but rather a time to reflect on the tragic history and ongoing impacts of residential schools in Canada and to honour Indigenous survivors, their families, and communities.
I will share a reflection that I have been iterating on every time I am in the facilitator role and need to do a land acknowledgement as a settler and trying to figure out how to make it meaningful rather than a checkbox exercise that it is sometimes viewed as. No idea if it’s good or correct or rings hollow, it at least feels true to me:
I live and work as a data practitioner in government, on the unceded Territory of the Anishinaabe Algonquin People. I understand that this type of role has benefitted from the histories of recordkeeping and data collection in both government and academia that enabled destructive and extractive colonial administrations and resulted in the theft of sacred objects, knowledge, and desecrated remains from indigenous peoples on Turtle Island and around the world. I believe that I have a duty and obligation in my work to critically assess and challenge our approach to data, the types of data we collect, archive, and share, and how those things have and continue to perpetuate systems of inequity.
Three books that have shaped my thinking around this:
Heart Berries by Terese Mailhot | Penguin Random House Canada - completely emotionally eviscerating processing of how childhood and intergenerational trauma shapes and colours your adult life in gorgeous prose
Maladies of Empire — Harvard University Press - a history of slavery, medicine, and empire that really hits home how huge of a role that quantification (dehumanizing humans into numbers) and recordkeeping played in colonalism
Love after the End | Arsenal Pulp Press - some queer Indigenous futurism because we spend too much time talking about Indigenous Peoples in the past tense. The authors in this anthology “show how queer Indigenous communities can bloom and thrive through utopian narratives that detail the vivacity and strength of 2SQness throughout its plight in the maw of settler colonialism's histories.”
Upcoming events
Speaking:
I’ll be speaking at Dama Days Canada in Ottawa in person on the afternoon of October 31.
You’ll also see me briefly (5 minutes) at IM Backstage Pass 2024 virtually on November 12, 2024 from 1:00 pm to 3:30 pm (ET). I’ll be doing an intro to the GC Data Community Team and demo some of our products.
Attending - come find me to hear about the projects I’ve got cooking and to get a sticker ;)
AccelerateGov Conference on October 21st, 2024 at the Shaw Centre in Ottawa
Shared Services Canada's Innovation Fair: October 23, 2024 - Canada.ca on October 23, 2024 from 8:30 am to 4 pm (ET)
Colleagues are organizing:
Data for Impact Series: The Importance of Data Stewardship virtual on October 02, 2024 from 1:30 pm to 3:00 pm (ET).
Policy and Service Conference: Delivering Impact Through Collaboration (csps-efpc.gc.ca) Hybrid (in person in Ottawa) on November 13 and 14, 2024
Links that I’ll try to remember to paste in every week if I remember
Connect with over 4,000 data people to share resources and job opportunities in the GC Data - Informal/Unofficial Facebook Group
If these weekly posts are not enough, here are almost 800 links to fill your heart. This is a data resource repository I maintain to keep track of all the things I looked at or relevant to doing data in government. The curation is skewed toward things I am interested in or am actively working on.