Two types of power
Why you can't separate truth from power, how leveraging values systems empowers you to own your truth, and why everyone wants to work like an analyst (they just don't know it yet).
👋 My favorite Roundup posts are the ones that are part of an ongoing conversation — it’s mind-bending to me that these conversations happen asynchronously across the blue bird site, substacks, medium posts, replies to this newsletter and for a week at a time or even more.
Sometimes though, a little synchronous is nice. Tristan and I are trying something different starting this week: we’re going to be hanging out in Slack (yes, even more than usual) in a channel for this roundup and the podcast.
The idea is simple:
bring a topic or jump in while we jam on the next issue in public
tear apart a current (or prior) issue/podcast
Looking forward to the spirited discourse! 🤓
PS. if you need to sign up, you can do it here.
Perennial Truth Architectures
by Stephen Bailey
Stephen went wayyy deep this week. I spent the last four days trying to decide if I’m going to join him down this rabbit hole, or commiserate with Elena on cross database syntax. And then Tristan, in Slack:
Ok I’ll bite. Geronimo! 🐇 🕳️
This week, Stephen took us on a multi-millennial journey on systems of knowledge from Roman legions, to the evolution of faith, the Republican party and the Scientific Method. His point: data professionals operate within a specific cultural context in their organizations, and that context can vary as widely as the timelines in Stephen’s post. So how can we begin talking about defining single sources of truth in data, when what is true and knowable can vary so wildly even within the same organization?
Enter clever chart:
Stephen describes two paths to growth in an organization that represent two different approaches to truth: the path of power, in which the word of
God CEO comes down and slowly diverges via apostles organizational hierarchy; and the path of consensus, in which multiple humans converge on a truth based on shared principles (like, you know, scientific method and stuff).
Y’all are not ready for how excited I am about this. I have been waiting a good 15 years to quote Foucault again on the subject of truth:
The important thing here, I believe, is that truth isn't outside power, or lacking in power: contrary to a myth whose history and functions would repay further study, truth isn't the reward of free spirits, the child of protracted solitude, nor the privilege of those who have succeeded in liberating themselves. Truth is a thing of this world: it is produced only by virtue of multiple forms of constraint. And it induces regular effects of power. Each society has its régime of truth, its 'general polities' of truth: that is, the types of discourse which it accepts and makes function as true; the mechanisms and instances which enable one to distinguish true and false statements, the means by which each is sanctioned; the techniques and procedures accorded value in the acquisition of truth; the status of those who are charged with saying what counts as true.
To paraphrase an old French dude — truth is always relative to the system of power, and they who control the means to define what is “true” get to sit at the cool kids table and do the jingle bell rock:
If you remember high school, you remember the two kinds of power that exist within those walls:
the explicit, hierarchical, formal and structured kind with uniforms, written rules, incentives for good behavior and the feeling that someone is always waiting for you to mess up and get detention; and then there’s
the implicit, tacit and intangible kind. The kind you can’t put your finger on, but everyone agrees on anyway. The kind you know as soon as you step into the hallway:
This second kind of power is driven by values systems and principles (some other old French dudes call this the habitus). In Mean Girls, those value systems are things like traditional gender roles, magazine cover concepts of beauty and consumerism. In your organization, these values and principles may be explicit or implied, and in both cases observable by the types of behaviors that are incentivized and rewarded in the organization. Put simply, what gets folks promoted and recognized?
Ok, Anna, I like Mean Girls too but what do they and old French dudes have to do with data systems?
If you’re a data professional of some flavor, odds are that you, your manager or someone in your reporting chain is constantly thinking about how to make data happen. That is, how to encourage more of the organization to leverage data in their decision making process. Any data. Even smol data. 🤏
If you’re lucky, you’ve already got the voice of
God CEO spreading the good gospel of data driven decision making to all their apostles. But something still isn’t quite right. Everyone has their own interpretation of what that means, and how to do it. NONE OF THE DASHBOARD KPIs MATCH! 😱
Turns out that a top down data driven approach by itself is not enough, for the same reason that adopting agile practices alone is not enough to transform engineering practice. It turns out you also need psychological safety, some amount of self-determination and some way to get everyone to march in the same direction. The end result of this looks like what Stephen calls the path of consensus:
”The reach of the company increases, but instead of breaking apart into factions, groups maintain lines of communication. As the company grows, it retains a posture of learning, not conquest, and assimilates new talent efficiently”
The way you get everyone to march in the same direction on a path of consensus is by defining a clear set of values and principles by which to enact them, and constantly working with the humans in the organization align on, reinforce and uphold those values.
Coming back to the agile example: it’s not enough to do two week sprints, daily stand ups and hire a scrum master. You must also align all your humans on the values system defined in the agile manifesto, a list written with a particular principle in mind — the experience of the customer comes first. If your organization doesn’t live and breathe this principle, agile is probably not the right methodology for you.
Analytics engineering is not unlike agile in that the practice and the community that has formed around it are both founded on a series of value statements (quality, human dignity, community) and principles (analytics is collaborative, analytics code is an asset, and analytics code needs automated tools).
So if you’re looking to make data happen in your organization, and speaking truth to power isn’t quite working out for you, try focusing on values alignment instead.
What does that look like?
If you’ve read Robert Yi’s post this week and are also thinking that analytics teams still need better alignment with business objectives, try aligning with what the business values instead.
Maybe instead of “Make sure to use some data to make your decisions”, we say instead, “Gain the ability to articulate, define and arbitrate your own goals”. Who doesn’t like some good old fashioned self determination?
And maybe this is why “quickly pulling some data” is an unexpectedly viral phenomenon across organizations that goes well beyond the analytics team:
“For decades, the industry has thought of those people the way we did at Mode in the early days: different, divided, and discrete. This belief is embedded across the entire data ecosystem. Products have technical and non-technical faces; vendors distinguish between analyst and non-analyst seats; brands are built around which side of the spectrum you favor; Gartner tells you you need to buy two reports and not one. […]
I’m now realizing that misses the real point: The divisions of labor between analysts and everyone else are fading. Analysis is getting bundled with other functions; the behaviors of analysts and non-analysts are overlapping; analysts are becoming positionless. The reason to build a “modern data experience” isn’t to unify the disjointed products of a bunch of startups; it’s to serve a world in which far more people want to work like analysts.“
-Benn Stancil, Work Like an Analyst
Far more people want to work like analysts than we expect because it’s ****-ing empowering! Because being able to define your own goals, measure your own success means defining and owning your truth. And as several old French dudes and Regina George already know — truth is power 💅
Don’t miss these too!
In What is SaaS debt?, Sarah Krasnik identifies a kind of debt more insidious than tech debt — process debt. The 3 principles she identifies to combat process debt are right out of the software engineering playbook: templates, testing and versioning. Her value system — technical literacy in scaling systems is more important than actually knowing how to code:
Just like philosophy is math without numbers, take out the code from system design and you’ve got the principles of building scalable operational processes
What started as a 💥 Twitter thread is now a 💥 Blog Post going into all the details of how Nate Sooter founded an analytics engineering team. I particularly enjoyed the section on making the case, because it so strongly resonates with the theme of self-determination I touched on above:
Something magical happened over the period of a couple weeks. We started asking even more “What if?” questions. They included:
What if we chose our own set of tools?
What if analysts could ship early and often instead of every few weeks?
Yes!!! Also check out Nate and team’s interesting column-level ownership approach and how they made this work for them!
I’ve already touched on some choice quotes from Work like an analyst by Benn Stancil above, and highly encourage you to read the rest to find out why the data industry needs a truck and for a well placed Back to the Future reference.
In Speed Changed Analytics, but Our Processes Need to Change Too, Robert Yi is making “Decision Scientist” happen. What do you think about adding this title to the growing pantheon of data jobs?
And better late than never:
Rules for making Airflow work in 2022 by @raphaelauv has promising meme game, and genuinely useful advice if you are still thinking about setting up Airflow in 2022 (say it with me: do not run compute directly in Airflow 👏):
And in Write that down: What I've learned about documentation for data teams find out what 3.5 weeks of infrastructure, testing and documentation work enabled Brittany Bennett’s team to accomplish going into 2022 (I’m a sucker for PR checklists that include docs ❤️).
That’s it for this week! 👋