Towards a digital progressive stack
I first encountered “Progressive stack” during Occupy Wall Street, where nearly all meetings ran by some version of formal consensus process. Progressive stack modifies the deliberative parts of a process by empowering the ‘stack taker’ - the member of the facilitation team tasked with ordering speakers - to shuffle the order such that members of traditionally disempowered groups recieve priority. It is intended as an instrument against the internalized forms of oppression, superiority, and bias that we all carry around, and its adoption by Occupy is a good example of how central the concept of intersectionality was to that movement's form and intended function.
I am a fan of progressive stack, because I am a superfan of fucking up the normative power dynamics that left unchecked (or simply unacknowledged) will inevitably limit, fracture and deplete movements for justice. I’m also someone who happens to be trying to build instruments of collective decision-making online, and much of my inspiration is drawn from the real-world toolkits we are continually refining in our movements. I think we’ll need a digital version of progressive stack if we intend for the tools we use online to make our movements the liberated and safe spaces we want them to be.
a problem with progressive stack
Perhaps the first obvious challenge seems to also be, to a lesser extent, a problem with the real-world version of progressive stack: its implementation often involves making a lot of assumptions. People don't always exhibit meaningful indicators of their various forms of privilege or oppression, and taking a survey is usually out of the question for various reasons of privacy, security and courtesy. Furthermore, by invoking categories of identity one might likely ask people to place themselves in, we would risk the likelihood of reinforcing the very systems we seek to undermine by asking people to situate themselves within their logic.
In the real world, this information gap is one person’s to solve — the stack-taker’s. To do their job, they make various sorting decisions based on a combination of presumption and familiarity, depending on the person. But, even in real life, a differentiation between someone with white-skin privilege and light-skin privilege isn’t always easy to guess at, and external markers of class are routinely fudged, to some extent in all directions — not to mention gender, where the most progressive thing to do is usually not to make assumptions at all. But the stack produced after applying these theoretical weightings to speakers is usually accepted without much scrutiny. It’s dealer’s choice — because, again, what else could it be?
This inexact science is pretty impossible to apply on the Internet, because so many of the markers of privilege and/or oppression in society are made invisible online. In some ways, this is itself a wrench in the usual gears, but it’s not a solution: in general, we bring the same internalized experiences, entitlements, and internalized oppression to the Internet when we sit at a keyboard that we’ll bring to a real-world meeting. But the way we do it IRL has its flaws anyway. Perhaps we can achieve the same goal -- of factoring in and countering for internalized oppression and superiority — with a different approach.
self-identification
The only real source of truth we have about the ways that people have been affected by oppression is people themselves. Online it might be marginally less awkward to have a webapp ask its users to identify themselves outright, but even categorizing people is still potentially oppressive, and once that information exists in a hard form it may constitute an acute risk to privacy.
So if we can’t ask people to tell us the ways in which they are privileged and oppressed, how can we ‘sort’ their voices in an appropriately progressive manner? I propose to experiment with an alternative approach: ask folks to identify how they believe they may be situated in comparison to each other, and then calculate a sort that most accurately reflects the group’s self-identification. It would be a sort of self-stacking.
Here’s an example. Let’s imagine the following four users:
Justine is a middle-aged queer black ciswoman who grew up lower middle class in Bed Stuy, inherited her familial house and now makes her living from renting the property to both long-time neighborhood residents and gentrifying white folks. Judy is a 20-something white trans woman who grew up in a wealthy household, graduated high school, dropped out of college and is now a library assistant. Joan is a young white ciswoman who grew up poor, was given a scholarship to a university and eventually found her way into the depths of academia. She receives a healthy salary. Jorge is a young brown-skinned cisman of color who grew up middle class, graduated high school, and receives straight and male privilege. He works odd jobs and struggles to pay rent each month. Joe is a middle-aged cis white guy who grew up middle class in Massachusetts and is currently a systems engineer at a tech startup in Manhattan. He is pretty much your vanilla straight white dude, looks the part, and gets the usual benefits. Let’s further assume, for the sake of controlling just a few of the many variables involved in the complex web of identity and power we seek to model, that every one of these people basically knows and claims their own identity and relationship to these linked oppressions. They have limited information about each other, though — basically, what is claimed on their online profiles or evident through their communication. Therefore:
Everyone knows that Justine identifies as black and queer. Everyone knows that Judy is white, identifies as a woman and that she is a librarian. Everyone knows only that Jorge identifies as brown. Everyone knows that Joan is white and may gather at a minimum that she is an intellectual person. Everyone knows that Joe is white, male and a techie. What a mess! Even if a stacktaker knew everything about these participants, manually ordering folks boils down to trying to approximate a hierarchy of oppressions. It’s not a stretch to put Joe at the bottom of the priority order: in fact, one might argue that the main function of progressive stack is to make sure that voices like Joe’s are deprioritized; everyone else has some kind of internalized oppression worth undermining. But what if Joe is poor? Now it’s up to the stacktaker’s sense of what matters more: race, gender, or class.
But what if we asked each person to situate themselves relative to everyone else, based on their best guess? Let’s say there’s a slider from 1-10. 5 = “I think we have about the same amount of privilege and power or I’m not sure”. 1 = “I’m sure I have less privilege and power in society than this person” and 10 = “I’m sure I have more privilege and power in society than this person."
Joe’s self-identified relationships might look like: Justine: 10 (certain he has more power and privilege) Judy: 10 (certain he has more power and privilege) Joan: 8 (thinks he has somewhat more power and privilege) Jorge: 10 (certain he has more power and privilege) Justine’s self-identified relationships might look like: Joe: 1 (certain he has more power and privilege) Judy: 3 (believes Judy has somewhat more power and privilege) Joan: 3 (believes Joan has somewhat more power and privilege) Jorge: 5 (believes they have about the same amount of power and privilege) Judy: Justine: 8 Joe: 2 Joan: 3 Jorge: 7 Joan: Justine: 9 Judy: 7 Joe: 2 Jorge: 9 Jorge: Joe: 3 Justine: 5 Judy: 5 Joan: 3
So, who agrees? (Map out agreements and then make best guesses about the rest).
This has a couple of benefits.