The Truth Can’t Save Us: Ethical IA through Epistemic Nihilism

Kat King
Hello, I’m Kat King, and this is my talk the truth can’t save us ethical IA through epistemic nihilism. This title is, was a lot when I wrote it back in October, and it is really a lot now. So I just went ahead and changed it a little, the new title, the truth can’t save us ethical AI and epistemic friction. My point is mostly the same. But I’ve gained some new vocabulary that better describes what I mean. And I hope this framing feels a bit more hopeful, because the message was always supposed to be hopeful and grounded in the belief that it’s good to be a person that things don’t have to suck this bad. This was always going to be a hard talk to write in a way that’s understandable and feels relevant not just as an intellectual exercise, but the decisions we make in our real lives that impact real people. It’s a bit a lot harder to write in the middle of what the apocryphal curse calls interesting times, as our information environments are public. Did with lies and 10s of thousands of people are dying from a pandemic that could have been stopped if the appropriate information had been shared, believed and acted upon. Truth is a tough word. Because how we measure it depends on what kind of true we’re talking about. Sometimes it is the correct framing. But I’m not convinced it always is. This talk is not about journalistic facts and the reporting of events or the appropriate use of scientific evidence to make claims about the world external to humans. This talk is about information architecture, and how we within this community talk about what we do. I’m relatively new to information architecture. And the whole time I’ve been a part of this community has been wrestling with the big question of what exactly is information architecture? How do we explain what we do? What is it that we do? What makes us different than other disciplines? What are the things we believe how do you know if you Information Architecture is good or bad, what is the value we bring? And increasingly, it seems the idea of truth is emerging, that information architecture can find the truth and architected into the things we make. And this will make the world a better place. I mean, deeply anxious by this rhetoric. And this talk is my explanation of why, and hopefully a suggestion for another way to think about our work. I thought it might be useful to start with a definition of information. This one comes from Luciano floridi was a contemporary information philosopher. It’s from his 2010 book information, a very short introduction. Sigma is an instance of information understood as semantic content if an only sigma consists of n data, and it’s greater than or equal to one, and the data are well formed and the well forms data are meaningful. Well, that’s Not 100% the most accessible definition. So let’s start by clarifying. First understood a semantic content is clarifying what type of information we’re talking about. semantic means related to meaning, and clarifying. And it’s clarifying that we aren’t talking about environmental information like the rings in a tree that are created in correlation with a natural process. And we aren’t talking about the mathematical view of inflammation that’s concerned with how much can be transmitted under what conditions.

Kat King
Those things are interesting and irrelevant to our work, but they’re not the focus of this talk. We are talking about content with meaning.

Kat King
Our definition next says the information consists of at least one datum. So the next thing we need to clarify is what data means in this context, data are differences between one thing and another, a lack of uniformity. So something might be a color sign. or light might be on, there might be marks on a page. The difference between the shapes of two letters. well formed means that the data, those differences aren’t just random. They’re arranged according to some rules for arranging that type of data. For instance, if the data is language it needs to be arranged according to the appropriate grammar rules. But the system need not be language could be drawings or diagrams, or hand gestures or whatever.

Kat King
Lastly, meaningful indicates that it complies with the meanings available in the system. We’ll just use a famous example to explain the sentence colorless Green Ideas sleep furiously is syntactically correct in English, the sounds that I made correctly formed words which were correctly arranged into a sentence, but it doesn’t mean anything. This sentence on its own has not informed you. It is not information. So your screen should be totally white right now.

Kat King
And now it shouldn’t be. Some of the spots are different now, that data, but I tried my best to be random about it. They aren’t arranged according to any rules. And I don’t think that meaningful. Now I’ve made some of the pixels different. But those different pixels aren’t just random. They’re arranged according to the rules for shapes.

Kat King
There’s a stability to that arrangement of the different pixels, a uniformity. And so this is a thing instead of just a collection of pixels, and that thing is a square, because that’s what we call it when this particular pattern of differences persists for some time. It’s not very meaningful to have these pixels arranged like this, but it does have enough information for you to answer this question. It carries the information square if you know how to read it. But right now, that only tells you about itself. And you probably don’t have any use for this information. There’s two shapes Now, each one is stable and his thing and its arrangement is different than the surrounding pixels. There’s a lack of uniformity between the two shapes.

Kat King
If we are comparing these two shapes, we might think about the data outside the yellow dotted line in the square as the difference that makes a difference. We are concerned if we are concerned with what shapes these shapes are. This is actually where the information is that makes this a square and not a circle.

Kat King
I could ask you which of these are squares, and you’d use the square making information in those corners to tell and ignore the green making information that also makes some of these shapes green, that the shapes are green does not make a difference to them being the square knot. You’ve certainly picked up that information as well though So for our purposes, we can just simplify to this less technical and less rigorous definition. Something is meaningful content information. If if it’s made of at least one difference, arranged in compliance with some rules in a way that means something. Now, it may not mean something to everyone who encounters it. It could even sit on encountered informing no one, but the potential for it to inform is there. And when someone comes along, who can understand it, they can receive the information and add it to their knowledge. Now, this definition makes a lot of sense for information architecture, as some of what we do is arranging things based on the lack of uniformity between them in ways that are meaningful. If you’re familiar with Richard Solomon’s latch, we can think of that as a set of rules for doing this. There’s another interesting layer to how we arrange information. So I have some shapes here. So things different arrange them into two groups based on the differences between them. The bottom shapes are solid, and the top ones are outlines. But that might not be the most useful way to arrange these things. Because they don’t just represent themselves. They are symbols. So let’s arrange them based on the differences between the things they signify, and not the specific representations. Can you guess which difference I used?

Kat King
The foods on the top are foods you can eat raw, and on the bottom are foods that are cooked, but maybe also thought the top foods were healthy and the bottom ones were not. Now, none of these groups of pixels are cooked or raw or healthy or not. Because those words don’t mean anything in reference to pixels. But the pixels are arranged into symbols which hold information about different kinds of things. For instance, the symbol that represents a pineapple represents pineapples You know what sort of persisting uniformity make a thing be a pineapple. And you can manipulate the idea of pineapple in your head in order to arrange these symbols in ways that represent the uniformity and lack of uniformity between the things they represent. This is an awesome ability. These particular symbols are much less standardized than words. So maybe you weren’t sure if that’s a cupcake or a muffin in the bottom row, because there’s nothing representing difference. Those could be sprinkles or sugar bits. But I assume you were sure that it wasn’t an instance of an eggplant.

Kat King
Let’s dig a little deeper into our definition of information though. Because while we can arrange groups of things according to the uniformity or lack of uniformity between them, and just leave them there in piles, when we’re making semantic content information, we usually do more than that. For it breaks down semantic information into two types, factual information and instructional information. factual information informs about a situation or the state of something or states of fact, things fall down when dropped, the doors closed, the berries are poisoned. Paris is the capital of France. Due to the recent cares act, your loan has been placed in deferment, or that number six in the chrome tab and in the header bar of LinkedIn. If I add labels to my piles of sorted symbols, I’m now making a claim about the types of things the symbols represent. Again, you would understand that this isn’t about the actual pixels being eaten raw or not, but the ideas represented by the pixels. One thing that’s important is that factual information Must be political or it’s not information. Feritty makes this distinction and he uses the word veritical and they think that it’s important.

Kat King
Vertical means it’s in compliance with the truth, that it communicates the truth or is truthful. Not that it is the truth. That’s a subtle but important distinction. The point is, though it can’t be false. 3d says a false information when speaks of false information. And the same way one qualifies a false friend, that is not a friend at all. And it doesn’t matter how long after you receive the information that you discover it was both, it was never information at all. This makes sense. Because the set of data a sentence, for example, didn’t actually inform you about the state of something if it was wrong. So once you know it’s false, maybe you can count it as new information about the reliability of its source. Instructional information is meant to bring about some state of affairs. Make sure you put the milk back fridge keep left. If the light is on, keep the door closed. First cream the butter and sugar, then add the eggs. Unlock your full potential, try premium free for one month. Instructional information is neither true nor false. And I found this a little difficult to grasp at first. But these sentences as chunks of information are not falsifiable. They don’t actually make a claim about the truth they just are. The reason it took me some time to wrap my head around is because in context, the context that one might typically use these, there might be a truth claim stated before or after, or it might infer one as I process the information in my head, for example, and that LinkedIn copy, just like we manipulated the ideas of those foods represented by the icons, unlikely to be manipulating the ideas of sentences absorbed Then connecting them with other things. I know this LinkedIn copies classic marketing text, right? doesn’t actually make any claims. But it implies them in a way that they hope makes me understand it as being information about how to unlock my full potential. Whatever it is that means. Information doesn’t have to be as direct as just words. In information architecture, we’re often arranging things like web pages. Here’s an example. LinkedIn has a mechanism for me as a user to open a ticket if I have a problem, and how they’ve structured the pages for me to get there isn’t information architecture decisions, and the arrangement of those pages has created instructional information. So if I click on my own picture in the header of LinkedIn, and then I click on help, it opens a modal window. At the bottom of that I can click on hope and help in any tab, which takes me to a page full of FAQs with some activity. In the sidebar, and a link at the top in the bottom to the user forums, I can go to the help forums and all the way at the bottom. Now there’s a Contact Us link. If I click on that Contact Us link takes me to a page with three choices, the forums, the FA Q’s. And now this third thing. If I click on that, I get to a page with some topics. And when I click on the topics, I get some FA Q’s. But if I click on other, and I type my own question, it tries to suggest me some FAQ answers. But it also provides me with a link to open a ticket. The choice to keep me running through the FAQ in the forum. And to not have that Contact Us link on every page is instructional information that I should help myself. It’s meant to bring out the state of affairs that I figured out on my own, don’t make the company have to pay someone to help me directly. This information is neither true or false. But it wasn’t an accident. It was either Soon driven by business goals to keep the cost of support down. It’s annoying that actually have a problem. But maybe those help resources are actually good enough that most people can solve their issue. The point is, this information architecture decision wasn’t about truth, but it’s a decision that had to be made. Truth as a metric couldn’t have made this less of a labyrinth. Sometimes, the same piece of information can be read both ways. flirty uses the example of a battery light on the dashboard of a car, which was illuminated, we can understand to be factual information that the battery is dead, and instructional information that we ought to charge or change the battery that we may only be interpreting it as one type or another at any moment. Notification icons are like this. They hopefully are factual information about the number of notifications you have. But they’re also meant to bring about the state of affairs where you click and open the app and engage with some sweet content.

Kat King
Let’s go back to the factual information for a moment may have noticed that some of these facts feel different than the others. All of them are encoded Information System humans made up the English language, but some of them seem to be making claims about things that are also made up. These we can think of is social facts that describe a social world.

Kat King
The philosopher Miranda Fricker describes it like this. Suppose we conceive of the social world as a cluster of social facts, I suppose also, there is many social worlds is there are societies the clusters of social facts that constitute them overlapping and permitting of no sharp delineation? What makes a cluster affects a cluster is the same thing that makes the society you one society enough another, roughly speaking relations of social contiguity People living in one society within and around each other, sharing social practices and institutions, most obviously, perhaps political and legal institutions. fricker says that social facts are ontologically dependent on collective interpretive practices. ontologically dependent means that their existence is things dependent on other ideas.

Kat King
In this case that a particular group of humans has a particular way of doing things, and understanding those things they do. A way of living their lives are a form of life, and that they have an interpretive practice of meaning to understand their lives. The truth of a social fact and what it means in relation to other facts depends on the ways we as a group, understand the fact is we engage with each other. The example she gives us this the little discs of gold or silver colored metal that I hate And over to the person standing behind the counter with the packets and tins on the shelves behind him or money. And my handing them over to him in that place in exchange for one of the packets is a part of an act of buying something from a shop. It being the act that it is depends peculiarly on the fact that around here, it counts as such. So the concept of money and buying and shops, all dependent on there being people who buy things from shops with money, and that the action in question meet the local criteria for being called that high as an individual cannot simply say Nope, pennies are not money and have that be the case. Once the social factor set, they exist independent of a single interpreter. They are real, and they have real effects for the people living in the social world. But if we collectively stopped treating pennies as money, they would cease to be money. We’ll come back to this later. Little bit. But for now, what matters is that social facts are facts and truthful claims about them our information and false claims about them or miss or disinformation. We also use social facts frequently in our work. A quick example, if you’re designing the navigation of a clothing website and divide things into the categories, men and women, boys and girls, those are social facts about the genders available in our social world. And the fact that people of each gender wear different types of clothing. So let’s review. Information is data that is arranged in compliance with some rules in a way that means something, and it might be about the physical world or social world can be instructional or it can be factual and true. But what about all the other meaningful well formed data? What is all that and how do we use it and how exactly Do we as individuals get the information from the well formed meaningful data we encounter into our heads? It turns out, maybe there’s a better concept to talk about this whole collection of meaningful arranged data. And that is semantic content. semantic content is any of the well formed meaningful data, whether or not they technically qualify as information. If we look back at our definition information, actually, until we get specific about it being instructional, factual, this can describe all semantic content. Something is meaningful content, if it’s made of at least one difference, arranged in compliance with the rules in a way that means something. So we can update our model or somebody encounters information and then creates knowledge. So the information becomes semantic content, and the knowledge becomes semantic capital. These concepts of semantic content and semantic capital From 3d is more recent work, and he defined semantic capital is any well formed meaningful data that can enhance someone’s power to give meaning to and make sense of something.

Kat King
So we can expand our model again, to show that our Samad capital is what allows us to access the meaning in semantic content. This might seem circular at first, like a chicken and egg problem. But 3d points out that we don’t start at zero figuring all this out. One of the things people do is in better world with semantic content and share it with each other. floridi points out that quote, one is born into a context where meaningful data predate her existence. Were even explicitly taught some of this some of its in books and some of it we hear repeated by others. And some of it we just interact with in our daily lives. And once we’ve got a little bit of semantic capital, we just keep gaining it by living our lives. This is how he Humans are not a thing separate from our living anymore than a fish swimming is a separate activity from their living. Doing this is our evolutionary specialization. Here’s what flurry has to say about semantic capital. semantic capital is made of well formed and meaningful data. And the data are some absences of uniformity. For example, sound in the silence, a light and darkness, a black.on, a white page already count as data. One may say that there are always data because total and absolute uniformity is rare in the universe as complete randomness is. Furthermore, data come together. And such clusters then constitute the game you played when you were a child, the sound of the bell and the village in which you spend your holidays, a song so intimately mixed with a significant experience, which you keep hearing again in so many different circumstances of your life, the inevitable taste of any prosti moment The movies, the movies, you have watched the gossip, you have shared the smell of a new car, the first feeling of a new wedding ring around your finger. Some of what you’re learning is factual information. And some of it is misinformation. And some of it is neither. Its instructional or its art, its beliefs. Some of it is about the physical world, and some is about the social world. Although those things aren’t really separate to a human. As you live, you gain more and more semantic capital, and are able to make more and more meaning and the semantic content you encounter. Each of your users is going to be using their own semantic capital to understand anything MC, because that’s how humans understand. But you’re probably making things for many users, many of whom you’ve never met. So we need a slightly different concept to talk about the resources or the capital, they might have to understand the thing you’re making. Some of your semantic capital is personal to you Based on experiences that you had, but some of it is not. We can call these things that you have in common with other people in your social world shared semantic resources. They might include the things in your social world, they might include the things that people in your social world are taught as children, or things that are simply known in your shared language, your values and beliefs, the models and metaphors you commonly use. If you’re making something that you want a group of people to understand, you’re going to end up relying on shared semantic resources. For example, I assume you watching this video can understand English, and I’m using a relatively standard form of English and attempting, hopefully succeeding at defining and being clear about any words that are less commonly used and therefore not always understood by English speakers generally. also assume you have some concept of the practice of information architecture that overlaps with mine. different individuals within the same system can have different interpretations. But there’s some set of things that allows us to understand together and communicate with each other.

Kat King
Remember, Fricker said, social worlds permit no sharp delineation. So we can think of our social world as being made up of many smaller social worlds. For example, I work in a Library at the University of Michigan. And there are facts about the way things are done in libraries and at the University of Michigan, that are different than if I were an accountant, at General Electric.

Kat King
And I have shared semantic resources, specific language understandings of relationships, ideas about how to measure or value things about what makes a member of one of these communities good. And I may or may not believe each of these things personally, but I understand them as being a part of how I engage in these spaces. And I can use them to make meaning of the interactions I have and to communicate with others, when you’re new to a profession or an organization, you may not yet understand the language or the processes or assumptions, and so can feel how this what is known is different than what is known at the larger social world of your society. But, since each of these spaces also exists within that wider social world, they inherit, and are influenced by many of the same shared resources that we have as a society. And these shared resources often include things which are false, or which are not directly falsifiable because there’s some form of semantic content other than factual information, but maybe they inflate false inferences, or they’re harmful to some people. And the inequality in our form of life means not everyone gets access to the same resources or to add to or to edit these share resources equally, even if they have directly relevant information from their experiences embedded into Our form of life are ideas about a racial hierarchy, about the available genders and the roles people assigned to each should play about what a relationship is, ideas about social class virtue, about bodies and health, mental states and consent about land ownership and property and success and punishment. And these ideas affect who is able to access these resources, and who is able to modify them, who we believe and how we as individuals and societies interpret new semantic content and add it to our personal and shared resources. This brings us to the concepts of epistemic injustice and epistemic oppression. Remember, the epistemic means related to knowing. And so these concepts are describing situations where injustice and oppression are happening related to knowledge and what we know. injustice as a single act and depression is the repeated pattern. Have injustices against a particular marginalized group. One way to think about these issues is as a side effect of general distributive and justices and oppression in our current form of life. That is that marginalized populations may have less access to what fricker calls epistemic goods, like education, or information sources or expert advice. And this can limit one’s ability to engage in the broader epistemic or knowledge making world. But we’re going to focus on the other type of justices and oppression, discriminatory epistemic and justices. And this list is a little harder to understand without explanation, discriminatory and justices have to do with injustice is perpetrated in moments of knowledge sharing, because of some prejudice we carry consciously or unconsciously. pricker notes that prejudice quote works on us below the level of belief without our permission. And so we aren’t just talking about over basis being ready Here, we’re talking about how all of us are being influenced by all of the racist and sexist and so on semantic capital that we have acquired by living our lives in this social world.

Kat King
Pre emptive testimonial and justice is simply not bothering to ask a member of a marginalized group for their perspective, testimonial and justice is failing to extend credibility to a speaker because of their identity in a marginalized group. Hear me medical injustice is not only more difficult to say, it’s a little harder to grasp. This isn’t justice related to our interpretive practices. If you recall, when talking about social facts and social worlds, I said that social facts are ontologically dependent on our shared interpretive practices. This is money because we say it is. But what happens in our form of life, there are created conditions that are regularly experienced by members of the particular marginalized community, but for which we don’t Have a shared way to talk about this is hear me medical injustice. Some examples used by Miranda fricker are women experiencing sexual harassment workplaces in the 1950s and black people experiencing institutionalized racism and their treatment by police. Before the concept of sexual harassment, institutionalized racism were included as concepts in our collective interpretive practices. How could a speaker who had experienced these things, express them to someone who not and be understood? testimonial smothering comes from the philosopher Christy Dobson and as the self censoring of a speaker, because their testimony is risky, and they do not think their listener will engage in good faith or understand correctly. Related to this translational exhaustion is translation exhaustion is an idea that comes from Dr. Twyla Baker, when a speaker from a marginalized group needs to spend significant time setting content texts that the listener lacks because of inequalities and how history is taught before they can explain something. The last type is also from Christie Dotson, and is the idea that simply because of the relative positions and the social worlds, some people have experiences that even when they have the resources describe them. Others simply will not believe because they have never experienced anything similar, or they have only had experiences that seem to be opposite. Labeling these types of injustice is a way of isolating aspects of epistemic inequality in order to talk about them. But in reality, people are likely to experience more than one type of injustice at a time. what creates the conditions of systemic oppression is the repeated in justices against members of a particular population, which means that not only are the individuals not able to add their perspectives or voices, but as a whole the perspective of a marginalized population Not Incorporated. This is not just a result of the sort of unintentional bias that fricker mentioned. But also because many people have an incentive for the current resources to be maintained as they are. Those whose comfort and positions of power are reliant on the continued inequality in our shared form of life may actively suppress these perspectives in order to maintain a structure that best suits them. What’s more, the structure of the system itself makes it easy for the current dominant ideas to prevail. Let me explain by starting with some examples from science.

Kat King
Science is often thought of as the most neutral and objective practice of knowledge making, because it has rigorous practices for inquiry, and mostly confined its focus to the physical world. But looking back at the history of science can be a good way to see how the scientists semantic capital, and therefore ability to make sense of the information they were accessing and creating was influenced by the inequalities in the shared semantic resources of their societies. The further away We are from their understandings, the easier it is for us to see their biases controlling how they understood. In the MIS measurement of man, Stephen Gould traces how scientists attempting to provide empirical evidence for a hierarchy of races, for example, by measuring the size of skulls often got the results they wanted. Not because there was objective truth to their claims, but because their bias seeped into decisions like how to choose samples, how to define the categories, how to aggregate results, and how to interpret what the results meant. Even scientists who were not trying to be misleading and who did not notice their own mistakes were influenced to get the results they wanted, but already having the answer they thought was correct. And then they’re supposedly objective proof of those differences. Could be To support the positions they already held.

Kat King
When Carl Linnaeus, after observing the differences in uniformity between many kinds of animals proposed an update to the shared resources for understanding the relationships between kinds of animals. He divided the then existing quadrupeds category into multiple categories, one of which included humans. He named this new group of animals with humans in it mammals. But why did he name the group after breasts when almost half the members don’t have breasts? Some members, male horses and male mice don’t even have nipples. And it was known at the time that members of the mammal group had in common hair and arrangement of three inner ear bones. monda shy vinner are using her essay why mammals are called mammals, gender, politics and 18th century natural history that is choice aligned with his cultural ideas and values about gender. Time and with Linnaeus his own opinions on things like the naturalness of breastfeeding one’s own children, rather than hiring a wet nurse. naming the category after the breast situated women is less evolved or more animal like breastfeeding as natural and good, and maybe inclusion of animals and the inclusion of humans excuse me and an animal taxonomy easier to accept for many of the people he was trying to convince. This was an act of information architecture, and it permanent permanently encoded the gender binary into the organization and centered breasts is the defining characteristic of mammals. This last example comes from a sleep article by john Vianney. Our understanding of Darwinian evolution is often centered around the idea of the survival of the fittest, and of competition between individuals and species is being central to this process. But increasingly, scientists are finding it useful to think of the same processes in terms of cooperation, that many species the individuals aren’t constantly fiercely competing with each other, but work together. And then often species evolved together in cooperative ways. So how did knowledge about evolution get to be structured in terms of natural competition being the path to evolution, another set of ideas popular at the time Darwin was working with those around competition and the free market. And it’s probable that these ideas shaped how Darwin understood. And once his conception of evolution was incorporated into our collective understanding is true. It became evidence to support the very ideas that shaped it. I’m sure you’ve endured at least one discussion on the free market, framed as being natural because of evolution. So when scientists trying to understand evolution leaned on the ideas of their shared semantic capital about race and gender and economic To make sense of what they were researching and their scientific work, they were then intentionally or not weaving those ideas into their new ideas in ways that persist, such that when we interact with them, it can be hard to separate their framing from the actual factual information, their documented observations. Just like it didn’t matter that some of the squares were green. When I asked you to identify the squares on the slide, you absorb the information about their greenness. Anyways, when we interact with ideas, we sometimes pick up the other semantic content woven in with them. And it can serve as evidence to reinforce the inequality in the structures in our form of life. Even the most mindful of us have absorbed all sorts of ideas that are now a part of our semantic capital and which we may be using without even knowing it. And which we may be including in our contributions to shared understanding, without realizing it. Remember, when I arranged these foods in a factual way based on whether they represented something you could eat raw or not. I thought it was possible. You could also read the arrangement as being healthy and unhealthy. How would that have happened? What makes a food healthy? These foods are healthy as a factual statement arranged like factual information. But what it means is harder to isolate, particularly in a culture with some complex social ideas about eating and health.

Kat King
So here we come to the point of my talk. I think it’s ethically risky to use truth as our professional aspiration. And as the framing for why what we do is valuable. It’s not always really clear what true means in a particular moment. Some portion of the information we know will turn out later to have been missed or disinformation all along, and many of the ideas and semantic capital we have are not directly falsifiable anyways, some are only true by agreement and what is agreed upon me be harmful to some people. And by upholding that truth, we are upholding the harm. A single arrangement of information may have layers of shared ideas woven into it. And even if we try to be mindful of it, our architectures may have the biases we’ve absorbed from our form of life, woven beneath the surface. Sometimes, the problem is not a particular fact being true or not. But in the context of how the boundaries were defined, how it’s presented, and whether it’s the appropriate truth for the situation. Yes, we should strive to create architectures that are vertical, or that align with the truth, particularly the narrow local truths that we can control the boundaries and meanings of like you have a notification or your privacy settings have been updated. But the assumption that we as individuals, or as a practice deeply embedded in capitalism, making things that serve the purposes of our various employers are going to overcome the inertia of The bullshit we’ve inherited strikes me is unlikely. And they think we aren’t careful, we could just align with the dominant structures and receive plenty of feedback that our work is good. And the information true. I think of information architecture work is being about ontological alignment, by which I mean we are structuring information environments that it aligns with some overlap in the sets of shared semantic resources and goals. Taking into account the reality of the processes involved, the business goals, the users goals, and with the understandings of the shared semantic resources of various groups of nodes that make up our users. How can we make this information environment help people understand and be effective? And we don’t need to make claims that our architectures are the truth to do this. This is not to say that explorations of truth or not worthwhile, I think they are I think we have to be careful about the rhetoric we use today. Talk about such work and make sure we aren’t simply a grand dicing ourselves as a profession by claiming the power to find the truth. I think it is ethically much safer to assume we can’t know the truth, but that we can do our best to be honest and clear and to create conditions for epistemic equality. We’re knowers from groups marginalized in our wider social world, and those marginalized within the smaller social worlds of our organizations and domains are taken seriously and their perspectives Incorporated. And there’s high stakes here, especially as we seek to grow our field into the areas which I agree we have something important to contribute artificial intelligence and organizational design. Because we are the people creating the information structures that support and reify our current form of life. The structures we make will be interacted with, in some cases by millions of people. And each time one of them interacts with it, the experience will add to their semantic capital. help them understand in the future, in some cases or architectures will be the source of authority for some sort of local facts. Whatever mistakes we make or harmful framing we build into our systems we have the potential to perpetuate forward, especially if we frame them as valuable because the true we should be transparent about the goals and intentions we have in structuring a particular architecture and about who is impacted and about whose perspectives we took into account. I think we already do this, and we even talk about it like this. The case study by Sam Raddatz and Claire Rock presented at the IAC last year about restructuring the navigation of the National Center for Transgender Equality website is a great example of framing work like this. I think one of the valuable things about information architecture is that our methods can help facilitate conversations about what things are dependent, what things can change what’s semantic resources we’re using or expecting our users to use in order to make sense of our information. And to suss out which of these we really need to make something clear. These conversations are the first steps to epistemic equality happen.

Kat King
Here’s an example. There was a time not that long ago, when marriage is a contract between a man and a woman was a true statement. This was a social fact, based on the law in the United States, but one which left many people disadvantaged, and which was based on semantic capital that many people had in the form of beliefs that this was correct or natural or moral or whatever. If you were creating an information architecture 1015 2530 years ago, would you have built into your assumption that this was true? Would you have designed a marriage application with the fields name of bride and name of groom? It’d be understanding It would make the form easy to understand for those who are allowed to fill it out. But even then, when it was true that a state sanctioned marriage could legally only be between a man and a woman, it was possible to build an architecture that could expand to support other definitions that as people used, it didn’t provide reinforcing evidence that a bride and groom were the constituting facts of the marriage, and it would have been ready to use without changes when the law changed. Truth as a metric is better served by labeling the field bride and groom, but equality would have been better served by leaving the fields on engendered. But to do so would have required incorporating perspectives beyond the current truth outside of those intended to use the form and to have some imagination about future possibilities. I want to close with some thoughts on the tension between epistemological resilience and epistemic friction. epistemological resilience is the idea that a particular system of knowing is going to be robust to change. In part because of the structural way old ideas are passed forward. Because our ways of knowing are pronged, only revealing things knowable by our ways of knowing. And in part because they’re prone to absorbing disturbances without actually changing. The system for knowing must have some resilience, because if it doesn’t, it loses its usefulness for structuring the world. And because systems without resilience simply cease to be over time. Strong declarations The truth is available in a system uphold the resilience of a particular system. But if we have a resilient system that we are confident in, it can mean that we miss alternative ways of understanding a situation. In an essay on credibility and authority. Jose Medina introduces the idea of meta blindness as the cause of suffering. systemic and justices. That is that knowers confident in their own way of knowing, and their own ability to see the truth are blind to their own ignorance. He uses as his, as his example, the courtroom scene in To Kill a Mockingbird. We’re an all white jury in the southern United States in the 1930s is unable to correctly understand the testimony. In the trial of Tom, a black man accused of raping a white woman. the realities of the social world they’ve had it limits what can be said and how and the jury ignorant of their own inability to understand cannot get to what actually happened through the testimony. This is on one level, an example apparently medical a justice or injustice based on a lack of shared resources for describing what happens. But Medina presents another way to think about this injustice is not only that the jury does not have the correct resources to understand the testimony. They are also unaware have their own ignorance. There are other white listeners in the audience who can understand and it’s clear to them. And it’s clear to the black members of the audience that there are two layers of meaning to what is being said. The epistemic meta blindness of the jury means that they don’t just not understand they are confident that they have understood and have gotten to the truth. How in our own work can reduce the likelihood of meta blindness preventing us from incorporating perspectives of marginalized groups. Medina offers the idea of epistemic friction. Here’s a quote. As an antidote to this meta blindness, we need to seek epistemic friction, that is to actively search for more alternatives than those noticed, to acknowledge them or their possibility, and to attempt to engage with them whenever possible. It is crucial to have more than one form of receptivity culturally available, but it is also important to have the ability to move Back and forth among alternative sensibilities, to look at the world from more than one perspective, to hold different viewpoints simultaneously so that they can be compared and contrasted, corrected by each other and combined when possible. It is important to entertain different perspectives without polarizing them, or they’re customizing them and presenting them as exhaustive. I think epistemic friction is something information architecture as a practice could support. Our methods and approaches already seek out the differences and similarities between things like the mental models and language of various groups of stakeholders. And embracing this sort of friction could help us not simply tell the truth, but to do what we can to adjust our shared systems understanding towards being more equitable

Transcribed by https://otter.ai