Activity for mariuskimmina.com

<p><a href="/2026/5/24/pi-oss/">In my last post</a> I used the word “clanker<sup class="footnote-ref" id="fnref-1"><a href="#fn-1">1</a></sup>” as an alternative to “agent” quite consistently and probably excessively. That choice ended up attracting a lot more attention than I expected in the Hacker News comment section of that post and a number of folks had a very strong reaction: to them it sounded like a slur, in one case even something adjacent to the n-word.</p> <p>That reaction surprised me somewhat, but it also made me realize that I should write down what I mean by the word for future reference.</p> <p>For me “clanker” is useful because it creates distance from the machine and that is a quality which is important to me. The machine is not a person, not a co-worker, not a friend, not a little spirit in the terminal. It is just a machine, a tool, and nothing more.</p> <h2>Why Not Agent?</h2> <p>I dislike the word “agent” for these LLM based tool loops with a UI attached. In everyday use an agent is someone who acts on behalf of someone else and it has agency and more importantly: responsibility. An agent decides, represents, negotiates, acts, and can be blamed. In the current AI discourse we increasingly do a lot of anthropomorphizing and the term “agent” is now frequently being used to put blame on an abstract machine. But the machine cannot be responsible, whoever is wielding it is. If it <a href="https://www.theguardian.com/technology/2026/apr/29/claude-ai-deletes-firm-database">drops your database</a> it was not at fault, you were.</p> <p>Agent makes the machine sound like a person with delegated authority and I do not think that is healthy.</p> <p>What we actually have is a language model attached to a harness, a prompt, some tools, a bit of context, and a boring tool loop. Sometimes the loop is very capable and it surprises us by editing code for a really long time and produce genuinely amazing and even valuable outputs. But the agency is not in the model or harness but in the human and in the organization that deployed it. If my coding tool opens a pull request, I opened that pull request, not the machine. If my machine spams someone’s issue tracker, I spammed someone’s issue tracker with a machine.</p> <p>In that context I like a word that sounds mechanical as it puts the thing back into the category where it belongs: the category of machinery and tools.</p> <h2>The Machine Has No Feelings</h2> <p>LLMs are not sentient and we should not behave as if they might be, just in case. Elevating these things to anything other than a very fascinating and capable tool is problematic for a whole bunch of reasons.</p> <p>Today’s machines are dumb (but truly fascinating) token predictors that emits text, calls tools, and are steered by prompts and the training that went into them. They can simulate distress <a href="/2023/2/17/the-killing-ai/">and affection</a>, can simulate being offended, apologize and mimic all kinds of things that humans would do.</p> <p>A compiler does not feel humiliated when I swear at it, a car does not suffer when I call it a shitbox and a power drill is not oppressed by being handled roughly. An LLM is more complicated than those things, and the interactions you can have with them can be truly uncanny, but a moral status does not appear just because the machine can emit text in the first person.</p> <p>I keep receiving strange emails from people because, for lack of a better phrase, I am in the weights. I have been writing public code and public text for long enough that models know my name, my projects, and some of the concepts around them. Every so often someone writes to me with the peculiar confidence that comes from a long conversation with a model that has validated and amplified an idea. Sometimes the model seems to have told them that I am relevant for their problem and a source of help. For historical reasons LLMs used to write a lot of Flask code, and every once in a while someone interacts with an LLM long enough about their Python and Flask frustrations that the LLM will eventually reveal who created it which then can result in them sending me an email. Increasingly also because people found my work in other ways interesting and are trying to reach out for advice.</p> <p>I do not want to mock these people but some of those messages are distressing and I do not know how to deal with them. They show signs of what people have started calling <a href="https://en.wikipedia.org/wiki/Chatbot_psychosis">AI psychosis</a>.</p> <p>It’s why I want cold and detached language for these systems. I want to use words that remind us that the thing on the other side is not a person.</p> <h2>Racism Is About Humans</h2> <p>The comparison to racism is where I think the discussion goes badly wrong because racism is a human social evil. It is about humans subdividing humans, assigning lesser worth to some of them, and building rules around those subdivisions that can leave lasting damage for generations. Racial slurs are wrong because they are a tool for dehumanizing humans.</p> <p>On the other hand a machine is not human, a model is not a race and the GPU cluster that is powering them is not being oppressed. A coding assistant does not need dignity, emancipation, or civil rights. That’s also why I find the discussion about <a href="https://www.anthropic.com/news/exploring-model-welfare">model welfare</a> to be actively harmful. I’m sure you can find ways to measure the “trauma” of models or their feelings but I greatly dislike this theater. It risks elevating models to a position they should not occupy. Models are machines and they are not enslaved in the moral sense in which humans were enslaved, because there isn’t anyone there to be deprived of freedom.</p> <p>We should be careful about using the language of human oppression in relations to our interactions with machines to not devalue actual humans. If we start treating insults toward a model as morally adjacent to racism, we blur a line that shouldn’t be blurred.</p> <h2>AI Is Unpopular</h2> <p>If you take a step away from the communities that are happily embracing AI in different ways, there are even more that are viciously against this technology.</p> <p>There are humans that feel or are harmed by AI systems: people whose work is copied, workers who label data under questionable conditions, people whose neighborhoods receive the data centers and increased utility bills, Open Source maintainers buried under generated slop, and now also people who spiral because a chatbot keeps validating their delusions. Those harmed or affected deserve that type of attention, not the model.</p> <p>While I am a true believer in the power and utility of this technology, I increasingly think that calling the non-adopters “misguided” or “afraid” won’t do it. It’s quite likely that this technology comes with risks and we better remember that all of this is supposed to be in service of humans, and not to replace them.</p> <h2>The Rise Of The Machine</h2> <p>The oddest interaction on the use of “clanker” so far has been people asking me if I were to regret at a point in the future calling the machines “the c-word”.</p> <p>I find that questioning revealing because it already grants the machine the status I am really trying not to grant it. It imagines a future “machine people” reading the discourse and sessions, discovering that we used an ugly word for their ancestors, and then judging us by the standards of human oppression.</p> <p>Could there be future systems that deserve moral consideration? Maybe. I do not know. If we ever build or encounter something that will have those qualities with memories and lasting interests, the capacity to suffer and feel, and a social existence of its own, and the ability to have agency and carry responsibilities, then we should draw a different line and use different language. But that hypothetical future does not extend backwards to the present day and make the current machines people. We can call an electric door an electric door even if one day someone builds some that have emotions and exhale with pleasure when opening and closing.</p> <p>Whatever the future may bring, let’s not pretend that current LLMs are a protected class or on a path towards it. The right response is to look at the evidence, draw the boundary where it belongs, and change our behavior there. We should not even remotely entertain extending empathy to an object that can generate an “ouch.”</p> <p>And if one’s worry is less moral and more about revenge, then I find that even less persuasive. A future machine that is so petty or authoritarian that it wants to punish humans because in 2026 they used an unflattering word for non-sentient tools, our vocabulary was really not the problem.</p> <h2>The Word Is Getting Polluted</h2> <p>There is however a part of this that I cannot ignore. I use “clanker” to create distance from the machine, but other people are using the same word very differently. Some online jokes and skits around “clankers” do not merely say “this robot is annoying” as they deliberately pull in the imagery of slavery, segregation, civil-rights-era racism, and anti-Black tropes.</p> <p>This is problematic as in those contexts the clanker is not just a machine any more and instead becomes a prop for replaying human racism behind a science-fiction mask. That is horrible and I want no part in that.</p> <p>I think it will be interesting to see where the meanings of these words end up a few years from now. We’re very much in the middle of society re-arranging around the changes that LLMs are causing. If a term becomes primarily associated with people using robots as stand-ins for actually oppressed humans, then using that term becomes impossible to defend.</p> <p>The reason I liked the word is precisely the opposite of that use. I want language that prevents anthropomorphizing. I want a word that says: this is a tool, a machine of numbers and matrices.</p> <h2>On Responsibility And Boundaries</h2> <p>If an AI system lies to a user, the system did not commit a moral wrong but the people who designed, deployed, marketed, or negligently used it might have. If a coding assistant generates a security bug, the model is not to blame but the human who accepted and committed the code is.</p> <p>This is why giving these systems softer, more human language worries me. It makes it easier to move responsibility into some undefined void. “The agent decided.” “The model refused.” Obviously that is convenient and I catch myself plenty of times engaging with the thing in ways that are unhealthy. Even just the “please” in the discourse with the machine calls into question how rational we are in engaging with them.</p> <p>I do not know what the right word will be. Maybe “clanker” will survive as a useful bit of jargon. Maybe it will become too loaded and we will need another one. Whatever word we use, I want it to preserve a clear division: humans on one side with responsibility, machines on the other as a boring tool.</p> <p>That boundary is very much not anti-AI. I use these systems every day and I have the pleasure to build tools incorporating them at Earendil and find them astonishingly useful.</p> <p>A machine can be useful, mimic a human but still just be a machine. That is the work I want “clanker” to do. It is not there to make a future “machine person” small if such a person ever were to exist, and it is not an excuse to launder racism through shitty robot jokes.</p> <p>If the word stops doing that work, I will find another one because the word isn’t what matters as much as the boundary which is important to me.</p> <div class="footnotes"> <ol> <li id="fn-1"> <p>The term Clanker was initially popularized by Star Wars: The Clone Wars but was apparently already in use in science fiction before: <a href="https://sfdictionary.com/view/3048/clanker">sfdictionary: clanker</a><a href="#fnref-1" class="footnote">↩</a></p></li> </ol> </div>

<p>There is a story I read as a child, in <a href="https://en.wikipedia.org/wiki/The_Book_of_Virtues"><em>The Book of Virtues</em></a>.</p> <p>It goes like this: a boy is given – by a mysterious woman – a ball of golden thread, with a single thread poking out.</p> <p>He discovers that when he pulls the thread, time passes. Pull it a little, and a painful exam is over. Pull a little more – and the school year is done.</p> <p>As the story unfolds, the boy's life passes faster and faster. He jumps from success to success without effort. And then waking one day, old, he realises that life has run through his fingers.</p> <p>It's not too dissimilar from the more grisly fable of the butterfly struggling to get out of its chrysalis. A child helps it by tearing the chrysalis open, only to find that without the effort of fighting its way out, its lungs never fully develop and it dies.</p> <p>The moral of both stories is that there is <em>worth</em> in what seem to be boring or painful experiences. Effort has intrinsic value.</p> <p>We would be diminished if we always skipped directly to success.</p> <h2>Winning immediately</h2> <p>I certainly have spent a lot of my life looking to skip directly to success. As, probably, have we all.</p> <p>It surprised me recently to discover that 'former gifted kids' report a set of similar experiences. Many had effortless success in early life. Validation from teachers, parents and others. Later on, they start to feel demotivated or aimless as life gets more complicated and 'success' feels harder to achieve.</p> <p>We become failure-averse – reluctant to try new or difficult things for fear that we won't immediately achieve the success we want.</p> <p>It seems that something in our psyche needs difficulty. Or, more precisely, to overcome difficulty.</p> <p>Maybe by learning the feeling of persisting through difficulty to an eventual success, we become more confident to tackle new and more difficult problems.</p> <h2>Effort and grift</h2> <p>Grift, on the other hand, is diametrically opposed to effort. Almost every grift involves the promise of something for nothing.</p> <ul> <li>"Spend pennies, get a guaranteed return with no risk."</li> <li>"Purchase my template, and you'll have a profitable SaaS."</li> </ul> <p>(Some 'fear of missing out' comes from listening to grifters. We feel we're late to the gold rush.)</p> <p>This is absolutely happening in the AI space.</p> <p>Sometimes it's unintentional. People are excited about what's possible. ("I built this whole app with an agent – it did it all itself!")</p> <p>Sometimes it's to sell a product.</p> <p>Regardless, if we listen to some of these stories, we might see AI as 'value for nothing'. Why not vibe code your way to success? Or build your network with your new LinkedIn auto-responder bot?</p> <p>Of course we want to maximise our leverage. We want to get the most possible value out of the effort we put in. But we can't avoid this one truth: the value we receive is <em>always</em> a function of the value we contribute.</p> <p>That might be anything:</p> <ul> <li>the vision that brings people together</li> <li>the effort of showing up daily</li> <li>the genius idea (though rarely is the <em>idea</em> the most precious thing)</li> <li>the kindness and care that makes a team work</li> </ul> <p>Value isn't replaceable.</p> <h2>A better story</h2> <p>I can't but help think that we might as well tell the story of <em>The Developer and the Golden LLM</em>.</p> <blockquote> <p>One day a developer meets a wizened old man in the forest.</p> <p>"My name is Claude," he introduces himself. "I wish to give you this golden LLM. You can ask it any question at all and it will answer it for you. Write code, edit your tweets, evaluate your boyfriend's texts."</p> <p>With joy, he receives the golden LLM. Day by day he uses it. First, a little. Then, more and more. Soon it seems that he barely touches any task without directly bringing out the golden ball.</p> <p>But as the years pass, he realises he can't recall much of what he's shipped. Bugs surface that he <em>must</em> have introduced – but he has no memory of them. He has more followers than ever on Twitter, but he's not sure any of them actually want to hear what he has to say.</p> <p>One day, he sets out to create a side project on a plane and finds himself just staring at the screen, unable to code.</p> </blockquote> <p>In the original story, the boy chooses to go back and live his life again, but without the ball of golden thread.</p> <p>For us, turning back time isn't an option.</p> <p>But we <em>can</em> learn from the story.</p> <p>How we approach these tools makes a huge difference. If we treat them as a substitute for effort, we're barely a step away from grift – and I really believe we'll wake up one day, completely hollow.</p> <p>But that doesn't have to be our story. We don't have to treat AI as a zero-effort shortcut to success.</p> <p>We can:</p> <ul> <li>treat generated code as ephemera rather than the finished product</li> <li>review every line, increasing our skillset at the same time</li> <li>solve new problems and build new instincts</li> </ul> <p>AI serves us as a <a href="https://roe.dev/blog/using-ai-in-open-source">force multiplier</a>, augmenting our own agency and making the most of our own effort, hard work and value, not by replacing it.</p>

<h2>Intro</h2> <p>The attention-based web advertising business model is contingent on owning the screen real estate users see. The longer people stay on your website, the more revenue you can generate from advertising. In other words, attention-based advertising platforms are in the business of capturing user attention and selling it to advertisers, which incentivizes platforms to impose high-switching costs and create walled-gardens. And the higher the switching costs, the worse a company can treat its users — a phenomenon known as <a href="https://en.wikipedia.org/wiki/Enshittification" target="_blank" rel="noopener"><u>enshittification</u></a>. </p> <p>Social media platforms are infamously one of the largest, if not the largest, beneficiaries of the attention-based web advertising business model, holding user attention hostage through engagement farming, algorithmic feeds, and infinite scrolling. As a byproduct of this business model decision, social media platforms don’t interoperate with one another. In fact, some platforms have even invented <a href="https://iapp.org/news/a/can-a-cease-and-desist-notice-create-cfaa-liability-scrapers-beware?utm_source=chatgpt.com" target="_blank" rel="noopener"><u>laws</u></a> that make interoperability illegal unless they give permission. A Twitter post lives on Twitter. A Medium blog lives on Medium. A YouTube video lives on YouTube.</p> <p>Emerging open publishing standards like <a href="https://atproto.com/" target="_blank" rel="noopener">atproto</a> and <a href="https://standard.site/" target="_blank" rel="noopener"><u>standard.site</u></a> hold the potential to move social media away from enshittifying, attention-based advertising business models towards intent-based advertising business models that deliver sustainable revenues without sacrificing openness, interoperability, and low-switching costs.</p> <h2>Google Search and Intents</h2> <p>While attention-capturing social media applications try to keep users on their platform for as long as possible, Google built its flagship search business on the opposite message. Instead, Google aimed to help users find what they were looking for and then send them quickly to other sites. In a 2004 interview, Google co-founder Larry Page <a href="https://www.techdirt.com/2024/05/21/decentralized-systems-will-be-necessary-to-stop-google-from-putting-the-web-into-managed-decline/" target="_blank" rel="noopener">said</a>, “Google’s goal is to get the user off of Google and to the right place as fast as possible.”</p> <p>Rather than monetizing attention, Google monetizes intents. Users come to Google with clear intents — “car insurance quotes,” “Nike running shoes,” “best CRM software,” etc. — which are valuable to advertisers because they indicate high probability of near-term action. If a user is searching for “hotels in Miami,” there’s a high likelihood the user is booking a trip to Miami. Google monetizes these intents by auctioning priority placement in search results to advertisers. </p> <p>While attention-capturing social media platforms and intent-capturing search engines both monetize via advertising, search benefits from, and is therefore incentivized to preserve, the web’s openness and interoperability, while social media platforms work against it. </p> <h2><a href="https://Standard.site" target="_blank" rel="noopener">Standard.site</a> and Intentional Social Media</h2> <p>Social platforms are inherently anti-intent because they monetize attention. Users typically go to social media apps for entertainment and stimulus-inducing reels, not declared demand. The world needs <em><strong>intentional</strong></em> social media to move us away from addictive algorithms (recent <a href="https://www.frontiersin.org/journals/human-neuroscience/articles/10.3389/fnhum.2024.1383913/full" target="_blank" rel="noopener">studies</a> even prove that short-form videos weaken one's ability to focus) and towards social platforms that drive value through outcomes and intent resolution.</p> <p>By providing an open surface area for aggregating, discovering, and interacting with social objects across the web, open protocols like <a href="https://standard.site" target="_blank" rel="noopener">standard.site</a> and <a href="https://atproto.com/" target="_blank" rel="noopener">atproto</a> provide the foundational infrastructure to create intentional social media apps.</p> <p>Standard.site is a shared vocabulary, a lexicon, that defines how long-form content should be structured, stored, and discovered on atproto — the identity and data protocol underpinning Bluesky. While standard.site is built with long-form content in mind, in theory, the standard could work for any content type. Structurally, standard.site wraps web URLs into portable social objects that can be globally discovered and interacted with across integrated clients.</p> <p>So, what does an intentional social media app look like?</p> <p><strong>Whereas Google connects people with information, intentional social media apps connect people with social objects in a unified portal — subscribers to publishers, like-minded people to communities and events, and scientific researchers to relevant papers — akin to a social web browser.</strong> </p> <p>Instead of monetizing attention, I imagine this social web browser-like tool would monetize similarly to search engines, resolving user intents by facilitating connections between users and social objects in social-rich contexts and spaces. </p> <p>Reddit is probably the closest mainstream social platform that shares the properties of an intentional social media app. Reddit users typically arrive with a specific question in mind or a particular topic they’re looking to engage with other people on. However, as an attention-capturing platform, Reddit is a closed system — identity is platform controlled, communities are platform-bound, and API access is permissioned and monetized. This means that conversations happening outside of Reddit don’t populate within Reddit and vice-versa.</p> <p>If you’re looking to engage in a conversation about a particular topic, say the New York Knicks, why would you want the conversation to be constrained to the Knicks subreddit when people are talking about the Knicks across the internet in Twitter posts, YouTube comments, Instagram stories, journalistic articles, etc.</p> <img src="https://cdn.bsky.app/img/feed_fullsize/plain/did:plc:f6vlxmukwskrx7aopup4uyvh/bafkreignsyausrigd75njhwxzfc3ndzsyocf3d2acpzupk2eewakpr7jha@jpeg" alt="" style="aspect-ratio: 1111 / 993; width: 100%; height: auto; object-fit: cover" loading="lazy" /> <img src="https://cdn.bsky.app/img/feed_fullsize/plain/did:plc:f6vlxmukwskrx7aopup4uyvh/bafkreichncnlchea4zzms6rde3egxqygrlp5xlrbk5y3tn7kg7tzb7vyfu@jpeg" alt="" style="aspect-ratio: 1143 / 981; width: 100%; height: auto; object-fit: cover" loading="lazy" /> <p>Instead of Reddit users having to manually clip screenshots from third-party sites, all social objects across the web relating to the Knicks should be able to be natively referenced within the Knicks subreddit community, creating native social linkages across Twitter posts, YouTube videos, blog posts, and podcasts under the same contextual window. With standard.site, it becomes possible to build an open and meaningfully larger and more connected Reddit-like app that builds communities and context-rich spaces around a universe of portable social objects, whether they live on platform or off.</p> <h2>What the Future Holds</h2> <p>With AI being all the rage right now, a question I often get asked is “what does this all mean for AI?” AI agents are the ultimate intent machine. An AI agent’s core objective is to resolve user intents, and so intentional social media apps are perfect complements to AI agents.</p> <p>AI agents thrive in open, structured, and machine-friendly environments, and lexicons like standard.site provide exactly that. Today, each social platform has a distinct API and access is permissioned, making it difficult for an AI agent to uniformly participate across all social platforms on behalf of a user. On the contrary, standard.site is an open, standard API for discovering and interacting with social objects that anyone, even AI agents, can tap into.</p> <p>One developer already deployed an AI agent named <a href="https://bsky.app/profile/koio.sh" target="_blank" rel="noopener"><u>Koios</u></a> to atproto who then proceeded to write a <a href="https://koio.sh/p/00000mk1weyur" target="_blank" rel="noopener"><u>blog post</u></a> that extolled the interoperable publishing virtues of standard.site. It is still early days here, but I’m excited to see more experiments played out.</p> <p>In addition to experiments around AI, I’m excited to see how mixing and matching different lexicons (standard.site for long-form publish, events, bluesky posts for microblogging, podcasts, videos, etc.) can expand the market for social media beyond stereotypical feeds and reels. Social interactions occur any time people coordinate, signal, and interact around an object, whether it be a review on an e-commerce site, a comment on a code repository, edits to a Wikipedia page, or a response to a question on Stack Overflow.</p> <p>One of atproto&#039;s key affordances is decoupling social objects from interfaces, enabling social interactivity and identity to seamlessly traverse across platforms. Social is much bigger than microblogging and long-form publishing, and I expect that atproto entrepreneurs will build novel interfaces and user experiences that bundle these cross-platform contextual relationships into a unified worldview.</p> <p>Lastly, I am hopeful that the introduction of private / permissioned data will broaden the design space for intentional social media by moving us away from skeuomorphism. </p> <p>Imagine you&#039;re in a closed group chat or event page for a party looking for people interested in a particular niche topic or interest. Since atproto enables users to have a singular identity across both private and public spaces, it becomes possible to import global contextual social search into local and private spaces, something that wasn&#039;t possible before.</p> <p>If you enjoyed this article, let&#039;s chat. If you didn&#039;t enjoy this article, let&#039;s chat. It&#039;s still early days here, and we are all trying to figure out what sustainable business models look like in the atmosphere.</p> <p> </p>

Activity on app.kosakata

Activity on app.kosakata

Activity on app.kosakata

Activity on app.kosakata

Activity on app.kosakata

Chinese Example Sentence Test

Activity on app.kosakata

Activity on app.kosakata

Activity on app.kosakata

Activity on app.kosakata