House of Nettles: #article

Becoming a Video Game Scientist Part 1: Archipelago

2026-03-04T11:52:46Z

The bulk of my hobby time for the past six months or so has been spent not playing video games, nor yet creating them, but autopsying them. Layer by layer I peel them apart, examining every cartilaginous connection and noting down how each muscle pulls on the bone structure beneath. I am building on the work of those who came before me, a great berth of knowledge at my back and many fine tools at my hands without which my task would be too overwhelming to contemplate. I give back to this world by crystallizing the knowledge I find into forms that may be re-used and built upon long into the future.

In plainer terms, I've been spending a lot of time reverse-engineering From Software games.

2023: Dipping my Toes

A couple years ago, my friends and I learned about Archipelago, a system of interconnected game mods and related tooling which supports what they call a "multiworld randomizer". You may already be familiar with the concept of randomizer mods, in which the items within a game are shuffled about at random while still tracking enough game logic to ensure that the game can be completed. They're popular as ways to bring fresh life to games one has already played many times over, and can be particularly fun to play in a racing context where the strategy around figuring out how to proceed can be more complex than even the game mechanics themselves.

A multiworld randomizer takes this concept and expands it beyond the boundaries of a single game. Archipelago is able to randomize items across many games and connect them all through the internet, so that a Hollow Knight player in Minneapolis can find bombs for a Link to the Past player in Seattle, which lets them blast through a wall and find a Super Mario World player's ability to run. The possibilities are limited only by the imaginations and hacking abilities of a community of volunteer developers.

My little group quickly decided we wanted to give this a try. Looking through the list of available games, much shorter then than it is now, the one that appealed to me most was Dark Souls III. I consider the From Software oeuvre to be largely masterpieces, and while DS3 isn't my favorite^[1], it was the one that was available at the time.

The Old Mod

Unfortunately, the implementation of the game was not very good. No shade on the dev—modding with these games is difficult, as I was soon to discover. They use a totally idiosyncratic game engine that was originally built for Demon's Souls and has haphazardly accrued new features ever since; the whole thing is a mass of many different custom file formats all held together with difficult-to-decompile (and in some cases intentionally obscured) C++ code. But the player experience was not so great.

The core problem was this: because it was so difficult to figure out how to do anything at runtime in these games, the only thing the mod was really able to influence was the moment the player received an item. They could see which items the player was getting and replace them with something else. What's more, this didn't work with items purchased from a shop.

The core principle of a randomizer is the separation of "items" from "locations". A location is a place or situation in a game that gives the player something; the item is the thing the player gets. Archipelago works by knowing all the items and locations in each game, and assigning new items to each location. This poses a problem when all you know is which item you received: unless that item is unique across the entire game, you have no idea which location it came from. The Dark Souls III mod's solution to this was simple: only unique items could be randomized.^[2]

The other critical problem was that the mod had no way of telling the player what item they were receiving for another player's game. Foreign items, as they're known, were simply replaced with Prism Stones in the local player's game. Many was the time I'd hear a player say "wow Natalie, thanks for that!" only to have to ask them what I'd even given them.

The New Mod

It didn't take much of this before my mental wheels began to turn. Thanks to my previous work dissecting enemy AI for Elden Ring, I was familiar with the basics of how From Software games were assembled. I had also done a fair amount of Sekiro randomizer playthroughs using thefifthmatt's randomizer, which operated entirely statically: rather than injecting a mod into the game which messed with items during play, it created new variations of the game's files which stored which items existed where.

Imagine all the data in the game laid out in a giant spreadsheet^[3]. Each item has its own row with all the information about it, each enemy type has its own row with all of its stats, and so on. The static randomizer works by rearranging cells in this spreadsheet and using a small off-the-shelf mod to inject the new data into the game. The result is that any item can be moved to any location, including shops and locations that originally held non-unique items, without any extra reverse-engineering required.

I put the pieces together and realized that you could do the same thing for Archipelago. It couldn't all be done by statically modifying data files, since at the very least you'd need to tell the Archipelago server which locations you found and receive items from other players. But you could statically add new items which included in their metadata which Archipelago locations they appeared at, and for foreign items you could even give them proper names like "Michelle's Amulet Jewel" so the player could see what they were and who they were for. Then the existing item-handling function would be able to distinguish any location in the game^[4] and swap in the correct item.

Once I'd had this idea, I knew I couldn't just let it sit stagnant in the back of my mind. I forked the DS3 Archipelago mod and thefifthmatt's randomizer and I set to work. It wasn't long before I had a proof of concept up and running: a DS3 Archipelgo mod that used a variant of thefifthmatt's static randomizer to load all the Archipelago metadata into the game items themselves. It worked as-is with existing DS3 Archipelago worlds, and as an added bonus it could randomize enemies as well (a feature we got for free from the existing static randomizer).

I could have left it there, submitted a pull request, and let the existing maintainer do with it what he wanted. But by this point he was largely inactive, and I've always found it difficult to let progress go unmade when I can clearly see how to do it. So rather than the end of my story, this is just the beginning.

To be continued in Part 2: Ghidra

That honor belongs to Sekiro: Shadows Die Twice. This was also the first one I played, but after replaying most of them numerous times I still have no hesitation putting this in my top slot. ↩︎
There actually was another workaround that could optionally be used. For certain classes of items, like upgrade items, the randomizer would count how many the player received over the course of their run, and with each successive item give the player a different item. It would keep track of how many Titanite Shards were guaranteed to be available before a certain point in the game to know where it was safe to put items that unlocked progression. This technique didn't play very well, but it was certainly clever. ↩︎
I haven't ever heard this explicitly verified by an employee, but rumor has it that From Software literally does just use a big Excel spreadsheet for all this data internally before it gets serialized into a more efficient format for including in the game. ↩︎
...except for locations in shops. Those still didn't work. ↩︎

So-Called "AI" Cannot Program

2025-06-20T09:21:08Z

I contend that it is impossible for so-called "AI", by which I mean the crop of convolutional neural network-based pattern-filling tools that is currently in the throes of a hype cycle that puts crypto and NFTs to shame, to meaningfully do programming. As more and more of the mind bogglingly rich tech oligarchs lead their followers into the delusion that these tools are useful in any way remotely comparable to their cost, I think it's worth taking the time to articulate exactly why this is, even if it's already intuitively clear to the more thoughtful practitioners of the craft.

We must begin, as always, with a clarity of the term under discussion. "Programming", in its broadest sense, is the act of making a computer do something—but "do something" is itself vague, so we'll need to dig a little deeper into that. To understand it, please bear with me as I establish some useful terminology.

A Brief Meander into the Philosophy of Language

Let's talk about syntax and semantics. If you're a programmer, you may be familiar with these terms as they're used to discuss programming languages. If you're not, that's fine too. I'll do my best to explain them in brief.

Syntax and semantics are two related ways of talking about an abstract structure. Although the concepts were originally developed to describe human languages, they can be used by analogy for all sorts of things, which is what I'm building up to here. They're very useful particularly for understanding how humans^[1] relate to those structures and relate those structures to the world as they understand it.

"Syntax" is simply the structure itself, in all the technical detail of how it fits together. The syntax of a human language is the way its words and sentences fit together, what's "allowed" and "not allowed" by the subtle and mercurial rules we all internalized as children (or adults, for second languages and beyond). In English, "ran boy" is not a valid sentence due to the language's semantics, although most native speakers would probably guess that it means "the boy ran". Noam Chomsky wrote "colorless green ideas sleep furiously" as an example of a sentence that is syntactically valid despite being nonsense.

The fact that that sentence is nonsense is the domain of "semantics". Semantics are the human interpretation of a structure, the meaning we ascribe to it and to the best of our ability share with those around us. For example, although the syntax of the sentences "みなさんはピッコロさんが大好きだよ！" and "Everyone loves Piccolo!"^[2] are completely different, their meaning—their semantic content—is exactly the same. And that meaning is something that humans bring to them, not something that is in any way intrinsically associated with those particular wiggly lines in that particular order.

Although syntax is a very useful concept in its own right, in this post I really want to focus on semantics. So remember: semantics is meaning, and specifically it's meaning applied to a structure from outside by humans.

A Better Definition of Programming

Let's go back and use our newfound understanding of semantics to improve our definition of "programming". Programming is the act of making a computer enact a semantic task. The computer's silicon internals, its RAM and hard disk and even its pixels, are the syntax here. It has no intrinsic meaning, just a set of ones and zeros and a very complex set of rules for transforming them. The semantics are the human interpretation of what it's doing and why, the understanding of those numbers and pixels as a map to the nearest ramen joint or a simulation of a puppygirl begging for treats.

At this point, the easy way out would be to stop here and say "LLMs aren't sentient and thus they can't produce semantics and thus it doesn't count as programming, end of blog post." But that's not what I'm trying to say here. After all, if a human tells an LLM "write me a puppygirl virtual pet", the human is still there to apply semantics even if the neural network has no mind to understand any part of that^[3].

The problem is that, although this is a good bird's-eye-view understanding of what programming is, it doesn't capture some of the most important aspects of the actual practice of writing code. Despite what dramatizations would have you believe, when a human sits down to write a puppygirl virtual pet, they don't just start typing as fast as their fingers can move and stop when they have a good girl who wants walkies. Most of the time one "spends programming" is actually spent thinking^[4], not typing.

There are broadly two categories of thing one might think about when programming. The first is what I'll call scut work: the tedious and annoying process of trying to figure out how to get things up and running, why stuff that you expect to work isn't, and how to use that damn API which looks like it was designed by someone whom you personally wronged and is out for revenge. You can even extend this into the realm of pure typing to include writing boilerplate, which may be more or less of a factor depending on the details of a given project.

Scut work, I will grudgingly admit, is something that LLMs can be useful for. I contend that their use is not at all worth the costs, not just in narrow terms of the environmental and moral externalities of creating and operating such tools relative to the alternatives, but also in broader terms that I'll elaborate on below. But this is not covered by my stronger point that LLMs can't program, because scut work is not the true heart of programming.

The true heart of programming is the second category of things a programmer might think about: the semantics of the program.

The inevitable outcome when someone sits down to write a program with some high-level semantic goal is, one way or another, hitting a wall. This can come in various forms: perhaps they run their code and see behavior they didn't expect, or they get a compiler error, or they can't figure out how best to name their classes. Whatever the symptoms, the cause is the same: they've run into the limits of their initial understanding of what they were trying to do.

The preface to the first edition of Abelson and Sussman's seminal introductory programming textbook Structure and Interpretation of Computer Programs famously says "We want to establish the idea that a computer language is not just a way of getting a computer to perform operations but rather that it is a novel formal medium for expressing ideas about methodology. Thus, programs must be written for people to read, and only incidentally for machines to execute." A program is an expression of semantic "ideas about methodology" in a syntax rigorous enough for that methodology to both be unambiguously understood by other humans and to be executed swiftly and automatically by a machine.

That level of rigor is far beyond what humans use to think about the world around them or the problems they're trying to solve, and it's the impact of the programmer's human semantic understanding on the computer language's need for rigor that creates the inevitable wall. But unlike the scut work, climbing that wall isn't a tedious chore. It's absolutely logically necessary. It is intrinsically the process of refining your loose semantic model to close some gap you didn't even know was there.

This is the beauty of programming, the core thing that makes it (as A&S say) a truly novel medium for communication, beyond even the mathematical rigor that prefigured it. Because a program will not work^[5] until it is a reasonably faithful syntactic representation of the programmer's semantic ideas, programming is a forcing function for making loose semantic ideas rigorous—to refine them. The code then is merely a static representation of the current best refinement of the semantics, if I may further abuse human language.

Let's use this to produce a new, even better definition of programming. Programming is the iterative refinement of a semantic task to the point of sufficient rigor that it can be executed mechanically.

Bringing it Back to So-Called "AI"

A human programmer isn't being paid to type, or even to do the scut work—although being able to do the scut work as well is essential, in the same way that a carpenter needs steady hands even though their job isn't holding their hands still. A programmer is paid to refine semantics. They are asked to have a clear semantic understanding of the goal to be achieved and to make choices about what achieving that goal really means in practice.

Avoiding this work by handing it off to an LLM means avoiding making those choices. These programs cannot choose, not just in a philosophical sense, but in a practical one. They are pattern-generators. Even granting the assumption that the technology will advance substantially beyond where it is today and be able to create large-scale programs that run, they will always produce the most likely syntactic structure given their inputs. In the best case, this will force the program's semantics to look like the average semantics of every similar program in the LLM's inputs. More likely, they'll produce a set of abstractions that are superficially applicable but subtly unsuited to the specifics of the task at hand.

Here's the thing: if a task was so similar to existing programs that the average structure would work, even with a few tweaks, the existing programs could have been used instead! There are many well-established means of abstracting and re-using repeated structures in programming. The entire industry is built on layers and layers of abstractions created from tasks that humans realized had enough in common to create a shared semantic base. LLMs bring nothing of value to this process—when all is said and done, they're just a fancy copy/paste operation, and like any copy/paste operation they may be quick do in the moment but they leave you with so many copies of essentially the same thing that they become impossible to maintain.

This brings me back to why LLMs are bad for scut work, too. The sheer unpleasantness of programming scut work is what drives the creation of all these abstractions. The C programming language exists because tracking pointers by hand in assembly is exhausting. C++ exists because writing boilerplate for objects in C is exhausting. Java exists because handling memory management in C++ is exhausting. And on and on—every innovation in tooling, frameworks, and programming languages is driven by the desire to reduce scut work. To shrug and say "I'll just use an LLM for it" is to stagnate those innovations, because LLMs make nothing better in their passing. They only paper over what is already bad.

I'm using "humans" here as a shorthand for "sentient beings with minds remotely comparable to humans", because humans are the only such beings known to exist. I'd use "people" instead but it doesn't have an adjectival form nearly as graceful as the adjective "human". For whatever it's worth, I think the same thing would apply to any non-human species or even digital sentience, which to be clear is very much still the domain of science fiction and is not a realistic end state for the branch of neural network technology that I'm discussing here. ↩︎
Thanks to Cassie and Erik for providing these example sentences. All the languages I know well enough to create examples in have annoyingly similar syntax to English, which makes them unsuitable for this example. ↩︎
This is the footnote where any other author would explain Searle's Chinese Room thought experiment. I'm not going to do that. The Chinese Room pisses me off. Searle pisses me off. If you ask me to elaborate one-on-one I will, but you're buying the drinks. ↩︎
Unless your compiler is really fucking slow, in which case maybe you spend most of your time waiting. I got a lot of reading done at my first programming internship. ↩︎
Depending on the laxity of the programming language, it may still run, but it won't work in the sense of doing what it's supposed to do. It would just be the programming equivalent of "colorless green ideas sleep furiously"—syntactically valid but devoid of meaning. ↩︎

A Non-Technical Intro to Webmentions

2024-10-22T02:00:25Z

I keep bringing up "Webmentions" in the context of discussing the sociable web and advocating for more people to adopt more social technologies on their websites, but I always run into a wall: there's no good place to link people to so they can understand more about what that means. All the existing explanations I've found are deep in the weeds of how Webmentions work on a technical level, which isn't a very helpful place to start for people who just want to post.

I want to fill that gap with this post, and give people who don't know the ins and outs of HTTP a working understanding of what Webmentions do and how to get them up and running for your site. To that end:

Webmentions are a way to let a website know that you linked to it.

That's it! At it's core, it's just that simple. If a website supports Webmentions, you tell it "Hey, here's the URL of a page with a link to you", it double-checks that the link actually exists, and then it does what it pleases with that information.

What can you do with Webmentions?

Notifications

The simplest thing you can do is just look at the Webmentions you receive like a notifications feed on a social media site, and appreciate that people like what you're up to. I get all my Webmentions delivered to me as an RSS feed (more on that below), and I'll always check out the links to see what people are saying.

My Webmention notification feed on webmention.io

In addition to being the easiest to set up, I think this is actually the most useful thing to do with Webmentions. Having a way to see when people reply to your posts makes conversation possible and seeing people's appreciation encourages performance. Even if you never go beyond using Webmentions as pure notifications, it's a great way to become more interconnected.

Replies as comments

If someone makes a post on their blog that's replying to yours and sends you a Webmention, you can display that reply like a comment underneath your post. This is pretty common for out-of-the-box Webmention plugins, like this one for WordPress^[1]. You can see it in action on Liz's WordPress blog where my reply on this blog shows up as a comment on hers, with my avatar and the original posting date and everything.

My post replying to Liz

My post as a reply on Liz's blog

Making this work nicely requires a bit of setup on the part of the page that contains the link, though. A computer isn't smart enough to take any old webpage and figure out which parts of it are the author's name, the author's avatar, the text of the reply, and so on. In order for all of that to work nicely, the linking page needs to use h-entry metadata to explicitly indicate all this information. Fair warning: h-entry is unavoidably a bit technical to set up. since it requires editing HTML. That said, some blog hosts like micro.blog do support it out of the box, and others like WordPress have plugins to handle it for you.

Even if a page is mentioned by something without any metadata, it can still display a link back; it just won't have much additional information. Here's what a bare, metadata-free mention looks like on my blog:

Reposts and likes

In addition to replies, which are basically reposts or links with additional content, Webmentions + h-entry can express a number of other relationships between posts. (Technical moment: which relationship you express depends on which class attribute you put on the link. For example, <a class="u-in-reply-to" href="..."> indicates that you're replying to the linked post. A link without a class is just treated as a generic mention.)

Reposts (class="u-repost-of"): This usually indicates that you're reposting some or all of the content of another post with minimal additional content of your own. This is modeled on the Twitter retweet more than the Tumblr reblog, which I think is better expressed as a reply.
Likes (class="u-like-of"): This indicates that you're expressing appreciation of a post, also usually without additional content. You might, for example, have a page on your website that's just there to contain "like" links. I personally don't use this one, because I find commenting to be a more personal way of expressing my appreciation, even if it's just a little ":yeah:".

There are more relationships between pages that aren't as widely supported, like RSVPs to events, reviews of media, translations of posts, and so on. But these are the main ones, especially if you're coming at it from a social angle.

How do I send Webmentions?

Some sites that support Webmentions will provide you a handy text box into which you can paste the URL of a page which links to the page you're on. My blog has this at the bottom of each page, for example. But not all pages have a form for this, and even if they did it would get pretty annoying to do manually all the time.

Blog software and plugins that support Webmentions will usually automatically handle this for you, sending them out every time you make a post (or edit it, so that websites know to update any Webmentions they display). But if it doesn't, don't despair—webmention.app has your back. It takes a bit of doing, but you can set it up to send Webmentions every time your RSS feed updates^[2], which will work with just about any blog under the sun.

Edit: Apparently the old documentation^[2:1] for setting this up now costs money, since IFTTT doesn't provide free webhook support anymore. Fortunately, Karma Chameleon has new instructions for setting it up with make.com instead, which is free.

How do I receive Webmentions?

Okay, this bit is unavoidably at least a little bit technical. Unless your blog platform supports Webmentions out-of-the-box or through a plugin, you're going to have to edit some HTML. But don't worry! I'll guide you through it.

I use webmention.io to receive Webmentions, and I recommend it. It's got a straightforward, sensible UI—the only wrinkle is that its signin process may not be what you're used to. But I'll guide you through that as well.

Choose an email address that you're okay exposing to the world. If you don't want to use your primary email, Fastmail and Proton both provide free accounts.
Let's say your domain name is me.example.org and your email is me@example.org. Add the following to your website's HTML:
```
<link rel="me" href="mailto:me@example.org">
<link rel="webmention" href="https://webmention.io/me.example.org/webmention">
```
If possible, these should go just before the </head> in your HTML. That said, if you can't edit your layout directly, these should work anywhere on the page. They won't be visible; they're just there to tell computers
- this is an email address that belongs to the person who runs this website, and
- this is the URL to which to send Webmentions for this page.
The only real constraint is that they need to be on both the root page of your site and on each page that you want to receive Webmentions.
Go to webmention.io, type in your site's base URL under "Web Sign-In", and click "Sign In". It's important that you don't include a path in the URL (good: https://example.org/, bad: https://example.org/blog/my-post). Otherwise the URL for the <link rel="webmention"> we added above won't be correct.
This will take you through a sign-in flow where a login code will be emailed to the email you put on your website above, which verifies that you are in fact the person who owns this website.
Once you're signed in, you'll be redirected to webmention.io/dashboard. I recommend bookmarking this, because for whatever reason there's not an easy way to get there from the main page without signing in all over again.

Once you're at the dashboard, you're good to go—webmentions will be visible there, or you can get them in your feed reader by going to "Settings > Mentions Feed" and copying the mentions.atom URL into your feed reader^[3].

Webmentions from other websites

Whether you're using your blog's built-in Webmention support, webmention.io, or some other system, you can use the brid.gy app to convert replies and comments on a bunch of other social websites into Webmentions on your site. I do this for the Fediverse, and you can see for example a bunch of Fediverse discussion below this post. It's sometimes a little shaky, especially because it doesn't always preserve threading accurately, but it works!

It looks like brid.gy can also act as a Webmention plugin for Tumblr, WordPress, Blogger, and Medium. I haven't tried this personally, so I can't vouch for it. But it might work great!

Go out there and get mentioned!

I really recommend getting Webmentions set up. It does involve a bit of silly rigamarole, it's true, but it's worth it to become part of the sociable web. And the more people start using this, the more it'll become a valuable feature for blogging platforms to support without any hassle. And now at least it'll be a bit demystified for those who want to understand what it's all about without reading W3C specifications.

WordPress has been the talk of the town lately for the WordPress.com CEO's non-stop heinous antics, so I hesitate to give it a full-throated endorsement. That said, it's by far the easiest way I know of to get up and running with all the IndieWeb technology from zero baseline knowledge, and fortunately the engine itself is open source so it's unlikely to collapse completely. Just maybe choose a hosting provider that's not WordPress.com. ↩︎
If you do choose to follow this documentation, note that it includes adding &token=... to the hook URL. You can get a token by signing into webmention.app with a GitHub URL, but I don't think you really need to. Without a token, you can only make four requests per URL per hour, but if you're only sending one request per page anyways that won't matter. So you should be able to safely leave off the &token=... bit. ↩︎ ↩︎
There's also a mentions.html URL, but you can ignore that. It uses an h-entry based feed format that's pretty cool, but not at all widely supported whereas Atom is pretty much universal. ↩︎

A Sociable Web

2024-10-19T03:27:07Z

The process of building out this site has inevitably involved design choices about how to engage with various technologies and other people's web presences. It's a truism that you can't solve social problems with technology, but social media has made it just as clear that technology does shape the social dynamics that emerge in the spaces it mediates. This drives me to wonder: as more of my friends and friends-of-friends move to individual websites and blogs, what social dynamics does this give rise to? And what different technical designs could improve those dynamics?

I think it's most interesting to approach this question from the social direction rather than the technological. Our first priority should be a set of social goals for interacting on the internet, and only with that understanding firmly in hand can we start usefully interrogating the way technology gives rise to or fights against the sort of interactions we want. My ultimate aim is to articulate a clear vision of a way to interact with people's websites that's not just a pastime or a research process, but that can meet social needs—to imagine what I'll call a "sociable web".

I'm using the term "sociable web" as a conscious contrast to "social network". Social networks have unavoidably shaped our understanding of what form of socialization is possible online, and I don't even think it's always been for the worse. But now they are fragmenting and rotting, and whatever was good about them—their casual usability, the massive network effect of "all your friends are here"—is falling away like so much decayed flesh. I want to envision something new, and when doing so I often find that a new name can help.

"Web" versus "Network"

I'm not interested in a "network" in the sense either of a single company overseeing many users or a decentralized collection of nodes like the Fediverse that is nevertheless tightly coupled technologically. I don't think either of these forms are sustainable in a capitalist world. They are either fed by boundless venture capital coffers which inevitably move to capitalize their userbase, or they're built on the backs of massive amounts of unpaid labor and poorly-understood power structures. The exploitative conditions under which these networks are produced run downstream and affect their social dynamics.

I'm interested in a social form that uses as its foundation the web itself. Where anyone can participate just by having a website. Where the existing networks are, to some degree, part of that form already simply by virtue of being accessible over HTTPS.

"Social" is unopinionated. "Social" is throwing a bunch of people into a room and seeing what happens. A party can be social, but so can a witch hunt. "Social" is thinking about the technology as primary and allowing the social dynamics to fall out from that. When things are built to be merely "social", they force the real humans who use them into chaotic interactions both healthy and harmful, a freeze peach nightmare that only a libertarian could love.

I want more than that. I want things to be sociable, to be friendly, to be fun. I don't mean to try to avoid conflict entirely, which is neither possible nor truly desirable. But to envision a sociable web is to envision a place where interactions that make your world richer are easy to realize, and those that make your life worse easy to curtail. I want to be able to chat in public without the world overhearing, to meet new people and be able to block creeps, and to have conversations without being drowned in a sea of bad faith.

The first step towards understanding what a sociable web could be is to understand the social goals we're trying to achieve. Social media, for all its doomed flaws, is a huge part of millions of people's lives for a reason, and it's not just because it exploitatively monopolizes its users' attention. It meets real needs people have for connection and communication in a space that's decoupled^[1] from physical proximity. If we can identify the healthy interactions and cut them away from the rotten habitat that is (for now) their only home, we can start building a web that is genuinely for us.

Conversation 💬

This is the core of what it means to socialize: saying things, having people hear them, and listening to responses. If you're not saying anything, you're just a lurker—not a bad thing to be, but also not a participant in a social dynamic. If no one hears you, you're just writing a diary. And if no one responds, then you might as well be. Conversation is the fundamental building block of human interaction. Without conversation, you have nothing. Silence. Void.

Conversation, like the libertarian "social", is unopinionated. It can be vapid or it can be enlightening; it can be uplifting or it can be miserable. But this flexibility is also an asset: when conversation can be both silly and serious, comic and tragic, it gives people room to bring their whole selves (or whatever slices or performances of themselves that they want) into the space. This is one reason I make a point of posting jokes on this blog, even when they're incredibly dumb.

Conversations are also (at least in physical reality) local. Even when conversing in a public space or a crowded party, most of the time you're just talking with a small number of people. People may drop in and out, but there's no expectation that everyone in the room will hear everything you say.

...and Technology

I see conversation as one of the key failures of the old-school "blogosphere". You could post all you wanted, but the means of actually turning those posts into a conversation was extremely limited. Comment sections met this need to a degree, but without the ability to make a long-form response to a long-form post and have the original author actually see it a culture of back-and-forth never blossomed.

The classic blog form also felt, for better or for worse, like it had to be long-form and serious—another failure of conversation. I think part of this was just a problem of snowballing cultural precedent, part of it was a failure of performance (which I'll get into below), and part of it was a vicious cycle with the inability to reply in kind. Small posts are compelling to write because they function as an invitation to casually interact with the author. Once that interaction is curtailed, there's far less reason to write small posts; if you write fewer small posts, you'll build up less of a network of people who will interact casually.

Tumblr, to its credit, did solve both of these problems. Its original stated goal was to be a blogging platform that encouraged smaller posts, but it also supported reposts^[2]—and in doing so, ended up metamorphosing from its blog-oriented origins into an out-and-out social media site. I think this is the critical distinction, in fact, between the era of blogs and the era of social media. Blogs are built around a one-way flow of writing, where social media is built around conversation. So how can we bring conversation back to individual websites, which are tied to the blog form at least inasmuch as they aren't part of a tightly-coupled network?

It won't surprise those of you who read my post on h-entry that I think reposting is a key part of the answer. Having a way to interact with people that keeps the context of the conversation intact and encourages other people to participate is extremely valuable. But equally valuable is making sure that the conversation is visible to the existing participants. If I can only see replies from people I'm already following, conversation is stifled. I think Webmentions are the best solution for this that currently exists, and I'm planning a more thorough non-technical introduction to them for a future post.

Performance 🎭

Although talking with people is the most important way of being sociable, I think there's another modality that's subtler but still very relevant: performing for people. I mean "performance" in a very broad sense, covering anything that's fundamentally about creating something for others to appreciate. This can be a long, well-reasoned essay or a silly offhand joke; it can be visual art, a poem, or a css crime. I call it "performance" because my heart is in the theater, and because a critical aspect is not just the creation itself but the two-way dynamic between the creator and the audience.

To be able to see the joy or catharsis or edification your creation brings to people in real time is tremendously heady. At its best, it allows the creator to see their work through new eyes and appreciate it all the more. It can be inspirational, motivational, and even open up whole new avenues of creation—for example, I would have never started helping Zandra with Her Majesty the Prince if she hadn't treated it as a performance she was writing live, chapter by chapter.

The potency of performance can be turned to harm, though. The desire for applause can wrap itself around your heart until everything starts to feel like a performance. Performances can be a great joy and a source of profound connection, and having room for them is crucial. But they cannot be the bread and butter of social interaction. Any healthy social situation must have plenty of room for simple conversations with no thought of an audience beyond the participants.

...and Technology

Performance is the handle by which social media grabs hold of your attention and refuses ever to let go. Performance is the food it feeds you as you scroll your infinite timeline, and the heart of the Skinner box it uses to keep you posting forever. This is harmful. Cohost absolutely made the right call in dramatically minimizing this effect by removing metrics such as like counts for posts and follower counts for people.

But like many vital nutrients, while an overwhelming focus on performance is toxic, you need some to thrive. And this is something that individual web sites struggle with^[3]. You can implement anonymous "likes" like Medium's claps or BearBlog's up arrow, but without names and faces this feels hollow and impersonal. And attaching an identity while still making "liking" a post simple and quick all but requires some sort of centralized account^[4].

I think comments are a critical part of the solution here. Even a simple "hell yeah" or "love this!" comment feels personal, and there are plenty of comment systems that make posting without an account relatively painless. And "making a comment" feels enough weightier than "liking a post" that a small signup flow reads as less egregious, even if the end purpose is the same.

Reposts are even more valuable as a form of "applause". Not only do they bring a post to a new audience, but they're the strongest possible endorsement. Nothing makes me feel more like my writing is appreciated than seeing it show up in a friend's link roundup, and I think the roundups people like Dante, Shel, Nicky, Caoimhe, and many more have been doing are among the most valuable contributions to building a sustainable culture for a sociable web.

And I do think culture is the deciding factor for nourishing performance without letting it become toxic. For sure, there are technological aspects as well: Webmentions are crucial for allowing an author to actually see reposts, h-entry makes it far easier to create a repost and adds a lot of expressive power to Webmentions, and the generally slower pace of web-based posting limits the potential for performance to become toxic. But none of that matters if we don't establish a culture where we tell people when we appreciate their posts and share them if we think our friends will appreciate them as well.

Meeting People 👋

Hanging out with your friends is lovely, but friends don't come from nowhere. Everyone wonderful in your life (other than blood relatives) was once a stranger to you, and there was a moment when you first met them. In order for a social context to fully function, to be more than just a group chat, there must be a way to build new relationships with people you don't already know—to have that moment of meeting and to give it room to grow.

There are two parts to this that are equally important. First, you have to have a way to actually literally see new people in your daily social experiences. If you're only ever interacting with the same group of friends, you'll never meet anyone new. There needs to be some amount of circulation: being introduced to friends-of-friends, going to a party, joining a hobby group for something you're interested in. What are the analogues for these the digital world?

But merely interacting with new people isn't enough. Building an actual relationship with someone requires a capacity for balance. There must be a way to break the parasocial asymmetry of merely "following" someone and to instead put yourselves on equal terms. That's not to say that you have to do this with everyone, or that it has to be explicit—it's rarely explicit in person. But if there are structural barriers from moving from "acquaintance" or "fan" to "friend", then all you'll ever be or have is groupies.

...and Technology

I have met the substantial majority of my friends, including many people I consider "in-person friends", either directly through social media or transitively through other friends whom I met through social media. My life would be immeasurably different if I'd never created a Twitter account, and most of that difference would be the connections I would have never established with people who are now beloved friends. But despite all that, I don't think social media is actually very successful at making it possible to meet people.

There were specific historical contexts in which it worked well, to be sure. The little slice of queer Twitter I was part of in the mid-10s, after the most vicious of the cancel wars but before the algorithmic timeline got really bad, was one of those contexts. I'm sure over the years and across the communities there were many more. These were never perfect—I have come to strongly believe that knowing for sure when someone unfollows you or doesn't follow back is intrinsically corrosive to real friendship—but they did function.

But those contexts were always contingent, small loopholes that were doomed to close. The structure of social media, the arc upon which it bends, pulls towards a world where parasocial relationships are everything, because parasocial relationships are profitable^[5]. The parasocial structure already mirrors the distribution of content for profit, and so is far easier to Midas-touch into the gold standard of the internet era: advertising. X dot com is no longer the hot new kid on the block; it has far fewer users than TikTok, Instagram, and YouTube, all of which are fundamentally about sharing media rather than communicating. The only outlier is Facebook, which itself has fan pages and a deeply pernicious algorithm.

This is one place where I think the classic blog structure fares a little better in one key respect: it's very easy to have a balanced relationship with someone. You "follow" someone via RSS or just checking their blog periodically, and this leaves no trace other than the comments or other conversations you explicitly choose to have. It can be parasocial—or at least it could back in the day when there existed very popular blogs with huge followings—but if a blogger appreciates a particular set of comments or reposts they can easily follow the creator's blog, post a few comments of their own, and gracefully unfollow if it doesn't click.

The crucial problem for blogs is the circulation. You need to be able to actually see other people in order to meet them. Friends' reposts and link roundups (which are functionally the same thing) can help you find new blogs to follow, although there's currently not a great equivalent of "hobby groups"^[6]. Comments can help you actually connect with new people you've found, although this really only works if they connect back to the commenter's own web presence, which many out-of-the-box comment solutions frustratingly don't support. I'm sure forbidding homepage links mitigates spam to some degree, but I think it's actively detrimental to establishing new connections, and I encourage everyone to search out comment apps that have a place to fill in the commenter's website (as am I at the moment—there's a lot not to like about CommentBox but this tops the list for me).

Having people repost and respond to your posts is also an excellent way to get a sense of them, because you're seeing them in their native habitat, as it were: on their own blog, writing primarily for their audience. But this does require actually being notified when those reposts happen, which requires the use of a technology like Webmentions on both ends of the interaction.

Avoiding People 🫤

The final pillar is I think the most difficult all around: in a healthy social space it must be possible not just to have healthy interactions, but to avoid unhealthy ones. This is, interestingly, something that's not even entirely solved in person—I imagine that the delicate dance of navigating which of your feuding friends to invite to which parties is a tale as old as the concept of a party itself. But in person you can at least always walk away when someone you can't stand walks up, and if someone is being belligerent enough you can toss them out on their ear.

A healthy social space has to be resilient to both small-scale interpersonal conflict and large-scale bad actors to avoid collapsing under the weight of humans being inevitably human. At the same time, it has to be acknowledged that a perfect solution here isn't possible—someone with a big enough grudge and enough VPNs can harass you through the most robust defense, and it'll never be possible to just edit someone out of your life or community no matter how much they piss you off.

...and Technology

Just like in person, I don't know that any digital platform has come up with a really robust solution here. The state of the art, I suppose, is mutes ("I can't see you") and blocks ("...and you can't see me either") for smaller conflicts and thoughtful human moderation along with suspensions, bans, and IP bans for bigger problems. Unfortunately, all of that depends intrinsically on having a centralized platform.

Even the Fediverse, for all its Twitter mimicry, struggles with this issue. Mutes work fine, blocks mostly work but only on the assumption that other instances play by the rules. Bans only sort of half work—in email terms, you can ban someone from making an account on your server or emailing anyone on your server, and you can ask politely that no one forward anything from your server to them. All of these cross-server conventions are enforced by the extremely coarse-grained threat of defederation, which itself relies on intermediate servers that are not defederated to play nicely. To a degree, it all comes down to little fiefdoms screaming at each other.

And that's still worlds more than you can do on an independent site. Most comment services provide the means of blocking someone from commenting on your blog, but unless it's self-hosted that's the limit. Anything more than that (like making sure they can't cause problems for other people) is up to the comment software's moderation staff, if they have any. If you run your own server and know the IP of someone who's a problem, you can probably IP-ban them, but there's no other way to block them from looking at your site (and even an IP ban is easy to circumvent). There is some intrinsic protection just by virtue of how relatively low-bandwidth the communication channels on individual websites are—it's a lot harder for someone to dogpile you when sending you a message isn't just two taps away for all their followers—but that only goes so far. I don't know if there's a great solution here, especially given that no public presence will ever be truly safe from dedicated harassment.

Smaller-scale avoidance is somewhat easier. RSS readers could in principle filter out reposts from people you don't want to see—the only one I know for sure has this feature is Inoreader (and then only with a paid account), but it's not hard to envision. If all my dreams come true and blogs do truly have a wide resurgence, maybe this will become a valuable differentiating feature. And again, the fact that blogs are much less of a firehose than social media means that it's that much easier to periodically just wince and scroll past a repost of someone you dislike.

Can we do it?

Is a sociable web even something that's possible to create? In 2024, it's easy to despair at computers in general or the internet in particular being usable for anything good at all, and yet here you are reading my words, hoping to make some sort of connection with me or with my ideas. I believe in the power of those connections. It may be true that no corner of the world is safe from the corrosive power of capitalism, but so too is no corner of the world impervious to human flourishing.

I think we can do it. In fact, I think we will—but whether that starts now or years from now depends on us. What precisely a sociable web will look like, I'm not sure. That's why here I've chosen to focus on the social goals first and make the technology secondary. But I do believe that enough technology exists today to at least take a stab at it, and I want you to join me in that.

The deepest virtue of independent websites communicating with loosely-coupled conventions and protocols isn't anything technical. It's the fact that you can just do it. You don't need a platform to exist, you don't even need anyone else to be doing the same thing (although it's more fun that way). You can write a blog and people can read it via RSS or just by loading it up every now and then^[7]. You can hook up as much IndieWeb technology as you want^[8], or just focus on the cultural aspect by writing link roundups and leaving comments on the stuff you read.

The crucial part is that you participate, and help to shape the web you want to exist in. Don't just let the dictates of some megaplatform constrain the way you connect with people. You have the power to make the web a place for people.

To a degree. Language barriers, time zones, networks of existing friends, and barriers to technological access all put some pressure even on the internet to make friends with people physically close to you. But at the very least it expands "close" from "your city" to something more like "your country" or even "your hemisphere". ↩︎
To the best of my knowledge Tumblr introduced the concept of "reposting", although I haven't done any research to verify this so I'm putting that statement down here in an uncertain footnote rather than boldly stating it in the main text. ↩︎
Adding analytics seems at first blush like a solution to this, albeit one that's usually quite invasive. But I think it doesn't address the core issue, because it only measures passive behavior, not active feedback. It's the difference between reading ticket sale numbers in an office and hearing the applause on stage. ↩︎
The Webmention protocol combined with h-entry can explicitly express a "like", but to like a post you must establish a long-lived link to it on a site you control which is absolutely not simple. You could make a generalized "Webmention like" app that's just designed to do this, and it would be usable enough as a browser extension or dedicated button, but that's basically just reinventing a centralized account. ↩︎
One might hope that the fediverse would avoid this issue by not having a profit motive. It certainly is better, to the degree that it explicitly avoids an algorithmic timeline—although I would contend that the global and local timelines pull in the opposite direction. But it (or at least all its most popular instantiations) is so heavily influenced by the structure of Twitter that it inherits its failures as well. And just so we're clear: even classic Twitter was never good, it was just not yet bad enough to completely preclude pockets of valuable community from forming. ↩︎
Back in the day there was a concept of "feed aggregators", often focused on a particular topic, which would take a bunch of RSS feeds—sometimes just posts tagged with a particular topic—and combine them into one output feed. I'd love to see this become a thing again. Blackle's Sortition Social is the only one I know of right now, although it's not at all topic-specific. ↩︎
I learned recently that this is how my girlfriend has been reading this blog and it blew my mind. I have since promised to give her a personalized introduction to feed readers. ↩︎
If anyone wants help with this by the way: open invitation. I would love to get more people up and running with h-entry and Webmentions. ↩︎

one last selfie for the road

2024-10-01T01:09:22Z

Solid chance this is my last selfie to go out on "social media" as such, and so I hope you'll forgive a bit of nostalgia. Posting selfies, especially early in transition, was hugely important for me. For most of my life up to that point I had regarded my looks with suspicion if not outright hostility, and the selfie became the medium through which I began to take ownership of my appearance and develop a sense of style. The first time I looked at a picture I'd taken of myself and realized "oh. I'm hot" was a huge milestone for me.

Posting selfies to social media was an inseparable part of this. As my style became something I actively put effort and care into, having people appreciate it became a reminder that that work was doing something. It wasn't the dreaded numbers, it was the people: people whose taste and style I also saw and appreciated, people who I was friends with and people I just saw around.

When COVID began, I went from taking and posting selfies almost on a weekly basis to nearly not taking them at all. I no longer felt I had any reason to dress up. Wearing lipstick felt foolish when it would just get ruined by a mask. I wasn't getting any new clothes, because I hate having to remember to send back online orders I don't like. My trademark blue hair was growing out, because even after the vaccine made it feasible to be masked indoors with strangers I couldn't bear the pain of trying to find a salon that still required them. I didn't realize until my roots had overtaken me how much the blue had become a part of my self-image, how much it hurt to look in the mirror and not see the self in my heart.

Little by little, I pulled myself out of that hole. Although the pandemic is very much still present and I am very much still taking precautions, I found ways to allow my sense of style to flourish in between those precautions. I'll put on lipstick just because I feel like it. Liz took careful measurements so I could do online shopping with minimal risk of send-backs. Every now and then I'll even strap on a P100 respirator and go to a physical store. I asked around and found a local salon that not only still requires masks, but is queer- and disabled-owned.

And I started taking selfies again. I took selfies to give myself a reason to care about my appearance again. I took selfies as a reward for putting in the work to make myself look and feel good. I took selfies to send to friends, I took selfies to flirt, I took selfies with pals, I took selfies at beaches and forests and birthdays and the precious few weddings that were safe enough for me to attend. I took selfies to remind the world that I existed, and to remind myself as well. And many of those selfies ended up here.

This is the last selfie that I'll post here. It's also the first selfie that I'll post to my website. I've had a few websites before, but I've never had a website where I'd post selfies just for the joy of sharing my style with the world. I hope that there, too, people will appreciate it. I hope the world will remember I exist.

COVID Denialism and Disability Justice

2024-09-21T01:21:00Z

This is a post whose seeds have been bouncing around in my head for years. I always intended to write it up and publish it on Cohost, and so the twilight of that storied website seems like as good a forcing function as any.

In this post, I seek to understand and explain the pervasive phenomenon of COVID denialism from the perspecitve of disability justice, specifically as someone who remains extremely cautious and anticipates doing so indefinitely. It's not intended to excuse this behavior—denialism is actively harmful to everyone the denialist interacts with and fundamentally eugenicist in effect whether or not in intention. But understanding and even empathizing with people who believe falsehoods and do harm can be valuable, especially when they make up such a huge portion of the world and for many of us are inescapably part of our networks and communities.

The first crucial thing to understand is that, if you're at least on board with the basic idea that COVID denialism is a pervasive problem, COVID-19 has already disabled you. Even if it didn't give you long-term side effects, even if you're lucky enough never to have caught it, you have been disabled by it. Or to be more precise: you're disabled with respect to COVID-19. The specific agent of your disability is the society that subjects you to snide remarks and outright harassment for wearing a mask, that closes off opportunities for social interaction and employment to you, that makes it impossible for you to exist within it without putting your health at risk.

This is an analysis based on the social model of disability, a major branch of disability theory that emphasizes the way disability is created by a society's failure to provide accommodations for certain bodies and minds rather than intrinsic aspects of those bodies and minds themselves. To use a lightly clichéd example: my severe nearsightedness doesn't function as a disability, because I exist in a society^[1] that accepts it as "normal" and provides easy access to socially unremarkable assistive devices (glasses), or even invisible assistive devices (contacts) if I so choose. But my sleep disorder is a disability—society doesn't consider it "normal" in the same way, and so it's seen as my personal failure and I have to work to make sure it doesn't affect my relationships^[2] or employability.

Using this model, if you exist in a society that has accepted the uncontrolled spread of COVID-19 as normal, the attempt to avoid catching this disease is itself a disability. Society is organized to systematically deny accommodations like mask mandates, sanitizing ventilation^[3], lockdowns and contact tracing, and free access to vaccines, prophylatics, tests, treatments, and protective equipment. Depending on the specific activity and your risk tolerance, public existence while taking reasonable COVID precautions ranges from requiring serious equipment and preparation to being outright impossible. Even if your body isn't any different than it was in 2019, you are functionally disabled by the society you now exist in.

If you have an existing disability, including Long COVID or even just the incremental damage that each COVID infection does to your immune system^[4], the situation is even worse. Your risk tolerances are likely to be lower, and you may already suffer from ableist limitations on your ability to engage in the world that are compounded by COVID. You become intrinsically multiply disabled just by virtue of the total failure of our social structures to create a world in which it's safe to exist.

The Category Myth of the Abled

I don't remember when exactly it was that I started taking seriously the possibility that I could become disabled at any time. It could have been in college, when my friends would share blog articles that often discussed disability justice. Certainly once my wife was diagnosed with type 1 diabetes in her 20s the point was driven home with terrible force. But for most of my adulthood, I have taken it as a fact of life that something unexpected could happen out of nowhere that would radically change the parameters under which I lived my life.

This is not how most people think.

The categories of abled and disabled are not broadly considered to be permeable. Certainly, people recognize that someone can get in a car crash and lose the use of their legs, but they don't expect it will happen to them. When it happens to someone they know, people desperately claw for reasons why unreasoning fate took this turn—were you driving recklessly? did you eat too much sugar? did you displease G-d?—because they feel a need to shore up their confidence in their own security in the category of "abled".

This is what I'm calling a "category myth": the tale people tell themselves that being abled is a rigid ontological division that cleanly separates those who are from those who are not. Although the categories of "abled" and "disabled" do exist in the ways people's relationship to the society around them is mediated by their bodies and minds, their blurriness and the ease of movement between them is critical to their structure. To insist on clean boundaries, to buy into the category myth, is not just to internalize an inaccurate understanding of humanity but to enter into a way of thinking that is predisposed to doing harm.

Because this category myth isn't just incorrect, it's oppressive. All axes of oppression are wrapped around similar myths. This is why sexists are so threatened by transsexuals, why racists invented the crime of miscegenation. An oppressive mindset demands a clear and permanent division between oneself and one's victims; a mode of thought that relies on clear and permanent divisions is at high risk of enacting oppression, knowingly or not. Those who are unable (or unwilling) to imagine themselves becoming disabled are the ones who do the most harm to people who already are.

Understanding Denialism

I have struggled tremendously with the emotional weight of seeing more and more people in my life succumb, to one degree or another, to denial about the risks posed by COVID-19 and the precautions necessary to proportionally mitigate those risks. At the end of 2021, I saw first Delta and then Omicron cause wastewater levels to rise higher and higher, while more and more businesses dropped mask mandates. Through 2022, friend after friend posted unmasked selfies out in public or invited me out to bars. I fought with my parents to try to keep them taking precautions, and then gave up and just tried to accept the heavy knowledge that their life expectancy would simply be substantially lower than I had thought.

I began to succumb to despair. I couldn't stop wondering, why? Why would people do this, when all the evidence is there for them to see? Succumbing to the "COVID is over" propaganda campaign was part of it, sure, but there was too much overt cognitive dissonance in the way people spoke about the disease for that to be everything. It was in seeking an answer to this question that I began thinking through the ideas I'm presenting here.

In our collective retrospective vision, the relatively brief period of lockdown during 2020 is spoken of as a deeply traumatic thing. And I think that's accurate, but not (or not entirely) for the reason it's usually framed as. I think it was traumatic specifically because it shattered people's category myths of being "abled". Everyone was disabled in that moment^[5]. Something unexpected happened out of nowhere that radically changed the parameters under which they lived their lives. Suddenly they found they had to use assistive devices (masks), they had limited access to spaces and people they had taken for granted, they were forced to consider the fragility of their own lives.

This was traumatic not just—not even primarily—because of the specific facts of the lockdown. It was traumatic because it forced people to confront just how thin the barrier was between them and someone who could only enter the grocery store masked at 1am even before COVID. And that knowledge was, broadly speaking, intolerable. People could not bear the enactment of themselves as part of a category that they considered to be lesser, weaker, broken.

And so they find the prospect of taking COVID precautions, which is to say voluntarily re-entering a state of disability, essentially unthinkable. Even to protect their loved ones, even to protect themselves, it doesn't matter. They cannot allow themselves to consider a world where that would be necessary, because that would be a world where they are all the things they think of disabled people as.

So when you hear someone say, "I can't let COVID control my life," understand that they mean they can't let themselves think about becoming disabled. They can't open the floodgates that would force them to reevaluate everything about how they conceptualize disability. They can't conceive of doing what millions of disabled people do every day and play with the cards they've been dealt.

Whether this will help you feel a bit less despair or not, I don't know. It helped me. I hope at least it can give you a path to a clearer understanding of why people arrive at these arational points of view and what keeps them there, and maybe even give you a few clues as to how to break them out of it.

And, it's worth noting, a minimally precarious position in that society. For someone without vision insurance, the same level of nearsightedness could be a severe disability. ↩︎
The people around me are very chill about this because my friends are wonderful people, but someone with less wonderful friends might have a lot more trouble here. ↩︎
Especially audited sanitizing ventilation. Venues can update their HVAC (or just say they updated their HVAC) but without accountability for achieving a certain level of ventilation or a clear scientific understanding of what different ventilation levels actually mean for the likelihood of spread, it's near impossible to factor that into a risk calculation. ↩︎
What Is COVID Actually Doing to Our Immune Systems? ↩︎
Or, arguably, some people who were already disabled briefly became much less so by virtue of the societal standard of "normal" shifting to the point where their access needs were unmarked and easy to fulfill. But as far as challenging the category myth goes, that's the exact same thing. ↩︎

it's all about the pandemic again

2024-09-10T18:00:00Z

When Cohost goes, with it goes the era of my life I spent on social media. Now that I've tasted the rich fruit of what's possible, I'll never be able to go back wholeheartedly to a site where my timeline is constantly deluged with the latest atrocity, where there's no room for me to write an essay, where I can't even see and share porn. I have loved ones I've followed since before we were even friends whom I won't follow anywhere anymore. It is, inescapably, a paradigm change in how I use the internet.

And in many ways, that's fine. Social media brought me many good things, but even before Cohost I was getting deeply sick of the Twitter model. This is not an intrinsically necessary mode of human interaction, and in a lot of ways it's better not to have it at all than to have it in an unhealthy form.

But. But. There's still a pandemic, deep as world may be in denial, and even with all the mitigations and precautions available it's still an order of magnitude harder to spend time with people in person and another again to enter new spaces and make new friends.

While I've been grieving Cohost, this is something my heart keeps returning to. This was the last great space where I consistently made new human connections. And the way the world is right now, I don't know what can replace that, not just in terms of technology but in terms of life as a whole. The world is so much smaller to me than it was five years ago, and so the loss of a deeply valued space hurts all the more keenly.

Apologia for RSS

2024-09-09T21:00:00Z

RSS^[1] was the original federated social media. RSS invented the "share" verb. RSS is easy, simple, and sustainable. RSS never died.

I'm sure a chunk of you already know what RSS is. Probably some don't. RSS is a protocol (really just a text format) for listing the posts on a website in a machine-readable way. Readers of your website post its RSS feed URL into their feed reader software, this reader periodically checks it for new posts, and it presents all the posts in a single unified list with useful features like "marking stuff as read". (Podcasts still^[2] use the same technology.)

Back in the day, everything had RSS. Any site that had anything that mapped kind of onto "posts", from blogs to webcomics to the nascent social media sites, would expose an RSS feed as well because not to do so was to cut themselves off from what was at the time a massive audience. Even Twitter had RSS for a long time. This was great as a user, because you could easily read stuff from all across the web in a single place. It was great as an author, because your self-hosted blog wasn't intrinsically harder for people to read than anything else anywhere on the internet. It was a kind of federation, but one that didn't require each node to rack up server costs and DevOps time. If you could host a simple XML file that only updated when you made a post, that's all you needed.

If you're reading this, even if you used RSS a bit in its heyday, you may think "oh I can see why that would be useful" but you don't yet understand. All of this is just preamble. What really makes RSS spectacular is the ability to share^[3] posts across feeds. Each post comes with a unique identifier as well as authorship information, which means my feed can just contain a post from my friend that I think my readers would be interested in. It can contain any post from anywhere on the internet! Before Twitter had "retweets" and every website had a "share on social media" button on every page, RSS had post embedding.

Not that most feed readers really took advantage of that, because there wasn't a particularly tight feedback loop between reading and authoring an RSS feed. The one major exception was the late lamented Google Reader, the first high-profile victim of Google's now-infamous penchant for devouring its children. Google Reader provided its own feed for each user, and for a moment there was a vibrant culture^[4] of sharing blog posts with friends and commenting and discussing. I want this back. I want to be able to share anything from anywhere, I want to see the fruits of my friends' excellent taste, and I want to be able to have a conversation about it.

I may not be able to do that yet (although I am hoping this blog can be something like that eventually). But I can and do use RSS, as I have continuously since the mid-2000s. You can too! These days I'm using NewsBlur, but I've also heard good things about Feedbin as well if you want something a little simpler. (I used to use Feedly, but it went all-in on LLM shit so I can't endorse it.) RSS may not quite be Cohost, but it's a way to interact with the web that's not just sustainable, but whose very existence makes it easier to make a web that isn't at the mercy of corporate cartels.

Throughout this post I'm going to use "RSS" to refer to the concept of semi-static machine readable news feeds in general. Atom is definitely the better actual format, but RSS is the more recognizable term. ↩︎
Unless you get them through something like Spotify, in which case they aren't really "podcasts", just sparkling audio shows. ↩︎
I believe this was originally added with the intent that there would be dedicated "feed aggregators", some of which did actually exist. I never found this particularly useful, but I absolutely valued having people I knew personally curate posts that I would later see. ↩︎
I maintain to this day that if Google had expanded this existing small but well-loved proto-social-media product, adding the ability to include locations (via Maps) and videos (via YouTube) and websites (via Search) and expanding the authoring tools for plain posts, it would have been wildly more successful than burning it to the ground and trying to create Google+ from whole cloth. Then again, it's probably for the best that they didn't. ↩︎

House of Nettles: #article

Becoming a Video Game Scientist Part 1: Archipelago

2023: Dipping my Toes

The Old Mod

The New Mod

So-Called "AI" Cannot Program

A Brief Meander into the Philosophy of Language

A Better Definition of Programming

Semantic Refinement

Bringing it Back to So-Called "AI"

A Non-Technical Intro to Webmentions

What can you do with Webmentions?

Notifications

Replies as comments

Reposts and likes

How do I send Webmentions?

How do I receive Webmentions?

Webmentions from other websites

Go out there and get mentioned!

A Sociable Web

Sociable Web not Social Network

"Web" versus "Network"

"Sociable" versus "Social"

Social Pillars

Conversation 💬

...and Technology

Performance 🎭

...and Technology

Meeting People 👋

...and Technology

Avoiding People 🫤

...and Technology

Can we do it?

one last selfie for the road

COVID Denialism and Disability Justice

COVID in the Social Model of Disability

The Category Myth of the Abled

Understanding Denialism

it's all about the pandemic again

Apologia for RSS