Community on ben's notes

Have some shame, NaNoWriMo.

Thu, 05 Sep 2024 00:00:00 +0000

Over the weekend, National Novel Writing Month kicked a hornet’s nest. They posted a statement claiming that any categorical dismissal or criticism of the use of “AI” tools in writing is classist & ableist, and implies some unrecognized privilege on the part of the critic. That’s a strong stance, and merited a robust response, which the community was happy to provide. Many published authors (Chuck Wendig¹, Daniel José Older², and others) jumped in to point out how the tools on the market are built by destroying copyright, but I haven’t seen an angle that addresses the biases inherent in the LLMs & neural networks that are being used for most of these “AI” products. (I’m drafting this longhand & hate writing quote marks, so just imagine quotes around AI anywhere I write it. The tone is going to need to be inferred as well.)

Who the fuck am I to be talking about this subject, anyways? I’m some writer, what do I know about machine learning and artificial intelligence? For some background on me, I work in software for my day job on network protocols, I’ve dabbled in the underlying technologies of LLMs and their predecessors, and my degree included work in applied mathematics. I’ve got a working knowledge of the underlying techniques and research. LLMs in their current form are effectively Harvard Architectures with no NX bit, and that’s a terrifying thing to have decided to ship at a global scale. While some number of readers will have already decided I do not have sufficient background to have an informed opinion on this topic, let’s assume I’m knowledgeable in the topic and move on.

Technology had historically had a blind spot around bias. More broadly speaking, scientific enquiry has historically had a similar gap in its recognition. The relationship between the observer & observed, and what information an observer includes, inherently relies on the discriminating filter of the scientist!

There’s an entire novel on the topic of identifying what facts are worth selecting, Zen and the Art of Motorcycle Maintenance. Pirsig had a lot to say on Quality, and bias plays a role in defining Quality. To cut a very long discursus short, LLMs are built on data sets that have been filtered for specific stances, and have incorporated some level of bias in their training. This can be seen very literally in an adjacent form of GenAI: when prompting image generating tools like DALL-E or Stable Diffusion with ungendered and racially ambiguous terms, users still got biased output (skin tones and genders for given jobs and social statuses, for instance.) The companies building these tools have done a lot of work to try and suppress this behavior―by making substitutions to users’ prompts to force diversity… which effectively proves the point. They haven’t fixed the bias, they’ve papered over it.

To put it mildly, pervasive bias problems in AI tech have negative downstream impacts on their applications, especially in creative arts. The dearth of material from minority artists, writers, etc. in the training set―much like documented issues in other technical areas―skews their output. Pervasive stereotypes in older works of literature, about any minority or marginalized community, are amplified by the “training” techniques used to tune and build these systems. Like an automatic suspect identification system that can’t distinguish Black men, a system trained on decades of misogynistic, racist books (stolen from their authors, even!) will perpetuate those stereotypes in generated prose and editorial suggestions.

Personally, as a queer writer of queer fiction, I’m concerned what the prudishness of Silicon Valley would do when presented with my writing! What adjectives would it seek to erase, what aspects of my experience would it stifle as a statistical anomaly, under the guise of being an editorial tool?

This is hardly an idle question. Speech to text tools routinely obfuscate swear words, because obviously we need to be shielded from profanity. “Fuck!” gets rendered F*ck! or F***!―neither of which is what I said. Tools like Copilot will often refuse to discuss politics, anything about sex, and sometimes even routine queer social activities or figures! Censorship of minority experiences is hardly new; resources on sex ed and LGBTQ+ experiences is routinely filtered by so-called “porn filters” as “adult” content, even after objections by some of the youth orgs that were hit.

Somehow, we’re expected to believe that tools built by some of the same teams and under the same direction as the people who built biased services, to be fair and correct in their handling of queer issues with GenAI? NO!

Beyond the facile argument in favor of using AI tools, NaNoWriMo’s counter, that my categorical dismissal of the tools is classist, ableist, and indicative of my unrecognized privilege, is downright insulting. This claim twists the language of the oppressed to support a venture capital-funded company’s defense³, who are also funding NaNoWriMo. The claim is clumsy, which is probably its greatest insult: it rejects some of the original goals of the organization and the project, and this leaves it a hollow statement. It wasn’t even made as a blog post, it was put up on their ZenDesk/help pages!

To claim that rejecting AI tools is classist is to ignore the actual cost of using those tools. Paying for a professional license to any of the AI editing tools is a not insignificant cost, for the limited utility of that program―and not a person you can maintain long-term interactions with, building up rapport and shared context! Those tools, too, are built using data taken from users and other sources, generally without or with minimal compensation for their effort. Using them won’t build your writing community, it’ll impoverish it by depriving you and others of connections and directing resources out of your community and into the pockets of the funders of those projects, indirectly enriching their investors to your community’s detriment.

Ableism… I’m not qualified to fully break this one down, but relying on a statistical model is hardly the only way to write, and certainly isn’t the aid these writers necessarily need. This argument is an attempt to claim an entire community as shields for corporate greed, when they don’t want the help.

Lastly, let’s talk about privilege. Again. I’m definitely privileged: my parents are both college educated, I was born in a state which prioritized education & had early exposure to a lot of culture. I first tried my hand at a NaNo in high school, even! I’ve had years of time to develop my craft, and I have a well-paying job that leaves me with leisure time to write. None of that detracts from my criticism, however, and I reject that my privilege is blinding me to the cost or the inaccessibility of writing and editing. Yes, hiring an editor is an expense. I’m still waiting to do that for my first novel, in part due to the cost! But part of the point of NaNo is to learn the tools & techniques of the trade. Yes, editing is a lot like work… isn’t that the point?

Ultimately, the statement from NaNoWriMo feels like the organization is trying to throw their support behind their AI tooling partner, to defend their sponsor from criticism from the community that they had to understand was coming. Unfortunately, the stance they took was so extreme and so directly contrary to the attitude of their community that it was bound to explode. We, artists and writers, are sick of being told how to think in furtherance of commercial interests, and are hardly going to be quiet now. When we’re seeing beloved internet groups like the Internet Archive attacked for improving access to information during an unprecedented geopolitical event and health crisis⁴, this support of commercial interests who are seeking to profit off of broad social goods & communical resources was definitely going to land poorly.

I hope the cash was worth it.

Chuck Wendig posted on his blog: NaNoWriMo Shits the Bed on Artificial Intelligence ↩︎
Daniel José Older posted on Twitter: “Hello @NaNoWriMo this is me DJO officially stepping down from your Writers Board and urging every writer I know to do the same. Never use my name in your promo again in fact never say my name at all and never email me again. Thanks!” ↩︎
In the Washington Post, Chris Banks, the founder of ProWritingAid said: “We fundamentally disagree with the sentiment that criticisim of AI tools is inherently ableist or classist.” I’m sure they didn’t ask for NaNo’s defense, but the whole thing looks bad. ↩︎
Hachette v. Internet Archive is a currently ongoing case (well, the IA lost on appeal by time of writing, but they’ve not run out of appeals yet) in which Hachette is claiming that the IA’s practice of buying a book, digitizing it, keeping the copy in storage, and allowing one person at a time to look at it is a violation of their copyright. ↩︎

The web I want is rough around the edges

Mon, 19 Feb 2024 00:00:00 +0000

A few weeks ago, one of the SDF’s Mastodon instances had its TLS cert expire. My understanding is, the cert had been renewed but the server had to be restarted to pick up the change, and for whatever reason, that just didn’t happen. In the conversation on BBOARD (the SDF’s virtual bulletin board) about the outage, was this post:

TACKER: mcornick (Mark Cornick)
SUBJECT: .. mastodon cert
DATE: 02-Feb-24 19:56:47
HOST: iceland

I really hope that this gets addressed soon, Mastodon is essentially “down” for a lot of people who depend on SDF right now.

As cseiler noted, a new certificate is in place, but the server is still catching up with its backlog.

(my opinions follow, feel free to disregard)

I would advise not “depend(ing) on SDF” for anything. This is not a professional, businesslike service with SLAs or guarantees. It’s a best-effort deal, and SSL certificates are a recurring issue that SDF does not seem to be interested in addressing beyond just telling everyone to be patient. If your use of Mastodon, or any other SDF service, depends on you being able to access it RIGHT NOW when a cert has expired, I hate to say it but you should look elsewhere for that kind of usage. SDF offers a lot of fun stuff, but that’s all it is - fun, not serious.

That resonated with me. These are services run by people, for people, on the timeline that people can keep, not an industry-driven timeline. There’s something amiss when services intended to be for a community, by a community, are placed under industry demands, and I think that how we got here is part of the problem.

I grew up with a web that was a little bit DIY. A little bit stitched together. Somewhere along the way, with the rise of the corporate web, we saw a shift towards five nines availability, the promise of 24/7/365 (and better) uptime of sites, the idea that you should always be able to get the shiny new thing, from a shiny website. The expectations of the commercial world have trickled down into community spaces, and we see it in all the myriad ways people try to replicate large commercial setups for their small, community projects.

I’m coming to realize, I don’t actually want that. I don’t think we need 100% uptime and availability, because that perfection has a real, tangible cost. I see it all around me, and I’ll be honest, it’s just not necessary. We can all deal if a website’s down for a bit, or a service doesn’t update instantly. Capitalism might not like it, but this isn’t about making money. It’s about having sustainable, long term viable communities and services to support them.

Sustainable services

Money talks

There’s a lot of ways to talk about sustainability, so I’m going to start with the most readily quantified: money. An expensive service is inherently less sustainable in the long term, because it necessarily draws more resources from its community to operate. Festivals and feast-days are rare, because they require intense resource allocation that draws from the community’s available funds without replacement in the near term—but they are themselves important. If you operate like every day is a feast-day, however, you’re draining resources constantly.

When looking at operating a high-availability, high-reliability service constantly, there’s a financial burden that grows superlinearly to the rate of availability intended. If you want two instances of a service running in parallel to ensure there’s better service, faster, well… now you’re not only paying to operate two instances of the service, you’re paying for the infrastructure on top of those instances to keep them both maintained, and to enable handoff between them. (I’ll admit, this example is a little weak, as I can easily see a reason to have a cold—or even warm—spare for handoff, but beyond that, you’re headed for commercial ops scale.)

Virtual machines, physical hardware

On top of burning money for a level of reliability we generally don’t need, that overspend means you’re driving many more cores and dedicating more physical resources to the replication; that’s an environmental burden imposed by your desire to be high availability. Keeping the lights on for each running copy of your service means allocating physical machines somewhere to your uptime. Is three times the CPU time worth it?

In the era of blockchains and wasteful power consumption for “proof of work”, it feels like I’m screaming into the void about reducing power usage, but realistically, we can’t just green-energy our way out of the cloud’s power budget. We need to reduce overall CPU usage for the future of our environment, and choosing to run your service just that little bit less reliably may just be the thing to do.

Beyond the monetary and environmental costs, there’s a human cost to very high uptime, which we experience most as burnout and all the social toils of “on-call”. Lots of my friends, if they want to go out on weekends while they’re on call, they’ve got a backpack along. The idea that we should all just have 24/7 jobs once in a while is absurd. By and large, websites are not so fundamentally critical as to require this sort of uptime, and … well, I’m writing this in Notion. Maybe stuff should be designed to work offline, too. (Side note: Notion actually doesn’t work for anything beyond simple note-taking, and this is reminding me why.)

As to how a rougher, somewhat less reliable web benefits its community: it builds in some need for patience, and a bit of concern for each other (and the welfare of the community as a whole.) If you’re petulant because your service is down, that can play out as a negative interaction with your sysadmins, even though you both want the same outcome. That’s no fun for the people doing the maintenance, and the users are mad. If the people using sites understand that the people running those sites are themselves human with their own lives and needs—and perhaps fewer people than you’d imagine run a commercial service (which may itself be fewer than you think)—it’s easier to understand when something isn’t fixed right away.

Simpler websites, lighter footprints

There’s another aspect of the smaller web that I want to touch on. Alongside the unsustainable website cost dialogue, we should also be encouraging site developers to build smaller sites with less tech. Websites simply are too big, too flashy, and most of it’s completely unnecessary. Even my blog routinely has a several kilobyte load size because I want to include a particular font—and I agonize over that. Meanwhile, news sites serve up megabytes of javascript just to … make something bounce as you mouse over it? (And serve ads. Can’t forget the ads, really.)

We live in an era with fantastically capable devices in our pockets. Why does my phone take so long to load some pages, and why are they so hard to read? Web design lost its way, and we’ve never really recovered.

Part of why I’ve built my website to be so minimal is because while I like some styling, I want to recapture some of that TUI look and feel—and I wanted to have a tiny footprint, ideally served by only my host (thank you, SDF). This all results in a site that looks like I want—and which loads in a pretty acceptable timespan, roughly a third of a second on my (admittedly) fast internet. The most embarrassing aspect of it is the fonts. I use Font Awesome Free for the Mastodon and Creative Commons logos, and for the RSS icon; that’s 259kB, more than my entire (un-minified) homepage, and only a little bit more than the Cascadia Code webfont that I include. I’ve considered just abandoning having one specific font and saying “typeset this in monospace!” but… I can have nice things too, right? (Looking at the MetaARPA transfer limits… I might well switch to monospace and minify my Font Awesome setup, honestly. Or see what else I can reduce, that’s pretty heavy.)

Critical services

I’ve been workshopping this commentary for a few days, and there is one argument that I’ve heard a few times: “but this doesn’t apply to critical things!” And that’s true and correct. If lives depend on your service, you should be maintaining that high availability. Medical services, key communications backbones, in some cases financial mechanisms: all of these deserve high uptime. To achieve that, however, those services should be paying enough and have enough staff to achieve reasonable on-call periods. Where teams are globally distributed, follow-the-sun models of on-call can reduce off-hours calls, and where they’re not, either have enough staff to handle shifts, or reduce workload during on-call.

There’s no excuse, in an industry so flush with cash, to have on-call rotations that destroy the people working them. Human effort is one of the most limited resources we have, we should be cautious about needless waste.

Against mediocrity

This is emphatically not a call to accept bad experiences, served in a lackluster manner. I’d argue that the vast majority of the commercial internet is exactly that right now: on a daily basis, I deal with slow-loading sites that have special interstitial graphics just to tell me something is working real hard in the background to load something you want. (The local movie theatre, for instance, took a lot of time to give me showtimes for Dune Part 2 when I was booking tickets yesterday.) Instead, I want excellent, simple, easy experiences that take almost no time to load—even if they’ve got some styling—and that just work. Maybe they’re offline when it’s 03:00 because they have to update a database. Maybe I should be offline at 03:00, too.

Fundamentally, I think the usability and utility of websites can be made better, with far less resource consumption. This isn’t a new gripe or statement, but dammit, I gotta say it myself too. Make the web more human-scale, for everyone.