⁂ L. Rhodes

https://merveilles.town/@lrhodes

Tinkerer. Fix your hearts or die.

pronouns: he/him/they
site: https://lrhodes.net/
tech: https://destructured.net/

Public Key

npub1e5z8avwreewgmjhugmusjpddnyrgjgnarahvfawh9m566pezkyysu2p90v

NIP-05 Address

[email protected]

Profile Code

nprofile1qqsv6pr7k8puuhydet7yd7gfqkkejp5fyf737mky7htja6ddqu3tzzgprpmhxue69uhhyetvv9ujumt0d4hhxarj9ecxjmnt0qx4z2

Publishing to

relay.momostr.pink

Author Public Key

npub1e5z8avwreewgmjhugmusjpddnyrgjgnarahvfawh9m566pezkyysu2p90v

Show more details

Last Notes

2026-05-19 18:11:06 UTC

- reply

"These agents can work in the background 24/7, to track changes on the web and alert you to new information."
Finally, an answer to the question: How can we make Google Alerts less reliable and more environmentally destructive?

2026-04-22 03:12:42 UTC

- reply

If 313 is a pro-Iran group, as reported, then the .social attack doesn't really make sense to me. Bluesky is US-based, so that at least has a tangential association to the war, but .social is EU-based, so like… why? And neither target would make as much sense as Truth Social.

2026-04-17 16:36:42 UTC

- reply

Apparently, the Bluesky outage was due to a DDoS attack: https://www.engadget.com/social-media/bluesky-blames-ddos-attack-for-server-outages-150515882.html
How do you DDoS a decentralized network? Tom's Guide says it was API servers that were primarily affected: https://www.tomsguide.com/news/live/bluesky-down-live-outage-updates-4-16-26
That's a chokepoint I haven't seen mentioned before. I assume it's part of the Bluesky appView, though I still haven't seen anyone confirm whether other appViews were affected by the outage.

2026-04-16 17:20:43 UTC

- reply

Is this what the kids mean by "infrastructure inversion"?

2026-04-16 17:13:17 UTC

- reply

"Bluesky operates an official status page which is built to give updates on outages, but that also appeared to be affected by the problems." 🤨 https://www.the-independent.com/tech/bluesky-down-not-working-twitter-x-threads-b2958863.html

2026-04-16 14:17:43 UTC

What exactly does everyone mean when they say Bluesky is down?
Presumably it's not the PDS since it seems to be affecting other PDS too. Is it the appview? (In which case, other appviews, like Whitewind, should still be working.) Is it the relay? (Which would stop just about everything.)
Do outages like this take out the other *sky networks as well?

2026-04-07 13:12:47 UTC

- reply

Quarantine periods for new accounts, especially on open registration servers. The details could be configurable, but basically, new accounts would be unfederated for their first month. They could follow any accounts already known to their host server, but could only boost, fav and reply to posts from accounts on the server they're on. Admins could waive the quarantine on an account-by-account basis.
Three benefits:
1. It raises the investment level for creating troll accounts, which serves as a deterence.
2. It allows mods to observe new account behavior before unleashing them on other servers, which helps contain reputational blowback.
3. It narrows new arrivals' focus early on so that they're encouraged to familiarize themselves with the local community first.

2026-04-06 18:37:44 UTC

- reply

We — the fediverse — have, in fact, handled inclusion poorly at times in the past, and we need to do better in the future. But the way to approach it is to be inclusive of peoples, and let diverse viewpoints crop up (or not) as a result of that.

2026-04-06 18:23:37 UTC

Part of where today's main character post went wrong is that there was a trend a few years back in which tech bros tried to pass "viewpoint diversity" off as a DEI issue so that they could justify bringing more conservative/libertarian voices to the table, and while that's not where he was going with that post, that's the frequency it vibrated.

2026-04-02 18:34:05 UTC

- reply

I'd like to see something like the approach that @npub1p09…m5vl and team have taken here extended to how the fediverse practices mutual aid more generally. Maybe in a less centralized form.* I've been thinking about human.json https://codeberg.org/robida/human.json as a model: something that would allow a handful of people to independently vouch for a crowdfunding campaign, so you can decide to get involved based on prior relationships of trust.
* not that gaza-verified should be decentralized; there are good, structural reasons for how they're operating

2026-04-02 18:27:40 UTC

Was surprised to see that I had 176+ pings on Street Pass before I realized it was because I had https://gaza-verified.org open in a tab.

2026-04-02 18:10:47 UTC

- reply

Every time someone says we should adopt a feature or approach from some other platform, network or protocol, I want to shout, "WAIT!" Because our differences from those other platforms is central to our value as a network, we should always go slow and make sure that whatever we'd gain from following their lead is worth that loss of value. We have to challenge their logic at every step, because once implemented, it will shape the fediverse to the patterns they've set for themselves.

2026-04-02 18:04:42 UTC

- reply

Even though quote posts ultimately won out, the period where Mastodon refused the rush to implement them was valuable as an approach to design. The debate over quote posts was valuable because it occurred in a lacuna that was genuinely participatory. And if the form that quote posts have taken on the fediverse is different and (hopefully) better than their counterparts on other networks, it's due in no small part to a determination to approach design as the domain of community, not profit.

2026-04-02 18:00:44 UTC

"But if the crisis is the product of specific design choices made by specific companies for specific economic reasons, then those choices can be challenged, regulated, reversed." https://aeon.co/essays/what-we-think-is-a-decline-in-literacy-is-a-design-problem via @npub17tt…mw52
That's one of the most valuable things about the fediverse. Even if it never attracts as large a population as other social media networks, it's absolutely critical as a space to challenge and (in some cases) reverse the design decisions that make them so hostile.

2026-04-01 13:55:00 UTC

- reply

It's one of the contradictions of modern publishing that poetry has been relegated to niche, even though there seems to be reasonably strong demand for "poetic" novels. I'd argue that the demand for that style is an indication that readers want poetry — they want the sort of highly compressed language, non-linear thought, and focus on qualia that poetry is much better suited to deliver. But, for some reason, the axiom remains that poetry doesn't sell.

2026-04-01 13:48:04 UTC

- reply

Novels with prose that strives to be poetic are not all that rare. It's an extremely fashionable mode in contemporary literary fiction. Novels that successfully integrate the poetic mode into the novel form are much, much rarer.

2026-04-01 13:45:13 UTC

- reply

I was basically uninvited from a book club for insisting that I didn't really care about beautiful writing at the sentence level.

2026-04-01 13:40:30 UTC

I have very little patience for most "poetic" novels.

2026-03-29 13:18:27 UTC

The popular understanding of enshittification gets things precisely backwards. It's not that digital platform startups set out to make valuable services for consumers, but end up making changes that tilt the balance toward exploitation. Rather, exploitation was the destination from the very start, and user satisfaction was a means of scaling up the pool of exploitable users. Enshittification is planned, not inevitable.

2026-03-29 02:45:25 UTC

Sorta disappointing how often the narrative arc of rock biopics ends up being, "I wanted to be a rock star, so I started acting like one until it made me rich."

2026-03-23 00:03:53 UTC

Seems like a lot of people were taken off-guard by the news that Bluesky had been sitting for nearly a year on a second round of unannounced funding worth nearly $100 million, but what really ought to give everyone pause is that both their Series A & B funding were led by cryptocurrency investment firms.

2026-03-22 19:33:34 UTC

- reply

Happy birthday, Devine! :moomin_hattifatteners:

2026-03-18 17:55:22 UTC

- reply

The whole thing just looks so sharp.

2026-03-10 20:51:04 UTC

- reply

Even without the midrash (if that's what the tradition is), I think there's a question of motive when it comes to the snake. Christian exegesis has tried to resolve that by associating the serpent with Satan, but all the text actually says is that the snake was the most cunning of animals; it doesn't say WHY they deploy that cunning against Woman, but jealousy fits the narrative. Tempting Woman into disobeying God would be a neat form of revenge. And if closeness to Man is what the snake wanted, then God's curse against the snake makes a punning kind of sense in that it forces the snake into closeness with Adamah.

2026-03-10 19:59:51 UTC

- reply

That would make the snakes motivation pretty interesting.

2026-03-10 19:51:07 UTC

- reply

Another fun bit: In the first narrative, God creates everything, including fish, birds and cattle, and only then creates humans. But in the second narrative, Adam is the first living thing, and God only creates the animals in a process of trial and error to give Adam a companion. So you're left to imagine God presenting Man (he doesn't have a name at this point) with each animal, and Man saying, "Let's call that one a penguin, but no, that's not really my style," until he God gets around to creating Woman.

2026-03-10 19:37:27 UTC

- reply

Incidentally, David Rosenberg and Harold Bloom made the case that much of the Penteteuch was written by a woman in Solomon's court, and that she gave traditional Levantine stories a witty gloss as a satirical treatment of women's status in monarchial Israel: https://groveatlantic.com/book/the-book-of-j/
There's certainly a lot of wordplay in Genesis, starting with Adam (alluding to adamah, the earth) and Eve (or Ḥawwāh, which sounds like ḥyw, living).

2026-03-10 19:19:14 UTC

- reply

The fig leaves are funny, btw. Adam and Even realize they're naked and immediately invent sewing. But killing won't be invented for a while yet, so they sew their loincloths out of leaves!
Literalists tend to be humorless about this sort of thing, but I suspect it was intended to be funny, not least of all because, having cursed pretty much everything, God then turns around and clothes them in "garments of skin." Like: "Here, you two look ridiculous in those leafpants."
And then, talking to an unnamed audience, God casually drops that there was also a previously unmentioned tree that would have made them immortal! You can practically see them camera pan to the other tree on that punchline.

2026-03-10 19:09:54 UTC

- reply

I really want to insist on the disagreement between the two creation stories, not as some skeptical gotcha, but because I think the disagreement itself is interesting. In the first, "God created humankind in his own image, in the image of God he created them; male and female he created them." And almost immediately, the second narrative steps in and says, in effect, "No no no! They weren't created at the same time, and they weren't godlike until later on, and those intervals make a big difference."

2026-03-10 18:41:33 UTC

- reply

I'm pretty familiar with Genesis, so I had thought about skipping it, but I'm glad I didn't. Certain elements stand out to me now, like the fact that the book contains two creation narratives that contradict one another, e.g. on the order in which classes of things are created. It's easy to see that one set of authors waa much more invested in patriarchy than the other.

2026-03-10 18:35:41 UTC

It never really occured to me before that Eden is not the name of the garden. Rather, the garden is explicitly introduced as being planted in Eden, which establishes Eden as a distinct place. Further, the garden is watered by a river that "flows out of Eden," and branches into the Pishon, the Gihon, the Tigris and the Euphrates, which suggests that the authors may have had a specific region in mind when they spoke of Eden.

2026-03-02 22:25:35 UTC

All of the most interesting stuff happening in computing right now is reparative.
Not "where can this take us next?" but "how can we make this humane?"

2026-03-01 23:47:38 UTC

Boosts are how you say "Check this out!" Favs are how you say "Life is overwhelming, and I am too tired to formulate a reply."

2026-02-23 23:17:45 UTC

I couldn't care less about IBM's stock value, but the idea that businesses and governments might trust AI-generated mappings of financial and aerospace databases and workflows gives me the heebie jeebies. https://www.cnbc.com/2026/02/23/ibm-is-the-latest-ai-casualty-shares-are-tanking-on-anthropic-cobol-threat.html

2026-02-23 00:56:02 UTC

- reply

Do people sometimes make needlessly sweeping, part-for-whole generalizations on the basis of one bad actors' behavior? Sure. But think about the role that the word "culture" plays in the argument against "purity culture," lumping all ethical critique into a unified "culture" that you can then reject tout court whenever it suits your prerogatives.

2026-02-23 00:35:49 UTC

- reply

The phrase "purity culture" is your cue to interrogate the speaker's position even more closely, because 99 times out of 100 it's just a thought-terminating cliché for shutting down ethical considerations.

2026-02-22 23:57:29 UTC

- reply

I guess I'm still mad.

2026-02-22 23:56:42 UTC

"Oh, you want me to avoid technology created by immoral people?? I guess you have to give up THE INTERNET hurr hurr!!"
Nobody cares that you use a technology invented by bigots who died half a century ago, my brother in TELNET. Just maybe don't lend rhetorical support to the technology that's funding and empowering the people who are currently commiting atrocities.

2026-02-20 22:32:22 UTC

- reply

That premise — that scraping for any purpose is ethical — also seems pretty out of step with changing perceptions of how the internet ought to work, as someone who has been on the fediverse since 2018 and presumably seen the controversies over privacy, search and opt-in v. opt-out should probably know. People are questioning the status quo when it comes to consent and consensus on the internet, and it would behoove AI enthusiasts to think about why.

2026-02-20 22:28:34 UTC

Among the many things Doctorow gets wrong in That Post is this:
"It's not 'unethical' to scrape the web in order to create and analyze data-sets. That's just 'a search engine.'"
Apart from the fact that AI companies are particularly malicious in the way they scrape the web, I'd say we accept search engine scraping mostly on the premise that it's done for the benefit of the scraped sites. There's no such principle of mutual benefit in AI scraping — the AI company gets the value of the data scraped and you get bupkis at best, and possibly DDoS'd

2026-02-20 19:20:40 UTC

People act like picking a server is the real hurdle for Mastodon onboarding, when really, it's teaching people how to set up a filter for all of the programming languages they don't use.

2026-02-19 18:55:32 UTC

Mastodon GmbH has launched a help center for Mastodon users: https://help.joinmastodon.org

2026-02-17 17:27:25 UTC

- reply

Seems to me that development on AT sped up *at the expense of* development on AP.
That is to say, a lot of the people who've gone all-in on AT were building here before, but have pretty much abandoned AP development. And it's worth examining why AT development was so much more appealing for those folks.

2026-02-17 14:49:00 UTC

Gen AI is an expensive way to create text and images. Right now, much of that cost is being subsidized and hidden, but it will eventually need to be paid. Because charging the actual cost will likely negate most of the value proposition of genAI, the companies that run the big models are focused on altering the social and economic context so that there are big external costs to opting out. E.g. getting companies to fire staff makes it difficult to pivot back away from AI because hiring and training replacement staff can be difficult and costly. Hence, the huge rush and hysterical sense of urgency around adoption: the demand for profitability is an approaching tidal wave, and they need to lock entire industries in before that wave hits.
That's why resistance IN THE PRESENT counts for a great deal. Right now, we have the approaching wave in our favor, and they're counting on cultivating enough dependence before it hits that we'll have no choice to accept the actual costs. The closer they get to locking society into dependence on hyperscale AI systems, the more difficult it becomes to opt out of even the plainly dystopian uses of the technology. And the longer we "wait and see," the less say we may ultimately have in how this technology shapes our society.

2026-02-15 05:44:47 UTC

Everyday, there are two good things on the internet. Your job is to find them. This job pays 0.083 good things/hour.

2026-02-13 01:27:21 UTC

"None of these alternatives are as polished as the big corporate platform that's doing shady things!"
Yeah, buddy. You'll never get free if you cling to polish. Embrace the scruffynet.

2026-01-28 13:15:19 UTC

- reply

Vibe coding is flooding the world with bespoke programs riddled with security flaws, memory leaks, and sleeper bugs that no one will catch until the damage is already done.

2026-01-28 13:09:54 UTC

"No one has ever made an app that does the specific thing I want, and I'm not a programmer, but maybe I can get AI to make it…"
My deepest condolences, friend, but if you couldn't code an app yourself, then you are the last person who should be overseeing an AI to generate one.
"But I just vibe coded a version and it seems to work all right!"
That's the most dangerous type of software: the kind that only seems to work correctly.

2026-01-28 12:42:50 UTC

The hard problem of decentralized social media is getting new people to appreciate what the fediverse is rather than dwell on what it isn't.

2026-01-24 14:17:49 UTC

One thing I didn't anticipate was that some of my leftiest friends would end up being the ones who embraced AI the least critically.

2026-01-18 15:18:33 UTC

- reply

Programs like "Buy For Me" reinforce Amazon's reputation as the ecommerce site where you can find literally everything you might buy, lowering people's tolerance for the need to search anywhere else. Ideally, the reputational damage from errors and complications would counteract that, though they tend to be rather teflon-coated in that regard. Creating friction is one way to fight back.

2026-01-18 15:01:22 UTC

- reply

If you sell anything on the internet, I'd recommend making a quick search of Amazon to make sure they're not selling your products through their "Buy For Me" program. You can opt out, and probably should, because this sort of thing is bound to cause errors and cost small retailers money, sales and trust.

2026-01-18 14:56:49 UTC

#Amazon has started generating product listings for companies that don't sell on Amazon, without their permission or knowledge: https://www.modernretail.co/technology/brands-are-upset-that-buy-for-me-is-featuring-their-products-on-amazon-without-permission/ "Buy For Me" links activate an #AI agent that attempts to buy the product through the retailer's actual storefront. The listings are AI generated and error-prone. Some allow customers to place orders for items that are no longer available. Companies are having to assure their suppliers that they haven't violated contractual agreements against selling on Amazon.

2026-01-16 14:34:10 UTC

Some of the most frequent questions I've seen coming out of the #forkiverse are about posting just to theforkiverse.com vs. to the fediverse at large. Which tells me:
1. They might have been better off using Hometown.
2. Local-only posting could actually help Mastodon attract communities.

2026-01-13 16:42:18 UTC

- reply

It is. They explain that in the episode.

2026-01-13 16:19:23 UTC

- reply

I find it particularly funny that they describe Kevin Roose in the episode as someone who "learns by doing," and then he turns around and has Operator set up the server. The obvious conclusion is that he was more interested in learning about agentic AI than fediverse administration.

2026-01-13 02:29:30 UTC

- reply

I don't want to blow the potential issues out of proportion. All new instances face unexpected challenges, particularly if their admins are also finding their way around the fediverse. Though I really do think the longevity question is the most critical from the perspective of the health of the fediverse. The forkiverse has brought 2k new people onto the network, and it would be a shame to lose them just because the operators or their employer decided that the experiment was no longer worth funding.

2026-01-13 01:50:21 UTC

- reply

I'm also curious as to how much they're involving OpenAI Operator in the moderation and administration of the site. Apparently, they relied on it so heavily in setting up the server that the account registration and approval process came as a total surprise to everyone involved.
Maybe more critically, what's their federation policy? Do they have a clear idea of how bad it can get out there if you're not diligent about blocking malicious servers?

2026-01-13 01:40:44 UTC

I'm listening to the Search Engine episode that birthed the #forkiverse, and I have questions. I'm morbidly curious about the server's configuration, since they handed that over to an AI that apparently neglected to configure them for SSL. But mostly I wonder how long their funding will last, since they're expensing it to their publisher.
Maybe that all gets cleared up later in the episode.

2026-01-11 22:22:21 UTC

- reply

Just gotta find the right one.
https://assets.merveilles.town/media_attachments/files/115/878/792/280/594/439/original/7d61261f22815000.jpg

2025-12-11 14:31:50 UTC

- reply

If we want software that is both personalized AND conducive to personal autonomy, the way to do that is by giving people clear, explicable options, and leaving it to them to decide what works best for them. Handing personalization over to AI shifts another degree of autonomy away from the person using the software.

2025-12-11 14:23:54 UTC

- reply

There's an affinity between the notion of software that adaptively personalizes itself to the user and the idea that a freemium model gives people creative freedom. Both present themselves as empowering the person while in fact tying them to corporate imperatives that are opaque and subject to change without warning. The tech industry strategy for selling submission is always to package it as individual empowerment.

2025-12-11 14:20:07 UTC

- reply

The algorithmic timeline is a good model for understanding how self-adapting software could (and likely will) subvert personalization to profit-seeking. Social media behemoths pitched algorithmic sorting as a way of tailoring timelines to individuals, but the actual effect was to give the company more control over what you saw and more influence over how you engaged with their app. Is there any reason to believe that "resonant" software wouldn't be put to the same use?

2025-12-11 13:35:39 UTC

"Where once our digital environments inevitably shaped us against our will, we can now build technology that adaptively shapes itself in service of our individual and collective aspirations."
Yeah, that's not how it's going to shake out. Software that adaptively shapes itself according to who's using it will almost certainly by used to target their particularities, making it a more effective tool for shaping them according to the designer's or provider's agenda. The whole history of digital personalization points toward that likelihood.

2025-12-08 02:26:07 UTC

There's a whole type of guy (mostly guys, anyway) who seem to think that books are overstuffed vessels for one or two deep thoughts, and that you can extract those thoughts and discard the useless husk of the book that surrounds them, and I can't figure out if they think that the authors are just too dense to convey their deep thoughts without all of those superfluous words, or maybe the book part is a hustle that they've seen through, or something else.

2025-12-07 16:04:49 UTC

- reply

Over time, the public library system has developed a rigorous set of ethics and practices concerning patron privacy and confidentiality, because librarians have seen first hand what can be done with information about what a person reads and what information they seek out, and here's the digital reading community plugging their personal libraries directly into surveillance dragnet services.

2025-12-07 15:53:48 UTC

idk, maybe don't flip the switch that gives access to your ebook reading to companies that train on books without permission and provide surveillance services to authoritarian governments. https://calibre-ebook.com/whats-new

2025-12-07 04:40:50 UTC

If you think leftists are afraid of capitalism, you should see how hard some capitalists work to avoid competing in a free market.

2025-12-05 20:47:42 UTC

- reply

The thing to understand about Garfield is that he is a cat, and thus is not subject to the strictures of the 9-5 work week, so his hatred of Mondays represents not the reality of capitalist labor, but its internalization as an ideology, even in the absence of the external pressures that condition human labor. Likewise, cats lack the physiognomy that would make regularly digesting lasagna tolerable; it is consumption as a mode, rather than lasagna as a substance, that compels him.

2025-12-05 18:41:48 UTC

I am to Decembers as Garfield is to Mondays.

2025-11-03 18:06:03 UTC

The cost of generating an ad personalized for one person is so much higher than the cost of making ads for thousands of people and just targeting delivery that I sincerely doubt any profit-seeking company will bother with this for more than a trial period. https://www.404media.co/the-future-of-advertising-is-ai-generated-ads-that-are-directly-personalized-to-you/

2025-10-31 14:53:53 UTC

Immortan Joe needed that water for the data center he was running.

2025-10-31 00:56:00 UTC

- reply

I missed it before, but the new #Affinity app includes an AI chatbot called Ritson.
There's also the last question on this FAQ: https://www.affinity.studio/help/canva-ai-canva-ai-affinity-ml/
"Can I disable AI in Affinity?
"No, you can't disable or remove AI features in Affinity, but it's your choice if you use them."
So, yeah… thanks, but no thanks.

2025-10-31 00:25:34 UTC

- reply

I wonder how far the Affinity team had gotten into developing v3 when Canva said, "Looks great, we'll make it freemium."

2025-10-31 00:03:17 UTC

- reply

Canva took a good, hard look at the best alternative to Adobe Creative Suite and said, "I bet we could sell a lot of Dave & Busters cards with that."

2025-10-30 23:58:02 UTC

- reply

All else aside, it's a bit demoralizing that a professional-grade suite like Affinity has been reduced to subscription-bait for AI tools.

2025-10-30 22:59:17 UTC

- reply

I'm half considering blocking the new #Affinity app with a firewall so that I can use it without sending data to Canva, but at the same time, why would I trust it after seeing it send data even after I opted out of analytics?
I'll keep an eye out for some reasonable explanation, but I don't expect one.
Affinity has said that the v2 apps won't get anymore updates, so maybe I'll block incoming traffic to those apps as a precaution against lockout, and just stick with v2. They do the job.

2025-10-30 21:19:03 UTC

- reply

Just sitting here with ActivityMonitor open, watching the new #Affinity app send packets out onto the network as I draw…

2025-10-30 18:53:56 UTC

- reply

If you signed up for Canva to download the new #Affinity app, go to canva.com, log into your account, navigate to Personal Account > Personal Privacy, and switch everything to off. Some of their AI permissions are tucked away there, rather than on the app.
You'll probably want to change some of your Message Preferences as well. And there's an AI Personalization panel.

2025-10-30 17:49:38 UTC

- reply

On the face of it, this is all much better news than I was expecting. My only two concerns at the moment are:
1. Is the separation between Canva/AI and Affinity as sharp as they're suggesting?
2. Is this financially sustainable?
re: the second one, it seems like Canva is banking hard on AI. If (ahem, when) the bubble breaks, what happens to Canva, and by extension, to Affinity?
So I'll probably try out the new Affinity, but I think I'll also hold onto the v2 apps, just in case.

2025-10-30 17:44:57 UTC

- reply

Announcement is up: https://youtu.be/UP_TBaKODlw
I'm still working my way through the details, but the tl;dr is:
• New #Affinity app with Publisher, Design and Photo all integrated into one app, all for free.
The "free" part is where I start to worry. The announcement positions it as freemium, with a paid tier for integration with Canva, but specifies that's really only if you want AI (and, presumably, cloud) features. Hopefully, they hold to that division in practice as well as marketing.

2025-10-26 15:10:23 UTC

- reply

The positioning of LLM-based AI as a universal knowledge machine implies some pretty dubious epistemic premises, e.g. that the components of new knowledge are already encoded in language, and that the essential method for uncovering that knowledge is statistical.
Maybe no one in the field would explicitly claim those premises, but they're built into how the technology is being pitched to consumers.

2025-10-26 14:50:22 UTC

- reply

If you want to know how the word "impactful" spread through written English over time, LLMs can help. It can help with that because it's a specifically linguistic question, but it's a pretty big leap from there to other types of knowledge. And LLM-based AI is being pitched to businesses and the general public not as a tool for conducting linguistic research, but as a universal knowledge machine. And that's where a great many of the technology's problems begin.

2025-10-26 14:29:31 UTC

- reply

In terms of cost, consider CERN's Large Hadron Collider, which had a budget of €7.5 billion (~$8.7bil), and was built to explore some of physics' most fundamental questions. OpenAI's Startgate project alone is expected to cost $500 billion, and it's only one of dozens, if not hundreds, of megawatt AI projects planned. Of course, not all of that is devoted solely to LLM processing, but still, the investment that we, as a society, are pouring into LLMs is a geometric increase on what we put into what was previously the largest and most expensive research tool ever made, mostly to misuse them.

2025-10-26 14:19:52 UTC

It helps to understand that LLMs are basically machine learning programs for finding statistical relationships in language use. It's an advanced form of the sort of ML that digital humanities researchers have used to, say, chart the incidence of sexist language in English literature over time. So there are logical uses for LLMs, but a) they're mostly limited to the study of language, and b) they don't necessarily justify the costs piling up around commercial startups like OpenAI and Anthropic.

2025-10-23 19:49:33 UTC

- reply

One week to go until the big Affinity announcement (derogatory).

2025-10-21 15:06:03 UTC

A DNS error at Amazon can hobble the internet because much of the internet is centralized on AWS servers.
Much of the internet is centralized on AWS servers because Amazon priced their hosting for aggressive competition, and companies couldn't pass up a deal.
So yesterday's big outage was also an object lesson about what's lost when you trust everything to the invisible hand of the free market.

2025-10-18 14:37:03 UTC

All data on the #Bluesky network passes through a server operated by the Bluesky PBC, so it's probably worth paying attention to the sort of relationships they form with the government.

2025-10-07 14:26:01 UTC

If you're just (re)joining us, the Big Rule on the fediverse is:
Help others be here, too.
The penalty for breaking the rule is:
Try harder next time.

2025-10-07 00:13:14 UTC

Just so it's lost on no one, the implication of the Bluesky Waffle House joke is that you can accomodate both trans people and activist transphobes in the same space, the same way that serving waffles doesn't imply that you hate pancakes. It's a false analogy that excuses bad policy, because activist transphobes have made it their business to drive trans folks out of every space they inhabit. The more you accomodate the transphobes, the less accommodating your space (not to mention society at large) will be to trans folks. And yelling WAFFLES at the people who try to point that out is just emphasizing that you're siding with the 'phobes.

2025-10-06 17:43:56 UTC

- reply

AFAIK, PDS don't publish posts to anything — the Relay scrapes every PDS in order to push posts to the firehose. So even if the PDS ignored DID blocks, I'm not entirely sure it would pass messages between accounts on the same PDS without an overhaul that breaks AT compliance.
Running your own Relay and Appview is technically the solution, but in order to connect to accounts on the rest of the network, they also have to subscribe to your firehose or a third party service that draws from it, which diminishes the likelihood that you'll be available to most people using the Bluesky relay.

2025-10-06 17:09:54 UTC

Someone said that some Blacksky users are staking out fediverse accounts in the wake of Graber's heel turn. Is that just hype? My understanding was that they had achieved enough independence that Bluesky's moderation didn't affect them as much.

2025-10-04 20:00:41 UTC

- reply

Is that what they meant by "creative freedom"?

2025-10-04 19:59:00 UTC

- reply

Apparently, the new Affinity Discord is just totally unmoderated on the weekends, which has not been lost on at least one racist/homophobic troll.

2025-10-02 20:48:06 UTC

Are there any examples of people who have been banned from Bluesky and moved to a different service on ATproto?

2025-10-02 01:26:55 UTC

- reply

I switched to Affinity because it was the Brother printers of publishing and design suites. If it ceases to be that, then I'll jump ship in a heartbeat.

2025-10-02 01:17:43 UTC

- reply

Needless to say, I won't be paying for any version of Affinity that makes generative AI integration compulsory.
If they sneak it into a v2 update, I'll just delete the whole suite.
And I definitely won't be switching to a subscription.

2025-10-02 01:11:41 UTC

- reply

One line of speculation that seems pretty convincing is that the "creative freedom" branding is meant to tease the release of a free tier with limited functionality, with seamless upgrades to subscription tiers for additional functionality. That's Canva's model, so it makes sense that they'd expand it to Affinity. Given that they recently acquired two AI companies, though, it's also a pretty good bet that integrating AI will be part of the "value" proposition.

2025-10-02 01:02:03 UTC

- reply

Shutting down the Affinity store with no prior warning has reportedly left free trial users in the lurch, some for as long as a month. If they can't continue their trial, and can't upgrade to the full version, many of them will look for an alternative, some permanently.
There had to have been a better way to handle the transition to whatever they're launching on the 30th.

2025-10-02 00:58:53 UTC

- reply

I wonder what the mood has been in the Serif/Canva offices today.
Seems like they expected this pre-announcement to be greeted with excitement, and that is NOT the vibe.