• 72 Posts
Joined 4Y ago
Cake day: Apr 19, 2019


I don’t know much about open source licensing. What’s wrong with the SSPL and what does that have to do with mongo?

I'm excited about about a more CMS-like approach to a fediverse server, especially one that leans on #IndieWeb building blocks.

Complaining that you’ve been banned for a seemingly innocuous post isn’t proof that you deserved to be banned. If the author is telling the full story, the admin(s) of their instance were in the wrong.

And this is a legitimate issue. If new fedi users try out an instance and get banned without warning, they’re unlikely to try another instance. I want the fediverse to have a wider array of people and conversations happening, but that can’t happen if admins are quick to ban instead of engaging their users.

Last time I used it, Fedilab also supported PeerTube. The same developer also released a specific PeerTube app, though so I don’t know if they split the video functionality out or left it in.

just because the (corporate) internet works this way now, doesn’t mean it should

The web worked this way before there was a large corporate presence. Scraping was common during the blogosphere period and robots.txt was the solution everyone at the time agreed on and that’s been the standard ever since.

I’m happy that you’re comfortable with this model, but I don’t want people who operate like this to intrude on the spaces we’re building to get away from it

We’re not intruding on this space. We’ve been in the fediverse for just as long or longer; the fediverse has been scrapable since 2008.

I think votes are represented by ActivityPub Likes and Dislikes. Likes are already federated by every other software so it doesn’t seem too much of a stretch to federate dislikes.

But unlisted toots are still technically public. If you scrape my profile, you will get them

Then that’s a scope issue with your server software.

the fact that they are public in the technical sense does not mean I consented to them being scraped etc.

This is what I was trying to say with the analogy to a public announcement. Public speech has no expectation of privacy. Nobody would find anything wrong with recording a public announcement. If you want to have a private conversation, it’s up to you to hold that conversation privately.

Just as wearing a short skirt is not blanket consent to sexual advances

This is a ridiculous analogy. Scraping public text, which is something that’s been widely accepted on the web for two decades, is not remotely similar to sexual assault.

Yes. Business that can afford it have security cameras. And more relevantly, nobody talking in a cafe thinks their conversation is private and that nobody will overhear it. We use a combination of location within the space, voice level, and body language to show how we want others to interact with us. If you walk into the cafe and make an announcement at the front, you have no right to expect that nobody will respond to that announcement, or tell others about it, or even record you while you make that announcement. That is what posting on the fediverse is like. If you want a quiet conversation in the corner, you can post unlisted.

Yeah, I know. I shoulda put a /s on that; my bad.

Disappointing. It’s too bad there’s not a standard API that clients could use to communicate with servers.

EDIT: That was a sarcastic response. ActivityPub has a Client-to-Server (C2S) API that would solve this problem if anybody would ever implement it.

But that helps the student get better at overseeing the AI, not at writing or critical thinking. I don’t even think it would help them get better at analyzing writing. Most students would just turn in the first result, unless the teacher requires them turn in the whole transcript of their session and then the teacher is just grading them on AI overseeing. And that’s one case. Every bit of homework I ever did (before higher education) has been shown by ChatGPT: analyzing literature, writing in various styles, physics problems, foreign language translation, etc.

I doubt that AI will increase the percentage of students that want to cheat, and it is easy to cheat.

What? Cheating right now requires a human somewhere to do the work. A student can steal their work from online or pay someone they know to do it, but basic work will have available answers. A teacher may be able to come up with a unique format for their specific questions in their subject, but that takes extra work for the teacher. If a student can literally just type the question on the paper into ChatGPT and get an answer, I can’t imagine many teenage students not doing that, at least some of the time.

It sounds like you’re suggesting they rewrite their curriculums around ChatGPT (or similar AIs). That would require the teacher themselves to have a good understanding of the AI. So they have to gain proficiency in a brand new technology and then design a way of teaching around that. This is a ridiculous ask of a group of people who are already under-resourced and not keeping up with their current goals (at least in the U.S where I am).

That’s the main point of my previous post. It’s irresponsible and immoral to develop and release a technology for your own profit and just say everybody else needs to adapt to it.

But how? Calculators can help you do arithmetic , but to solve real problems you have to know how to apply the mathematics. A calculator cant solve a problem for you until you break the problem down into discrete operations. You still have to learn how to break that problem into those operations.

If you’re trying to teach students the basic skill of writing, there’s nothing to break down. Write an essay is a pretty atomic operation.

Altman’s response is an incredibly typical response from a silicon valley style technologist. This will be really beneficial one day and the downsides aren’t really bad because you can just adapt to it so my company can continue making money

This seems to be relying heavily on the mastodon markup and prolly won’t work with other fediverse software.

EDIT: After looking more closely I realize all the markup was the author’s own markup, not the mastodon markup. As the author is using the Masto API, this should work with any software that implements that API. I still think this is not a good solution because it prioritizes mastodon over the rest of the fediverse. Comments from some people will be displayed, but not those of people using other software.

You could use deno. It has support for using node modules but is a much more sensible runtime and it adheres to web standards.

I’m not a lemmy contributor so everything i’m saying here is conjecture.

The problem is when any software does a webfinger request to get the Actor (which could be a user or community) they send a request that looks like GET lemmy.ml/.well-known/webfinger?acct=community@example.tldand lemmy’s response may contain multiple actors but mastodon expects only one actor. Lemmy handles this by fetching both actors in the webfinger response and checking their type, which will be User or Group (which is what lemmy calls communities).

Note that the webfinger request doesn’t include the leading @ or ! character because those character are just UI affordances to linkify the community. The !community@domain.tld format isn’t a part of any standard and is never exposed to other software.

Yes, lemmy allows communities and users to have the same name. Other software may not be able to handle this, e.g. mastodon. This causes compatibility issues with mastodon if you search for a user and a community with the same name. I think mastodon will just ignore the community because the name is already reserved on that instance. But if you search for a community that doesn’t overlap with a user, mastodon will resolve it correclty

Copy the post’s link and search for it within pleroma. This is also how I upvote and comment on lemmy posts from pleroma.

Sure I’ll try mocking something up in the browser devtools. On mobile, lemmy displays body text for posts (p.s. I’d like that on desktop too). What I’m talking about is just showing that in the position the title would be, still in an anchor tag so it links to the post but with body text styling instead of in a header.

EDIT: Here’s a quick mockup.

The first and third posts are title-less posts and, to me, look good interspersed with titled posts. To get this, I replaced the entire h5 element that serves as the post title with <a class="preview-lines" href="{{post.link}}">{{made up body text}}</a> The preview-lines CSS class applies a visual truncation so even if the text we render is too much it won’t overflow.

I don’t know how to write Inferno templates but in Vue this would be

<!-- this is the current markup -->
<h5 v-if="post.title">
    <a class="text-body" title="Comments" :href="post.link">
        {{ post.title }}
    <button class="btn btn-link text-monospace text-muted small d-inline-block ml-2" data-tippy-content="Expand here">
        <svg class="icon icon-inline">
            <use xlink:href="#icon-plus-square"></use>
            <div class="sr-only"><title>plus-square</title></div>
<!-- this is new markup i'm suggesting -->
<a v-else class="preview-lines" :href="post.link">
    {{ post.content.substring(0, 250) }}

If they wanna do both, they can. I’m just saying I don’t think it’s the best use of their resources because running a mastodon instance is a huge task. And companies having a Twitter profile was only necessary to get their blog posts noticed. On the fediverse, there’s no reason for that extra step when you could just follow the blog directly. That’s less work for the reader and less work for the poster and no extra service required.

It probably makes it more likely.

I disagree. Running an instance doesn’t give you any insight into how to implement ActivityPub and anybody can study the source code without running an instance. There are quite a few orgs running instances, but, as far as I know, WordPress is the only widely known software that has actually integrated AP and they never ran their own instance.

I’ve never personally received any messages from a pleroma or akkoma instance, that I’ve noticed. My understanding is that they only send Deletes to followers of the deleted acct.