Seems like the solution is: don’t do exams this way

This year, the economist decided that both the midterm and the final exams for his course would be of the take-home, closed-book type (there is a certain tradition of this at Ivy League schools). “It’s a very nice kind of exam, because as you’re giving students practically unlimited time to complete it, it lets you make it harder than normal, to see how far they can go.” In this case, Serrano changed some of the model assumptions they had seen in class, and asked students to demonstrate whether certain statements were true or false under the new assumptions.

I use a browser extension that sends me to a reddit frontend on the off chance I come across a post while searching. It works pretty well

Libredirect + the fastest instance

From the admin side, I do understand what they are saying but I think there’s a better solution to it. We run the old frontend on our instance, and it gets hammered with bots and scrapers. So if we assume that old.reddit works the same way, then it would be attractive to scrapers. But instead of locking it down, I think they could set up something like Anubis or put it behind some other anti-bot measure.

Should have called it the “polluter tax and rebate”

Or even “carbon tax and cash back” system

That’s what I was thinking, especially with the headline

But in this case it looks to be an actual data breach, where the images are the less interesting parts of the leak

https://www.androidauthority.com/iphone-18-pro-color-dynamic-island-leak-3682857/

While previous iPhones have also had many details leaked in the run-up to launch, usually in the form of spec leaks, CAD files, and dummy units, this iPhone 18 Pro leak is the biggest in recent Apple history as these detailed schematics, blueprints, and pre-release test videos are ordinarily not available (even after launch!). We expect more information on the iPhone 18 Pro and other upcoming Apple products to flow through as the leak by cyber group World Leaks reportedly comprises over 200,000 files totalling over 630GB.

You would think that an organization like this would have a better understanding than the average person

https://en.wikipedia.org/wiki/Entertainment_Software_Association

The Entertainment Software Association (ESA) is the trade association of the video game industry in the United States. It was formed in April 1994 as the Interactive Digital Software Association (IDSA)[1] and renamed on July 21, 2003. It is based in Washington, D.C.[2][3] Most of the top publishers in the video game industry (or their American subsidiaries) are members of the ESA.

IMO demonitizing the AI music is the important part. It lets them skip the debate around the AI while removing the incentive for the slop uploads

They didn’t specifically say if they will let users have a toggle to hide it all, that’s one of my concerns with this

Sweet, thanks for bringing it up to them. Looking forward to being able to follow some of the boards 😊

Out of the different mainstream sports, soccer/football is one that’s pretty popular in the global south. There are lots of teams coming from Africa and South America, and it’s pretty popular in South Asia too

FIFA corruption aside anyway

It must be an amazing experience for the kids playing in youth soccer teams here

Could this be solved by having two renderers, and only using the proprietary Adobe one needed?

So what do you do when the pedantic gold standard of epubcheck says your book is fine, when it works without issue on Amazon Kindle, Apple Books, Thorium and everywhere else and yet Kobo says it’s “corrupted”?

I dug into this matter and found out that Kobo uses RMSDK, “Reader Mobile Software Development Kit”, Adobe’s proprietary ebook rendering engine.

Once the stylesheet was identified as the source of my woes, I could finally drill down to find which specific line was causing the issue. After creating a dozen more variations with different subsets of my stylesheet I eventually identified the culprit. It was this line:

.copyright img {
    max-width: min(150px, 30vw);
}

Once I changed it to the more old fashioned max-width: 150px; ADE opened it just fine.

But what is the problem here? The above code is perfectly valid CSS level 4, it’s just not supported by RMSDK, because its CSS parser is frozen in approximately 2013 — no flexbox, no grid, no math functions, no custom properties. Just good old float, bad font handling, and silent crashes when it sees anything it doesn’t recognize.

It’s the year 2026. Thanks to the horrendous RMSDK which Kobo decided to use as their backbone for all book rendering (probably for DRM reasons), a single line of perfectly valid CSS turns a perfectly valid EPUB file into a “corrupted file” on Kobo and just drops the whole book. No clear error message, no fallback. Just a massive fail.

What kind of info?

If you are scraping the project sites and having an LLM put together guides, they will likely go out of date over time and likely contain vital errors already. It would be much better if you simply direct users to where they can read the current up to date information.

How would this be better or more helpful than selfh.st? It is more reputable and more likely to stick around

Some sources are still better than others.

They do?

https://www.cbc.ca/news/fifthestate

On the consumer specific side: https://www.cbc.ca/news/marketplace

Oh I see, that makes sense :)

Maybe these communities then:

• !opensource@lemmy.ml
• !opensource@programming.dev

You could also post about the community in !communitypromo@lemmy.ca

This sounds good to me, thank you for putting in the time for this!

I didn’t know about your project, it looks cool! I think you could share it outside of your community some more, maybe you can crosspost them to !fediverse@lemmy.world?

I feel that some further refinement is needed. I agree with the sentiment behind the latest version of the rule, but I think it still doesn’t address the recent issues.

The way I see it, there is a very specific type of post that has started showing up very recently, and is getting lots of downvotes. Users here are justifiably suspicious of the pattern.

The ones that get downvotes are usually:

• from new accounts
• the user makes one post, and at most they only responds to comments in that one post
• the software uses the help of LLMs, while the post and/or comments are also helped by LLMs
• the software is made to look “professional”, whether it is the UI, the demo, or the README

I’m not sure what exactly the end goal is, but I don’t believe the story that they all use where they “had this problem and now want to share their solution”. I’m concerned that there is some other end goal, whether it is link farming, SEO manipulation, LLM search result manipulation, or it’s the setup portion of a cyber attack where questionable code will be added later (if it isn’t already).

Normally I would suggest to just moderate it based off of “you know it when you see it”, but in this case it’s difficult since it’s very similar to legitimate posts. There are real users that want to post with a new account, such keeping their professional life separate from their main account. It’s also hard to differentiate it based on licenses, because those recent accounts almost always license it as FOSS. I also don’t think it’s fair to exclude all AI assisted code, since it’s very common to have that now.

Perhaps instead of a rule, we could even try some of the following:

• To reduce the risk of OpenClaw style bots creating content here: AI is ok for the code and external text (ex. the README), but the post here should be written by a human. It’s not like the post needs to be that long to express why someone should look at it, and it won’t go through that many edits. Translations should be done through traditional translation software.
• To prevent driveby posts, we could automate a comment on new posts see if a user knows where they are posting. Asking about their favourite threadiverse community, or how long they have been a member here, or even how they learned about the community might separate bots from real users. It works pretty well for our registration applications on lemmy.ca / piefed.ca etc.

On top of being suspicious, I think it boils down to “projects that have a future” and “projects that don’t have a future”. People in this community want to run software that is likely to stay useful and safe over time, and that’s at the core of why these recent ones are downvoted.

In the workplace, there are still a lot of domain specific programs that don’t have Linux support. Companies don’t have much of an incentive to port that stuff over. As for the people who just need a web browser, they probably would use Linux just fine if they could buy a computer at BestBuy that comes with Linux preinstalled

Compare that to LLM programs, where it’s a matter of “download this app instead of that one, because this one is free and that one costs $25 a month”

Having an option vs not having an option

Also Linux and Windows are pretty different in use cases and capabilities. Meanwhile, local AI models have a very similar user experience. If hardware was cheaper and people could run better LLMs locally, they wouldn’t pay monthly for it.