The Fetaverse is wholly and entirely unprepared for bots. Data scraping is a given. It’s completely open to the internet. There’s nothing stopping it.

Things like Lemmy are ill prepared to handle bots from the mid-2010s. Never mind bots of yesteryear, and never mind bots of today.

Problem is access outside your home for family and friends.

There are serious security gaps that make it a non starter to expose to the internet.

I’ve been using Jellyfin ever since they forked out of Envy, and honestly, it’s the biggest complaint that I have. It is incredibly difficult to make it available to friends and family who are on various devices, networks, so on and so forth.

Whereas Plex “just works.”

Oh yes, the routers and gateways that most people have that are isp provided that may not actually have open VPN or wireguard support.

Those ones?

Also putting a VPN in someone else’s house so that all their Network traffic goes through your gateway is pretty damn extreme.

Nor will the VPN work on things like their TV or Roku or game console. You know the things that people typically sit down and watch media on…

Which doesn’t work for The grand majority of devices that would be used to watch said media.

Tvs game consoles rokus so on so forth typically don’t support VPN clients.

The Jonathan clients for these devices also typically don’t support alternative authentication methods which would allow you to put jellyfin behind a proxy and have the proxy exposed to the internet. Gating all access to jellyfin apis behind a primary authentication layer thus mitigating effectively all security vulnerabilities that are currently open.

It never made sense why you shouldn’t expose jellyfin to the internet?

I mean I think your link tells the story right?

I mean, fundamentally, yeah.

But we live in a corporate controlled, corrupt, world and now of these larger companies can be trusted with this process.

Some smaller communities and platforms DO this right sometimes, as they build in house processu that respect privacy. But governments world wide are making this impossible through increasingly strict compliance requirements that actually increase data privacy risks and funnel these needs to 3rd party services who just lie about what they do with the data.

===========

I’m not kidding when I say this is a REAL BIG PROBLEM.

bot based traffic and astroturfing will supplement and replace human communication on platforms like Lemmy. Driving the narrative and how we engage to the whims of a few rich people. Bots are relatively cheap, and easy to deploy at scale across many platforms.

There will be no open corner of the internet safe from manipulation and forced division. More people will be forced into walled gardens from corps that implement human verification, as they are the only ones with the resources to do something (While also being the source of the problem, see how that works?)

How do you carve out spaces that are protected from that? Well, you need to determine who’s a bot, and who’s and actual person.

But we can’t do that, so the alternative is we are ran over by bots and astroturfing till we’re at each other’s throats like good culture war puppets.

The future is bleak…

It’s largely considered ineffective these days. Detecting elements that don’t affect layout is trivial, or elements that are occluded, transparent…etc

Capchas are one of the best options. But even then, LLM users bypass those relatively easily, and LLM users are one of the biggest risk areas for astroturfing.

Biggest problem for Lemmy and similar applications is scalability and controls and detection for bots.

The compute costs to operate instances are astronomical compared to the actual user load they receive.

That’s a bit of a side problem compared to bots though. Bots are a real problem that services like this are not equipped to handle.

They are already arriving to some degree.

The difference being is that Lemmy and other similar services have zero controls or ability to handle bots or bop traffic if those bots were bots from 2014.

Not bots from today.

It’s a bit of a problem and honestly with increasing bot traffic across the internet and fedaverse being extremely vulnerable to it It’s absolutely bat shit insane, but I don’t see any other option than somehow having some form of human verification.

It’s a problem

I’m guessing I must have missed something here when I made that comment. I visited the link in the body of the OP not once, or twice, but three times to verify I wasn’t losing my mind. Even went into reading the readme, some issues…etc to verify.

I’m now realizing that in my Lemmy client the link in the body is more obvious to click on than the actual article itself.

Did you go to the repo before running your mouth? It’s awesome-selfhosted data.

What AI slop?

Edit:

I’m guessing I must have missed something here when I made that comment. I visited the link in the body of the OP not once, or twice, but three times to verify I wasn’t losing my mind. Even went into reading the readme, some issues…etc to verify.

I’m now realizing that in my Lemmy client the link in the body is more obvious to click on than the actual article itself.

Makes sense. I appreciate your replies

The commit history is 1 day.

Which is incredibly suspicious.

The maintainer you and said that they tirelessly tested, reviewed and verified changes over the course of 3 weeks to make sure that things were running and operating correctly.

This is how it should be done. It’s not like they’re vibe coding this.

Pretty much.

I’ve started using AI on a project last week and the first thing I do is write tests. Lots of tests.

With enough guardrails, you could actually get pretty decent quality output out of it and with enough regression tests, you can ensure that nothing’s actually breaking.

Similarly, reviewing its changes and actually reading the code that’s being generated to ensure correctness is necessary. However, I am finding ways to automate that and reduce the incident rate of problems to even lower than my co-workers.

Dude, my team members put out code that’s like this or worse on a regular basis that gets caught in PR review without using AI tooling…

I’ve supported legacy projects that of course were built without tooling that didn’t exist. That are structured and written in ways that are far far worse than this.

Nothing here screams vibe coated.

Completely agree. If this is a skilled Dev who’s built products like this before and you can build something like this in your afternoons and weekends in like 6 months without LLM tool assistance.

With basic assistance you can definitely cut that time down to 4 months or less easily.

And if this is a full-time project, you can probably get it out the door in 1 to 2 months with llm assistance. (Not vibe coding, two very different things)

I’ve built projects of the size in 5-7 months before we had LLM or ML coding tools.

With tabbed completion (Which most devs enjoy), and before full LLM code gen, 4-6 months.

With llm assistance, not vibe coding, it’s possible to build projects like this in 1 to 2 months without sacrificing quality or safety. If you are an experienced engineer and have built projects like this before. A lot of these are boring, boilerplate, stuff.

So the time spent doesn’t necessarily say that it’s vibe coded but if this is an inexperienced engineer then it very well might be and may be full of holes and issues.

Yeah, it should balloon out to 15TB or more I think