Hacker News new | ask | show | jobs
by cwal37 125 days ago
LinkedIn is also a great example of this stuff at the moment. Every day I see posts where someone clearly took a slide or a diagram from somewhere, then had ChatGPT "make it better" and write text for them to post along with it. Words get mangled, charts no longer make sense, but these people clearly aren't reading anything they're posting.

It's not like LinkedIn was great before, but the business-influencer incentives there seem to have really juiced nonsense content that all feels gratingly similar. Probably doesn't help that I work in energy which in this moment has attracted a tremendous number of hangers-on looking for a hit from the data center money funnel.

8 comments

Yeah I've been collecting some of the weirdest ones I've seen floating by. It's really the only thing that has me visiting linkedin.

https://www.marginalia.nu/junk/linked/games.jpeg

https://www.marginalia.nu/junk/linked/json.png

https://www.marginalia.nu/junk/linked/syntax.png

(and before anyone tells me to charge my phone, I have one of those construction worker phones with 2 weeks battery. 14% is like good for a couple of days)

The red apple streams one is good. It shows how developers chase shiny new stuff with no respect for fundamentals. They will say it's less code, and then show you more code.
The apples one is LLM nonsense: the left example doesn’t include any code for the loop, whereas the streams version actually is iterating over a collection.

Regardless, FP-style code isn’t “shiny new stuff”—it’s been around for decades in languages like Lisp or Haskell. Functional programming is just as theoretically “fundamental” as imperative programming. (Not to mention that, these days, not even C corresponds that closely to what’s actually going on in hardware.)

>> apples.stream()

>> .filter (a -} ajsRed());

>> .forEach(giveApple); [sic]

> The red apple streams one is good. It shows how developers chase shiny new stuff with no respect for fundamentals.

The problem isn’t streams, it’s slop.

Care to explain the last one? The presentation is weird and stupid, but I don't see any obvious (technical) issue other than the missing bracket on the left, unlike the first two
Iterative example doesn't iterate, mismatches parentheses and brackets. Because of this, the iterative example is shorter and simpler than the "short & simple" lambda example.

Lambda example is to the best of my parsing ability this:

  apples.stream()
    .filter(a -λ a.isRed());  // <-- note semicolon
    .forEach(giveApple);
Should be

  apples.stream()
    .filter(a -> a.isRed()) // or Apple::isRed
    .forEach(a -> giveApple(a)); // or this::giveApple
It's also somewhat implied that lambdas are faster, when they're generally about twice as slow as the same code written without lambdas.
It's interesting to see how LLMs make mistakes sometimes: replacing `->` with `-λ` because arrow sort-of has the same meaning as lambdas in lambda calculus. It's like an LLM brain fart replacing something semantically similar but nonsensical in context.
Probably morphed > into λ because they're similar shapes, and lambda was in the prompt. Image models are often prone to that sort of hallucination.
The 'long' code for checking apples is shorter, but it's missing the external for loop. So I guess you could say it's not (ahem) an apples to apples comparison.
I'm not OP but:

- missing ")" on the left side

- extra "}" on the right side

- the apples example on the right side ("Short code") ist significantly longer than the equivalent "Long code" example on the left side (which might also be because that code example omits the necessary for loop).

- The headings don't provide structure. "Checking Each Apple" and "Only Red Apples!" sounds like opposites, but the code does more or less the same in both cases.

No “for” loop in the example purportedly showing an iterative approach.

Not mentioning the pain of debugging the streaming solution is also a little disingenuous.

Those are so funny that I was forgetting to breathe as I was laughing so hard, man that's excellent haha. Thanks for sharing them, even if we are cooked as a society...
Minecraft JAVA --------> C++

that one gave me an actual lol.

LinkedIn is a masquerade ball dressed up as a business oriented forum. Nobody is showing their true selves, everyone is either grinding at their latest unicorn potential with their LLM BFF or posting a "thoughtful" story that is 100% totally real about a life changing event that somehow turns into a sales pitch at the end...
LinkedIn is a fucking asylum populate by the most unhinged “people” and bots. I don’t know a single serious technical person active on LinkedIn.
There's a whole community devoted to pointing out LinkedIn Lunatics!

https://sh.itjust.works/c/linkedinlunatics

I short the stock of companies whose leadership is wasting time posting to LinkedIn instead of… y’know… leading their org. The more they post the more I short. Similarly, the less-attached-to-reality the post is the more I short.

I wish I could say I’m making bank off this strategy - but pretty-much all the slopposters (and the most insufferable of the AI boosters) are all working for nonpublic firms, oh well.

Maybe not a winning strategy because a lot of public companies have a comms team that manages the CEO’s LinkedIn. Thereby saving the valuable time of the CEO themselves.
> a comms team

Right; and those PR/comms/social-media-managers know better than to post LLM slop to LinkedIn.

There are people who write genuinely interesting stuff there as well.

I use block option there quite a lot. That cleans up my experience rather well.

Can to share some of them? Genuinely curious.
I don't know if I'm helping make things better or adding to the problem, but here's the sort of thing I share with my audience: https://www.linkedin.com/pulse/day-life-hft-developer-two-de...
I guess people one would follow on other platforms, plus bunch others posting in my native language.

Daniel Stenberg Jason Fried David Heinemeier Hansson Nick Chapsas Laurie Kirk Brian Krebs

> Brian Krebs

Even Krebs has switched to posting uncritical AI slop now.

https://krebsonsecurity.com/2026/02/starkiller-phishing-serv...

A rewrite of slop post at https://abnormal.ai/blog/starkiller-phishing-kit

In the meanwhile, the exploit.in thread where this phishing kit is advertised is just full of users making fun of the author for selling vibecoded garbage. One user seems to have paid for the phishing kit, and says it just doesn't work at all and says the author blocked him on telegram after he complained.

It's a marketing piece hyping up a phishing kit that doesn't exist. Krebs gave up a long time ago, if he gave a shit he'd be more than capable of going on these forums to try and verify this story before uncritically repeating AI company marketing materials.

> LinkedIn is a masquerade ball dressed up as a business oriented forum. Nobody is showing their true selves

That's the main trait of almost all social media. A parade of falsity, putting on the show for everyone else, being what you wish you were and what everyone else dreams of being or envies.

LinkedIn is about boasting and boosting the professional life, other social media is for the personal life. More or less equally fake.

> these people clearly aren't reading anything they're posting

I'm surprised they are able to care so little. Somebody actually published this and didn't care enough to even skim through it.

Of course they aren't. The text to go with those diagrams is also machine generated.
The comment you’re replying to already stated that.
Yep! Quit LinkedIn when it went downhill. Has only gotten worse since then. Most social media is filled with AI slop. For someone who grew up in the 90s-2000s BBS/IRC era this sucks!
LinkedIn and GitHub, hmm. Wonder if there's a common thread...
And LinkedIn is Microsoft as well...

IMO Microsoft is right at the nexus of opportunity for solving some of the the large _problems_ that AI introduces.

Employers and job seekers both need a way to verify that they are talking to real identified people that are willing to put in some effort beyond spamming AI or wasting your time on AI run filters. LinkedIn could help them.

Programmers need access to real human-verified code and projects they can trust, not low-effort slop that could be backdoored at any moment by people with unclear motives and provenance. Github could help.

etc. etc. for Office, Outlook ...

But instead they've decided to ride the slop waves, throw QA to the wind, and call every bird and stone "copilot".

totally. I'm really getting behind the slight replacement of TL;DR to AI;DR If you can't be bothered to read your own AI slop, then I'm not reading it either.