Yes, GP was discussing Ryanair flight prices and claimed that a "similar flight in the US" would be more expensive. I'm pointing out that similar flights in the US are actually cheaper, and we have budget airlines too.
I'm assuming this is vibe coded, because it's got a bunch of the usual tells, so to the people who do this: can you please stop making stupid scrolling presentations where I can see less than a slide of information at a time? Please tell your clanker to just write a blog post instead, or better yet, write it yourself.
My corporate firewall blocked this due to it being a newly registered domain.
So I can't even see it, I care less about "vibe coding" but it sounds like someone registered a domain just to get attention on their amazing take about why they think they're qualified to tell the world the future.
> My corporate firewall blocked this due to it being a newly registered domain.
Was surprised mine did not - usually a toss up with HN links. I don't get reasoning just "NONCOMPLIANT ACTION". It is interesting to have a flag telling you the domain is new, though
So a realtime whois lookup is performed when the request to the DNS server is made, and if the domain was only registered within X days/weeks, then return 0.0.0.0 (or other such blocking method).
See, I've outbuilding tried compiling lists of newly registered domains to use as block lists, bit they're very large lists that my under-spec systems struggle to deal with. As such, I scaled back / shelved the project.
Looks like Adguard DNS and NextDNS offer blocking NRDs as an option in their paid services. I shall be looking into this further.
Ive been out of the authoritative dns game for a while, but asi recall…
Larger providers can also get bulk zone access for TLD’s and whois/registrar data. For this use case it’s relatively easy to create a time based filter on that. Anything that’s “new” will be de facto absent from your “allow” check and create an implicit deny.
Then your large IT provider or recursive DNS system will probably layer in RPZ where they can insert explicit denies at resolution time. Either based on QNAME, RDATA, zone, etc.
Vibe coding has made them a lot more common. Before, you'd need to put a lot more effort into making a website that worked like this, and it wasn't worth it for a random post. Now this person's entire website is posts like this, and I've seen many more in the past few months.
But now with vibecoding it feels like the default for articles to have fancy animation, rather than the exception. I guess that by having a fancier presentation it subconsciously legitimizes the content more so you're less likely to critique it as compared to a simple blog post where you pay more attention to the words and can realize that it's very surface level.
This one is good but the content doesn’t speak to me like the above. Would have been nice if OP had added a writing style suggestion to the prompt. Vanilla LLM text is sad
eh, am very biased as I design similar sites but I honestly prefer these to what would be likely string of random social media posts.
I like having relevant graphics stickied while text is displayed alongside it (assuming by blog post you mean the typical page-like top to bottom approach).
edit: damn, if these designs are hated what modern approaches do people like? I feel like scroll based text is a relatively unexplored idea compared to the typical blog post.
When it comes to web content, I vastly prefer web like interfaces that you can't reproduce in print.
While I'm not that against to such a web page (when viewing it on a monitor, not a phone) I'd say follow these points:
• Don't hijack any browser functionality. Scrolling shall scroll the document, the end.
• Don't scale something to the screen size, breaking zoom! Especially don't do it so that when zooming causes a different scroll position and then it all jumps to a different slide. WTF!
• Make it accessible!
If you want to make flashy graphics and animations make a game. That is not meant as to belittle games, I love games.
I don't disagree with those, and I'm able to do them myself but what do you mean by this?
"• Don't hijack any browser functionality. Scrolling shall scroll the document, the end."
How would you consider the page hijacking scroll functionality? You can scroll down normally. There are animations based on scroll position, maybe that's what you meant?
I mean the animated graphics and some other hackery going on when you try to zoom. It thinks to detect a different scroll position and jumps to a different slide. Also means you can't convert it to a PDF or print it, if you want to (though I wouldn't want to do that).
> edit: damn, if these designs are hated what modern approaches do people like? I feel like scroll based text is a relatively unexplored idea compared to the typical blog post.
The New York Times generally does the "lots of data, text and graphs in a scrolling presentation" tastefully.
Any articles that demonstrate this? Haven't read the NYT in like 15 years. Only vaguely familiar with the data viz that Mike Bostock created while there, and only because he shared them on a defunct blocks site.
I want it to be something you could reproduce in print, though. As close to what I'd get putting it in reader mode as possible.
We've spent a long time optimizing the printed word, including pages with diagrams and illustrations on them. You don't need to reinvent the wheel for every blog
I mean if you want print views (reader mode in firefox works well if the dev took the time to structure everything semantically), definitely use the print functionality. Maybe I'm one of the last people that still makes print media queries, but I do find value in them especially since they're quite easy to make.
Tired of you vibe coding detectives. Who the heck cares? If you don't use AI these days you're not smart. Why call it out, as if you caught them doing something unethical?
You know there's like an entire generation of devs raised hearing "copyright violations are theft, don't do it, it's band"? Unsurprisingly many of these people indeed think that when Anthropic and OpenAI does industrial scale copyright theft that's bad.
Also, calling most of humanity "stupid" is pretty stupid.
Because a lot of us don’t want to see AI slop and it’s very useful to have that pointed out in the comments before I waste time clicking on the link and slowly coming to that realization myself.
Yet. But they are probably working with Chinese partners (including the government) on releasing something (maybe with Alibaba models instead of Google models, on a Chinese-local cloud rather than google cloud).
A quick check showed it is estimated that Apple gets about 18% of it's profits from China but only maybe 7% from EU countries (ignore Apple's definition of Europe!).
Maybe China is easier to work with - perhaps their rules are made clearer?
China has 1.4 billion people and they are rapidly increasing their wealth. The only surprising factor is that Chinese cell phone producers haven't eaten up apple's marketshare yet.
Almost too much so, it often feels like opus is pushing back for the sake of pushing back. The way old models used to add disclaimers to every message regardless of content
People are weird about their cars and make major errors in judgement as a result (e.g. we tolerate incredibly high rates of people getting killed because they were "hit by a car", as though the driver had nothing to do with it). Pushing back on that is absolutely worthwhile.
No, in general I don't buy this idea that if we start using awkward phrases like "died by suicide" everywhere or avoiding phrases like "car accident" (which, despite what advocates claim, is a literally accurate description of unintentionally hitting someone or something with your car) but avoid changing any of the circumstances that cause the behavior it changes anything.
That's a completely different claim from the one you were making in your previous comment.
> avoid changing any of the circumstances that cause the behavior
The normalisation of unsafe driving is the circumstance that causes the behaviour. Just look at how the cultural shift in how drink-driving is perceived over the last few decades has changed the rate of it happening.
It was mind blowing the first time I got a refusal, and retorted "yes you can" and had that work, but now it's just another reason to move to a different model.
Part of the idea behind a moratorium is to allow time for impact analysis so those regulations can be written appropriately.
Part of the reason we have big tech getting away with regulatory capture in the US is because they have enough capital to entrench themselves in the economy faster than regulators can react.
Will they also pay for rolling back the infrastructure and restoring the land to its previous state after they go bust and no longer need the data centre?
I don't understand why people look at bugs and go "oh it must be AI". Did humans never write bugs? I've seen plenty of platforms where bugs like that were the norm, bad software engineering has always existed.
Before it reached universal usage, I repeatedly saw teams I work with adopt AI heavily and then immediately start shipping more and worse bugs. It's probably not eating up all of their productivity gains but it's eating a lot. Just today I saw someone merge a PR whose description was clear that it would cause an outage if deployed, and I really don't see any explanation other than the humans who were theoretically in the loop not reading it.
Bugs did exist, but bugs that come from AI generated code can be easily avoided if a good enough engineer designed it. Also, the rate at which AI generates code means that there's A LOT more bugs now than there used to be.
I'm sure the engineering teams of every company in the world is just "good enough engineers" and not, you know, regular devs who ship bugs at the same rate that has been now. And I'm sure your personal anecdote of "I saw a bug" shows that there's more bugs being shipped now.
The evidence that per-token inference _is_ subsidized is (a) competition is a bloodbath (b) these companies are raising more money than any company has raised ever (c) a maybe-profitable quarter is maybe-coming for Anthropic after maybe-signing a compute deal with SpaceX that legitimizes both companies.
The evidence that per-token inference _is not_ subsidized is... a quote or two from Dario and Sam Altman
yes, and theres no evidence that they arent (or can't) use profitable inference to subsidise those other expenses. Some companies will keep spending massively to train better models, and some other companies will not, and offer good api prices. Which will end up being used? That depends on whether the spending turns into better value models
The fact that Anthropic models are offered at the same API pricing by not just themselves but AWS, Azure and Vertex despite Anthropic taking a major slice on licensing along with the cost an open weight 1T parameter model like K2.6 costs to run on any third-party provider, make it unlikely that API inference cost are subsidized by the labs.
Openrouter? i.e. Even excluding Deep Seek inference for very large open models is way cheaper. Maybe these providers are not very profitable but its highly unlikely that they are losing $4 for every $1 they make since selling inference is their only product...
reply