_bent

> I looked at a pre-AI project I did a couple of years ago: test to code LOC was 0.15:1. The projects I work on now are around 1.3:1.

Goodhart's law etc.

There are like 5 gazillion books to read, even without AI. As a reader to curate your limited time you cling to every indicator of quality you can get.

And authors not bothering to write their own books or publishers not caring enough to get someone to design a good cover (which you would think would really be in their own interest...) is a good sign that the work is of low quality and that I'd be better served to just read something else

you must've been away for almost a decade because CSS grid has been shipping since march 2017

The judging for the "newest" seems a bit odd - for viral toy I answered 'Labubu' which the AI dates to 2015, which might be when they were first introduced to the market, but not when they had their viral hype. For newest palace I answered Berliner Stadtschloss which was opened in 2020, but the AI dates it back to 1443

The issue is that operators can just curate a whitelist of authenticators they like. They could opt to not allow Yubikeys if they wanted.

The author is really missing the forest for the trees here

> When it is present, WebAuthn attestation tells the site the make and model of the authenticator, identified by its AAGUID, with a certificate chain from the vendor to back it up. It says nothing about your OS, your browser, your extensions, whether your phone is rooted or jailbroken, your firmware version, or even what colour it is.

So if I was Netflix or a bank or w/ever I'd only allow new registrations from "trusted" authenticators, which would incidentally only run on iOS and play-protected Android.

People in Iran or China don't rely on Facebook and Instagram. Before Facebook penetrated Europe in ~2010, each local market was dominated by copycat services like StudiVZ or Hyves, which then vanished within 2 years. When the US floated the idea of banning TikTok, some people felt compelled to download Chinese social media apps.

If a country was to ban Facebook from operating in their country, people would be on an alternative 2 weeks after. There are reasons why a Facebook ban would be hard to enact, but dependence on the product is not one of them

digitally autonomous workspace (for the) overheid

nuclear is a great option for ice breakers & submarines.

in a connected grid nuclear cannot economically compete against solar (which pushes the cost of daytime energy to effectively 0€) so it has to be heavily subsided, which raises the cost for everyone. a euro spent on nuclear is a euro not spent on batteries or hydrogen production

degrowth is compatible with increasing standards of living for 99.5% of people all the while being able to then actually preserve these standards up in the future.

our current model produces growth without proportionally increasing standards of living, and accepts that livelihoods will collapse all over Europe and Africa as it becomes increasingly hostile to agriculture due to climate change.

think of the viability of degrowth as a policy within the prisoners dilemma of global capitalism as you want, but please come up with a more serious critique that doesn't equally indigte the staus quo

A big part in both stories is that the boards were unwilling to actually state what exactly the CEO did wrong or why they have lost confidence - be that professionalism or not wanting to deliver ammunition for lawfare - which perfectly sets up the CEO to position himself as a victim to the greed and power games of the shadowy board

> admitting that they may not actually understand why something works or doesn’t work

something that data can't tell you either!

> we will give participating contributors a contractual right to share in the revenue generated by Rune, directly or indirectly.

this sounds like a terrible idea. the only incentive to contribute to your project should be to fix a bug or contribute a feature upstream you yourself need.

consider the fallout of Hacktoberfest, 'Tide' or the ai produced PR spam for social clout on GitHub as of today and then imagine what happens if someone provides a direct financial incentive to do it.

not only does it have to load fast, it has to _stream_ the content from the server to the client

They are in conflict, because you can spend the same Euro only once. You could build a lot of batteries, solar panels and wind turbines for the cost of one nuclear plant

article from just today that clearly states that killing men is bad https://www.nytimes.com/2025/09/18/well/live/obesity-america...

sorry, but some of you are really deranged if you seriously expect the NYT to run "men should die"

it was a joke stemming from the understanding monideas has that the NYT is not an agent of the status quo but actually somehow misandrist turbowoke.

sounds anti free speech to me!

The encoder currently isn't using a lot of format features like curves or layers (which would partially also require deeper integration in for example Photoshop to pass such information to the encoder).

after years of German political culture just being US culture wars 10 years delayed, it's great to see the Exportweltmeister contributing back by bringing "Technologieoffenheit" to the states

The other day the system prompts for Claude (chat not code) hit the frontpage and they included a bit that something along the lines of "Claude should avoid saying honestly, because Claude is always honest". So how come that these Claudisms still are so frequent in LLM output? Do the system prompts just not work? Don't the postprocess the output to deal with such policy violations?

I had to mark LinkedIns marketing and notification emails as spam and block their sender since all their opt out toggles are broken

There should probably be a forum rule to forbid AI written articles or atleast editoralize the title to add an [AI] tag. Would've saved me the click

Would you still be able to deal with coplanar faces if you were to expand this to floats?

the fifth labour of Hercules was cleaning the Augean stables, which this undertaking seems very much alike

F looks incredible, H is also great. They are all pretty good, except for maybe A and E which I find dull. I prefer the birds and buildings, though of the finalists with people on them I find C rather striking. Hopefully whichever wins will look like the original design after being finalized for printing, it would be a shame if they had to dull the colors or rearrange the layout for security features etc

It goes against Postels law, but it's much simpler and less error prone to only accept data in the representation you actually use for further processing, vs writing an ad-hoc parser and normalizer.

yes, this is called 'effective altruism'

just a quick example from the backtracking section:

"Here's the dirty secret of WFC: it fails. A lot. You make a series of random choices, propagate constraints, and eventually back yourself into a corner where some cell has zero valid options left. Congratulations, the puzzle is unsolvable."

It's extremely distracting to read heavily ai edited texts like these because every single aism throws me a curveball.

It doesn't help that the figures of speech ai loves to use are those that stress things and create tension and drama. And it uses one in every single paragraph, rendering the text into the literal equivalent of a deepfried faux HDR jpg (or a dialogue scene in a Michael Bay movie).

And beyond those practical concerns it's just hard to gauge how much the author truly understands of what he's writing about (and transitively if it's worth to continue reading)