> The scrubbing, Meta teams explained in documents regarding their efforts to reduce scam discoverability, sought to make problematic content “not findable” for “regulators, investigators and journalists.”
This seems to be the "smoking gun"... but it's unclear from the article what the source or context of the quotations are.
Needs verification on the normal user side of the experience. Like noted, unclear whether the ads are completely deleted, or just deleted from what regulators would return in search results.
Later part makes it sounds like the ads were at least deleted from the Japanese / Taiwan results areas completely, and then just redistributed to other geographic areas.
"If an unverified advertiser is blocked from showing ads in Taiwan, for example, Meta will show those ads more frequently to users elsewhere, creating a whack-a-mole dynamic in which scam ads prohibited in one jurisdiction pop up in another."
However, still difficult to tell from the story what the experience of "normal" Japanese / Taiwan users is relative to what regulators view.
Notably, from a different article about fake accounts, and how much they cost, Japan's regulation seems to be working (at least somewhat). [1] Fake Accounts for Facebook Available (US, 157,401, @$0.13/account; Japan, 16, @$3.00/account, Dec 31st, 2025 data)
> Thinking of Kubernetes as a runtime for declarative infrastructure instead of a mere orchestrator results in very practical approaches to operate your cluster.
This is a pretty good definition.
I think part of the challenge is the evolution of K8s over time sometimes makes it feel less like a coherent runtime and more like a pile of glue amalgamated from several different components all stuck together. That and you will have to be aware of how those abstractions stick together with the abstractions from your cloud provider, etc...
Sometimes the interfaces are clear and consistent (loadbalancers, for example) and the line between, e.g., an AWS NLB and your ingress gateway, is clear-cut.
Other times there is a significant degree of portability pains and sparse feature matrices (CSI, IME).
they did, judge told authors to get better representation to prove harm. I think it's dicey cause if anthropic loses then it could be catastrophic (i.e. if judge jury thinks reward is 5x of what the proposal is would mean they would need to raise a new round)
> That's a far cry from the 7 million works that he initially certified as covered in the class. A breakdown from the Authors Guild—which consulted on the case and is part of a Working Group helping to allocate claims of $3,000 per work to authors and publishers—explained that "after accounting for the many duplicates," foreign editions, unregistered works, and books missing class criteria, "only approximately 500,000 titles meet the definition required to be part of the class."
They're likely taking the deal because they think there's a good chance they'll lose, and that the final judgement would be more -- possibly a lot more -- than this settlement.
But sure, I bet us randos on HN have a better feel for this than Anthropic's legal team.
The problem is that the agreement does not seem to specify that the electricity has to come from fusion![1] This is actually common in renewable PPAs - if the specific project doesn't come online then the provider has to find an alternative. But those are usually done with established providers which have > 0MW overall capacity. Helion does not. If the PPA is fixed amount vs fixed price, Microsoft might end up on the hook for inflated wholesale prices instead of cheap fusion.
FWIW I agree with the author of that Data Center Dynamics post, it's quite likely that MSFT and Helion are essentially in cahoots by stoking investors with vaporware. But it also seems like Altman might have sold Nadella 50MW of magic beans.
Investors gets ruined on PPAs all the time. With renewables, there's usually some sort of wind turbine or solar park attached, so after declaring bankrupcy the new owners can settle out of court and continue production.
Here there's realistically no way to continue electricity production so the assets will likely be chopped up and sold, if possible. Nobody owns anyone anything and Microsoft doesn't have to pay a dime. They won't get their power of course, but there's no downside for them.
These types articles where a PPA contract is confused with an investment is really common, mostly for nuclear and renewables, but that doesn't make them any more true. Microsoft hasn't invested anything, likely because they know this is (pardon the pun) hot air.
Preprocess the documents to extract certain key features like "Cited Risk Factors", and have the AI model use that instead of each whole document as context?
This seems to be the "smoking gun"... but it's unclear from the article what the source or context of the quotations are.