Last Notes
Not to mention that DeepSeek keeps substantially innovating in public. Astra is literally built on an old DeepSeek paper!
Ha, maybe that too. But I get the sense everyone has been racing ahead on days quality roughly in parallel, including synthetic data.
They didn’t even bother to steal all the money, a huge amount was left vulnerable. Also, specifically to liquid, they have back doors to halt the system so reporting it to them can likely get the issue locked down faster than you can exploit it.
Finally, they kept 15%? That’s not even remotely white hat behavior. They just wanted to use the term “whitehat” to reduce risk of FBI.
The claim that Chinese models are “just distilled western models” is obvious nonsense. The Chinese models often end up better than the ones they’re supposedly distilled from, and K3 came out when Fable had only been accessible for a very brief period.
The distillation claim seems to basically be a fantasy of the western labs that want to think they’re just that far ahead and Chinese people are all dumb copycats - no different than what tech thought of China 15-20 years ago.
Well, yes, an individual tx that was invalid, just not crypto code failing to check it.
Yea that’s the bug, but that’s not the fix. That was an attempted fix but in fact it was incomplete, the exploit, afaiu, exploited the patched code.
I saw someone saying mempool didn’t accept that block. Unclear what’s going on I haven’t really dug into things
Nah, it was not an individual tx that was wrong. Based on a few people’s LLM’s analysis that looks right it was a higher level consensus failure not the crypto code.
No, means “boy this shouldn’t have happened”
https://blockstream.info/liquid/tx/ce4caece413cd9d444ce7ed9f54e5b328b3da5e4af301aff59a3571f76e988f2
I’ve seen that claim, but zero evidence for it.
There was supposed to be a whitelist of withdraw addresses. Maybe it was also a break in the range proof but there would have had to be a second bug. Seems more likely there was a software bug in the peg out verification more generally.
Qwen3.8-Flash-Next is smarter and needs less VRAM anyway 🤷♂️
This is the thing that gets me. That should have been an easy protection against this kind of hack.
Oof 3bpw? That’s usually pretty lobotomized.
Wait until they start hacking banks and laundering it through Bitcoin…
This is your regular reminder not to use Wireshark to read untrusted packets.
https://image.nostr.build/540253d80ac412578485d31cd6dbc814a2e9943b49dcbae4551943296bdcc57c.jpg
This is far from the first time wireshark has had a critical RCE vulnerability 😭. In fact far from it, they gave them every year or three.
I mean fuck even if it were unlicensed money services if it’s not illegal where it’s hosted and it’s not trying to scam people why would you take down a domain?
Whether they care about Bitcoin or not I’d really expect something from a Pirate Bay founder to understand that what’s illegal in one country may not be in another…
Probably tomorrow, and also again next week, and definitely for real about a month and a half from now.
Damn that’s so disappointing from njalla.
#nevent1q…ccew
Got on a plane last night and my wife joked “and Asia hasn’t even woken up yet”. Indeed, guess we’re living in 2018 and Asia woke up 😂
It was a refrain in like 2018 cause the price would always pump during Asia trading hours.
All the more reason for Luke to select a PoW he (or a friend) can premine?
Apparently he’s feeling it…
Problem is it needs a fork to reduce the difficulty. You’d need to convince people to follow that…
I mean hard to argue with honestly.
Yea it is, it’s “fair” in that anyone with like a single S9 can select the outcome. The only reason to do it in such an obviously broken way is if you want to game it.
Wait so let me get this straight. Luke’s (obviously) trying to rig the PoW function selection for his coin, probably cause he has ASICs for one option.
So….he’s just trying to create a coin he can premine so he can dump it on the BIP 110 followers? Man that’s low af.
Weird, I’ve asked everyone and no one seems to be having issues.
Almost certainly hardware failure, though if you have the specific error I’d be curious to see it.
lol, no. The security model hinges on it being *hard to change* Bitcoin’s rules. That applies equally to miners as to random nodes.
People got this weird idea after segwit that a “UASF” is this magic thing where a bunch of people on social media get together and start making threats and suddenly Bitcoin’s consensus rules will change.
That’s nonsense. I’m sorry if you were told that the thing that activated Segwit was a UASF and nothing else but you were lied to.
Yea, maybe they thought if they got lucky and reorg’d one block they’d scare other pools into joining.
For the good sports, yea.
I will never forgive the subset of the BIP 110 community who drove gloria out of bitcoin with their massive harassment campaign, though - she did more for lightning from the core side than anyone since segwit and her leaving bitcoin was a massive loss.
Probably but you can’t buy 50% of the network….
Ha well that turned out to be wrong. Directionally right, but they got lucky fast.
I missed that 110-coin isn’t even banning the subset of data BIP 110 seeks to ban, and probably never will.
bip110.fail
I’m betting any remaining 110 miners will turn off before they get a block. Boring, but certainly the better outcome for Bitcoin.
I’ve had enough drama this week anyway.