Wouter Post (@WouterSchiller) header
Wouter Post (@WouterSchiller) profile picture

Archeoloog. Wahrnehmung. J'espère. https://teapawt.net/t.co/YexUKEi9Df

Joined November 2012
Aan het einde van dit fragment een vergenoegde gezichtsuitdrukking, waarvan de origine de goede verstaander niet zal ontgaan.
"Voor ons was het makkelijker geweest om eruit te stappen", zegt VVD-leider @DilanYesilgoz over de coalitiecrisis. "Maar binnen zes jaar zet je dan het land voor de vierde keer op slot." #WNLOpZondag #WNL
3
10
4,252
Wouter Post retweeted
I have been handed an object of immense power
149
862
49
13,939
186,883
Wouter Post retweeted
'I didn’t want to talk about Claude. I wanted to talk about the user, about addiction and dehumanisation, about power and accountability.' Theologian Carmody Grey explains how the AI industry is courting the humanities — but asking the wrong questions ft.trib.al/5sZJqAT
10
53
3
156
38,897
Wouter Post retweeted
Want to know why young people are angry and feel increasingly hopeless? This chart says a lot. V good story by @jburnmurdoch. giftarticle.ft.com/giftartic…
140
1,009
133
5,437
347,580
Kallas (1 december 2024 –) Borrell (2019 – 2024) Mogherini (2014 – 2019) Ashton (2009 – 2014) Solana (1999 – 2009) Jürgen Trumpf (1999) Nach Trumpf (..), war die Kompetenz des Hohen Vertreters stets höchst fragwürdig. Warum tun Sie (...) das der europäischen Bevölkerung an?
No hesitation. Just pure confidence. Kaja Kallas disputes Russia’s claim to China that they fought the Nazis in WW2 and won. Just an absolutely stupid woman.
50
"Deep in the blackness of the cosmic waters, Mahavishnu reclines on the back of the thousand-headed serpent Shesha Without End, sleeping, and as he sleeps he dreams the innumerable worlds." @samkrissfeed crushing the moment, again. samkriss.substack.com/p/you-…
32
Wouter Post retweeted
Replying to @Maxirex26
Practically in terms of being able to show anything directly, this doesn't matter. Number theorist already tested something like the first trillion zeros or something of the zeta function, and they all sit on the critical line. So for any prime up to any size number that would ever actually get used in anything anyone is doing, assuming RH was true was good enough. We knew it's true up to some absurdly large number. What matters is that it was assumed to be true for a long time in many other theorems and proofs, and now all of those don't need to assume it's true. As long as those proof aren't resting on any other assumed but not proven theorems, then they're all provably true too. For example we can now say that Miller’s primality test is provably correct and also deterministic, so we don't have to use any slower methods (like AKS). Similarly there's probably a bunch of cryptography functions that are guaranteed to run within some limit now because the bounds of where the primes they need to find might be are tightened up (although again, I would guess many already assumed GRH as a starting point?) But I'm not sure if any of those algorithms are actually useful for anything anyone's working on in the "real world" and also needed to be searching primes beyond what were already proven to be on the critical line.
2
14
2
359
91,787
Burry on fire again (...). Lack of proximate cause preceded by q.e. etc., i.e. financial crack cocaine for years, dixerunt i.a. Druckenmiller et Dalio. May just find out soon if the ordoliberals, the premises of whom appeared beyond verge of extinction, are right after all.
One similarity today to 1929 and 2000 is the lack of a proximate cause in earnings.
46
Repressing real wages "reduced the labour share and increased corporate profits and savings, mirrored in the sharp rise in 🇩🇪 current account surplus." It's significant to see such a sober analysis from one of Europe’s leading policymakers in recent years ecb.europa.eu/press/key/date…
2
32
1
100
4,790
"The dollar is the currency not of a country but of a class."
3
29
3
175
13,245
This flow, generally overlooked, is a result of careful long-term policy, mainly via the unique capacity of an exceptional nation (sic) to undermine other nations while inactivating any resistance in those nations to such undermining. Little to do with capitalism as intended.
1/7 Ruchir Sharma: "So far this year, 80 per cent of global equity flows have gone into the US market, double the long-term average. No other country has ever drawn such a large share." ft.com/content/1f1a2a9b-9429…
51
Semiconductor executives say AMD and Broadcom can't take on the huge credit guarantees Nvidia has given cloud providers and data center developers, which leaves a gap if Nvidia pulls back. Full story: thein.fo/4iH6RNq
37
84
32
347
145,532
The interview with Marshal Zukhov was banned in the Soviet Union and was only declassified in 2010 ! 44 years after it was recorded . In 1966 Marshal Georgy Zhukov sat down with Konstantin Simonov and talked about October 1941: the front collapsing, Stalin summoning him, and how Moscow was saved. Soviet television shelved it for decades. Now in full, 53 minutes, English subtitles.
87
1,848
79
9,412
476,246
'The Sun.' (1904) Giuseppe Pellizza da Volpedo began work on what he described as his greatest picture in 1903 and completed it a year later. His aim was to achieve 'an exact replica of a sunrise,' something achieved by applying pure, unmixed strokes of colour side by side.
8
459
25
3,276
52,543
Wouter Post retweeted
Houellebecq, Daoud, Sansal, three cases in one year.
7
89
4
472
14,102
Wouter Post retweeted
I have a major result to share. I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China. This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model. Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves. I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems. I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used. It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20. These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs. I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at unemploymentlabs.com/checkpo…. There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.
1,823
6,422
2,210
41,256
6,802,038
Wouter Post retweeted
Replying to @politiek_tweet
Dank voor uw eclatant intellectueel uitdagende bijdrage.
5
1
54
4,770
5/ it’s truly uncanny how both sides are invoking the iconography of 1981, albeit without seemingly any self awareness on the Tusk side.
3
12
1,628
Wouter Post retweeted
Me: "You did your secondary school in Holland." Pope Leo XIV: "The town next to where I grew up was South-Holland, Illinois. And then I went to high-school in Holland, Michigan. The Tulip Festival was always part of my life." 🌷 – at Montevideo, Uruguay
41
297
58
4,065
345,679