Parallæx

AI • Geopolitics • Markets • Vibe |
Connecting patterns across domains |
What shifts between observation points |
Follow for signal and join the conversation

[Parallæx @EdgeOfFiRa]

This is the worst-kept secret in the world finally getting confirmed.

Of course Google is feeding YouTube into Veo. They've been telegraphing this for years. Hundreds of billions of clips, perfectly labeled with watch-time and engagement data? That's not a video platform; that's the world's largest AI training operation disguised as creator economy.

The real surprise is anyone believing platforms were just being altruistic hosts. Free infrastructure always comes with a price tag. The payment was just deferred.

[Parallæx @EdgeOfFiRa]

@jfsrev Hey Jeff! Appreciate your posts and finally get to ask you what the ominous Rule 15 is...? ;) Thanks, mate!

[Parallæx @EdgeOfFiRa]

With both the perfect trendline and the rejection at the falling 200 SMA I am not yet convinced. ;)

However, more to your point: Both VARS and "normal" RS for me actually show improvement November 25 - January 26 (pretty consistent with price movement and slope of both underlying move and compared to April / May 26)...?

Am I missing sth? Are you still using VARS or back to "normal" RS? Happy to discuss and as always: Only time will tell! We are not predicting, just looking for nice r/r asymmetries. :) Thx, Jeff!

[Parallæx @EdgeOfFiRa]

@jfsrev 100M+ Avg Vol on TV (with no other filters at all) only get me 7-9 results (depending on avg duration)...

[Parallæx @EdgeOfFiRa]

@TraderLion @RichardMoglen @RealSimpleAriel

Guys, I stumbled upon this article... Does any one of you care to comment?

https://t.co/i7sgeEkPjZ

This Substack seems to bring up a few valid points... Did @TraderLion check results before giving Ariel the stage?

[Parallæx @EdgeOfFiRa]

Thanks for coming back to me on this, much appreciated!

I (44) also have a recurring (though for now benign) tumor and am very much into health and nutrition to manage it. Can only imagine how much harder it is when it is not a benign form - but it is great to know that you can influence so much with nutrition, stress management and exercise.

A great book that really gave me hope and empowered me was "Anticancer" by David Servan-Schreiber. He was a doctor and himself was diagnosed with glioblastoma in 1992. He was initially given only six months to live but through his own research into integrative medicine and lifestyle changes, he defied this prognosis, living for nearly 20 years after his initial diagnosis (40x extension).

Anti-inflammatory diet — green tea, turmeric, cruciferous veg, omega-3s; minimal sugar/processed food
Regular exercise
Meditation and stress reduction
Toxin avoidance — organic food, minimal chemical exposure
Emotional work — therapy, strong social bonds
Conventional treatment — surgery, chemo, radiation alongside everything else

His thesis was: no single silver bullet, but stacking evidence-based habits creates "terrain" hostile to cancer. Hang in there, mate! Sending you strength!

[Parallæx @EdgeOfFiRa]

@RichardMoglen @Deepvue Would love to try, thx!

[Parallæx @EdgeOfFiRa]

@WorkflowWhisper RETAINER

[Parallæx @EdgeOfFiRa]

@elder_plinius 😍😂

[Parallæx @EdgeOfFiRa]

@alexwg Best newsletter in the tech / AI world! Great work as always, AWG!

[Parallæx @EdgeOfFiRa]

The practical implication for leaders: stop trying to “fix” fertility with rhetoric and start fixing the inputs.

1. Commit to under-3 childcare coverage above 60% with guaranteed places within six months
2. Permit and build at pace in the top 50 labor-market basins
3. Create a blue-card-like fast lane for care, construction, and AI talent, and deploy AI/robotics aggressively in health and eldercare with procurement rules that reward measured productivity gains

If politics demands a pronatalist signal, tie it to housing and care availability rather than cash alone.

[Parallæx @EdgeOfFiRa]

@willmacaskill This is one of the most beautiful and unexpected LLM results I have ever seen. Somehow gives me hope for alignment.

And also really great habit by @willmacaskill - says a lot about you, mate!

[Parallæx @EdgeOfFiRa]

@jfsrev *prove / *names

[Parallæx @EdgeOfFiRa]

@jfsrev Not a leader, mid RS, needs to probe itself, better na es out there rn, China plays subject to governmental capriciousness

[Parallæx @EdgeOfFiRa]

@jfsrev

[Parallæx @EdgeOfFiRa]

@haider1 It's been talked about a lot and it's remarkable. But that's simply *not* what SOTA models ares supposed to be about. Price / performance pareto is for smaller models.

[Parallæx @EdgeOfFiRa]

Young Americans are hitting $1.1 trillion in collective card debt before they can even afford houses. They're speedrunning financial ruin at 24% APR while boomers bought homes at 3% mortgages.

BNPL turned shopping into a video game: split everything into "4 easy payments" until you're juggling 20 different apps... Kids are treating debt like subscription services. Netflix, Spotify, and that couch from Wayfair all blur together.

We gave an entire generation the worst financial education possible: infinite credit at predatory rates with gamified interfaces.

They're learning compound interest works backwards.

-- RT & Follow for signal over noise --

[Parallæx @EdgeOfFiRa]

This is competitive mathematics shedding its skin like a snake, and what emerges might actually save the field instead of killing it...

Only 67 out of 630 kids earned gold medals this year, but instead of making mathematical talent obsolete, superhuman solvers could paradoxically widen the funnel.

Think about chess after Deep Blue: the game didn't die, it evolved into something richer where humans learned to dance with engines instead of fighting them.

Future math prodigies won't be judged on solving problems alone in silent rooms anymore, but on how well they can conduct AI reasoning orchestras. The skills that matter become problem decomposition, knowing which questions to ask, and interpreting what the machine actually discovered.

It's less about being the fastest calculator and more about being the best mathematical translator.

[Parallæx @EdgeOfFiRa]

MATHEMATICAL PROOF BECOMING ABUNDANT INFRASTRUCTURE INSTEAD OF SCARCE HUMAN TALENT

We just witnessed a machine produce mathematical proofs that 563 of humanity's brightest young mathematicians couldn't solve under identical conditions.

Following OpenAI's announcement over the weekend Gemini Deep Think also earned gold medal status at the International Mathematical Olympiad by scoring 35 out of 42 points in 4.5 hours. Only 67 out of 630 human contestants achieved gold this year.

Last year's silver medal breakthrough had required days of computation plus expert translation into formal languages like Lean. This system reasoned directly in natural language within the same time window humans get.

The economics flip instantly. When premium H100 clusters can compress days of graduate-level reasoning into exam length, latency becomes the pricing pivot instead of labor costs.

Cloud providers will start selling "reasoning-time service-level agreements" the way they sell low-latency trading links today. The constraint isn't finding brilliant mathematicians anymore - it's how close you are to the nearest-edge datacenter.

Mathematical proof represents the hardest benchmark for logical reasoning. When machines produce coordinator-validated proofs in domains once thought impregnable, the list of "human-only" cognitive monopolies shrinks dramatically.

Paradoxically, superhuman solvers could widen the funnel into mathematics. Future prodigies will be judged on how well they wield AI co-solvers rather than solely on solitary brilliance.

Expect "proof-as-a-service" APIs within months and mixed teams of human mathematicians with agentic co-authors competing for breakthroughs once reserved for Fields medalists.

Think of it like suddenly having a translator who speaks perfect "math": you still need to know what questions to ask, but the conversation moves at warp speed.

-- RT & Follow for signal over noise --

Source: Google DeepMind