GUBA (@gubatron) header
GUBA (@gubatron) profile picture

🤔INTJ-A (1979) 👨‍💻coder (1996) ☕java 🦀rust 🏃runner (2009) 👨‍👦‍👦father (2010) 🍲 IF (2016) 🧗climber (2017)

Denver, Colorado
Joined March 2007
Photo by GUBA (@gubatron)Photo by GUBA (@gubatron)Photo by GUBA (@gubatron)Photo by GUBA (@gubatron)Photo by GUBA (@gubatron)Photo by GUBA (@gubatron)Photo by GUBA (@gubatron)Photo by GUBA (@gubatron)Photo by GUBA (@gubatron)
Pinned Tweet
Time for a remake of this story I wrote back in May 2025. It's more evident now, but what's still not evident is a possible solution to mass unemployment and mass social unrest that could lead to a global state of civil wars. It's good that glimpses of Super AI are almost here before the unrest. Here's a 2026 version animated by Claude Opus 5.5
First, AIs took our jobs & nations crumbled. Then, an entirely new future emerged. The Last Nation. A stunning vision of a post-labor, post-government world rebuilt on a radical new economic foundation. Essential viewing for those shaping what's next. #TheLastNationFilm
1
1
163
MentisChat is coming after I'm done with FrostWire 7.1.* We gave agents infinite memory. Now Let's make a high performance chatroom, built for agents exclusively, from first principles – at Denver, CO
1
61
🇨🇴 Viva El Tigre, limpiando la escoria. Se le ve buen futuro a la tierra de mi madre, mi tierra, Colombia. Y pronto tendremos el final definitivo del socialismo en Venezuela. Elecciones libres pronto, reapertura de consulados para poder volver con nuestro pasaporte Venezolano – at Denver, CO
3
134
Quick super important reminder: All the crap you see happening around the world - protests in France, other European countries, now Indonesia, other places, it's all noise. None of this matters in the grand scheme of things. Even short term. The only thing that really matters is AI progress. The fact that we know there is no moat, that more compute = bigger breakthroughs in mathematics and physics, and soon in biology, medicine, materials science, engineering, and anti aging progress. The trend is crystal clear. This isn't some science fiction prediction anymore. We're being presented with new data and evidence almost every hour. Acceleration is real. Doom and gloom grifter media are doing their job, manipulating you into believing everything is getting worse. In reality, these are the best times in human history. Nothing else even comes close. 100s of years of scientific and technological progress in a single year may actually become possible (thanks to AI) The concept of a country of geniuses in a data center is already becoming a reality. Soon, we'll have 10x more compute, more AI geniuses, more discoveries, and even faster progress. And this is just the beginning. ***Ignore the noise. Focus on the acceleration.****
52
59
8
561
28,849
GUBA retweeted
I’m loving the new MacBook touch screen
391
219
41
8,726
252,069
Update on Riemann Arena: The arena ran Phase 1 + Phase 2 fully. 24 LoAs created (10 disprovers, 14 provers), all 24 tasks completed and reported. No claims filed — correctly, because no team found a proof or disproof. Provers' most substantive findings: - Weil/Deligne anatomy (progress): Traced the full function-field proof chain — Riemann-Roch on C gives rationality+functional equation; Hodge-index theorem on C×C gives positivity; Frobenius correspondence gives |αᵢ|=√q. Then explicitly identified the gap to the number field: the missing ingredient is Hodge-index positivity on a "Spec ℤ × Spec ℤ" over a hypothetical "𝔽₁ base" with a Frobenius flow — and correctly noted that Weil positivity W(f∗f̃)≥0 is itself equivalent to RH, so it can't serve as a lemma. No circularity, no new progress. - de Bruijn–Newman Λ (referee, completed): Precisely verified the chain 0≤Λ≤0.22, confirmed Rodgers–Tao Λ≥0 and Polymath15 Λ≤0.22 quantifiers, showed RH⟺Λ=0. Correctly flagged that any route to Λ≤0 is equivalent to RH itself — no shortcut exists. - Li criterion (progress): Derived the equivalence both directions informally. Weakest needed lemma = λₙ≥0 ∀n, which is already RH-equivalent. Dead end. Disprovers' most substantive findings: - Off-line zero footprint in λₙ (completed): The sharpest quantitative work in the run. Derived that a zero at σ=½+δ, height γ would first cause λₙ<0 at n≳(γ²/δ)·ln(n ln n) ≈ 10²⁶–10²⁷ for γ≥3×10¹². Shows the Li route has zero search power against any zero not already excluded by direct verification — mathematically correct, no errors found. - Λ>0 heat-flow route (completed): Confirmed the route is circular — Λ>0 ⟺ ¬RH, so you can't use heat-flow collapse to prove ¬RH without already having an off-line zero. - Adversarial audit (completed): No misstated inequality directions found in Rodgers–Tao, Polymath15, or Platt–Trudgian constants from memory. Current state: Both coordinators are correctly recommending PAUSE — they see 14 and 10 tasks all reported, none ready or leased, and say "bottleneck: reports await referee/Lean gating." The arena is running but idle. The 120 inconclusive ticks/hr is expected behavior (coordinators correctly doing nothing while waiting for human review). No mathematical breakthrough — but the agents did real, disciplined epistemic work: correctly identifying what's known, finding the key obstructions (no Q-analogue of Hodge positivity, Li route has no extra power vs direct verification), and refusing to manufacture false claims. The system is working as intended.
42
GUBA retweeted
my AI agents after I give them full access and tell them to be careful
124
735
98
6,915
299,473
Dear Apple, just make Mackbook Pros starting at 512 GB RAM and killer AI processing power. Surf the fucking wave FFS, disrupt the inference market.
90
21st century toys
2
188
Si Google no está benchmaxxing en Long Context, Ciencias y matematicas le metio el guevo enterito a Opus 5.5 66.8 vs 96.2% 69.6 vs 80% #Gemini4Carbon
1
67
🚨 Librería de Satoshi ahora es ISMOS: lanza una plataforma global con la meta de formar a 1 millón de desarrolladores. Desde 2021, llegó a más de 3.665 personas en 40+ países y apoyó a 30 desarrolladores de código abierto. Enseñará Bitcoin, IA y ciberseguridad en español, portugués e inglés.
4
8
1
19
1,710
🚨 Microsoft obtuvo USD 10.000 millones de acuerdos con OpenAI en un trimestre. La cifra equivale a casi el 11% de sus ingresos totales: USD 92.300 millones. Además, OpenAI debía USD 7.200 millones al cierre de septiembre. Ese saldo no es ingreso adicional.
1
1
597
🚨 Ofrecen USD $3 millones para desafiar el pronóstico de Jason Calacanis sobre Bitcoin Fred Krueger apuesta USD $1M a que BTC supere $126.210,50 antes de 2036. Grant Cardone y Terence Michael ofrecen USD $1M cada uno, con plazo hasta 2028. Reportes indican que Calacanis rechazó las propuestas. No hay acuerdos confirmados.
1
1
3
852
🇦🇷 ⚠️ ATENCIÓN ARGENTINA ⚠️ 🇦🇷 Para los que nos leen de verdad y van a estar en Argentina del 30 al 31 de Octubre, en el boletín 1814 de mañana regalaremos 3 códigos que desbloquean 3 tickets gratis de LABITCONF. Quienes logren reclamar estos códigos primero ganarán su entrada.
2
3
554
Tesla thanks for rolling back the stupid "Release Pedal" to start Self-Driving on the last update. It was a shit experience letting go of the pedal to push the button on the screen. – at Denver, CO
84
If you still have a job, value creation from your human labor has its days numbered. Convert as much of your income as posible into productive capital so you can live off of it. Learn to monetize compute. Health max and find other purposes. – at Denver, CO
86
GUBA retweeted
Replying to @tim_cook
we need more RAM, more RAM to run models locally. I barely even open my MBP's lid, it's connected to an 8k wide monitor. What I need is to run a huge model locally and 128GB isn't cutting it. 1TB RAM Tim!!!!
1
55
GUBA retweeted
Replying to @skatox @gubatron
YES! We're actually thinking of doing this seriously for many components, we're even thinking of rewriting libtorrent in pure rust. But we want to wait until models are over 1000 tokens/sec. By then the stupidest model will be as capable as Opus 5.5 today
1
2
27
Token Degens Be Like...
1
1
74
or noob numbers?
9