Doks

Contestonomics - the power of Bittensor

If you are under 30 do reverse barbell: 90% high-risk bets 10% ultra-safe @nntaleb

Trump securing the Iran deal

@MaxScore 🎯

eval design will separate the top 1% of ai builders from everyone else

The ultimate moats are grit and agency

he decribes Tesla's moat

pareto x retardmaxxing a system for structuring your life and time 20% → active ownership: think deep, plan smart, lock in hard on the few moves that actually matter 80% → retardmaxxing: drop perfection, get cool with ugly messy shit, and just hammer away with fast steady moves every single day Result: low stress, high output active ownership on what matters retardmaxxing where it doesn’t

@frihetspenger bull

Targon er, etter min mening, *det* mest spennende subnettet i Bittensor-økosystemet Markedet for compute i dag, er i all hovedsak kontrollert av de tre hyperscalerne Amazon, Google og Microsoft, som investerer i, og eier datasentre og maskinvare selv Dette er selvsagt vanvittig kapitalkrevende, og koster dem flere hundre milliarder $ årlig i capex Det er en bransje som på mange måter ligner mye på taxibransjen forut for Uber, i den forstand at den er sentralisert, kapitalintensiv og de facto monopolisert Samtidig finnes det millioner av GPU-er verden over som i stor grad sitter ubrukt, enten det er i nevnte datasentre, gaming PC-er, universiteter, småbedrifter, etc Jeg ble sjokkert da jeg fant ut at det er estimert at over *halvparten* av GPU-er globalt til enhver tid sitter ubrukt Potensialet for å bygge et marked rundt denne kapasiteten er med andre ord enormt, og kommer bare til å øke i takt med akselerasjonen innen AI og dermed behovet for stadig mer compute Targon gjør i så henseende for compute det Uber gjorde for transport; de bygger en markedsplass som kobler sammen de som har overskuddskapasitet med de som trenger compute Med TargonOS kan nå hvem som helst med en GPU koble seg på denne markedsplassen, som er fullstendig desentralisert og tillatelsesløs, og leie ut compute Og i motsetning til en aktør som feks Lium, som også driver med GPU-leie, har de også to lag med sikkerhet i Intel TDX (Intel co-signet for øvrig et whitepaper med Targon nylig) og Nvidia Confidential Compute (Targon er også innlemmet i Inception-programmet til Nvidia) Det gjør det, forhåpentligvis, trygt nok til at seriøse aktører kan legge arbeidslaster der etter hvert som tilbudssiden blir bygget ut, og introduksjonen av TargonOS er i så måte et stort steg i den retningen

@rob_svrn @Jason @markjeffrey @frihetspenger @const_reborn

karpathy’s autoresearch loop specify problem → evaluate on metric → iterate is the fundamental system in which Bittensor subnets operate

If you put David Goggins, Richard Feynman and Napoleon together you get @elonmusk. "He's David Goggins' level intensity, he's Richard Feynman's level of unconventional technical brilliance, and he's Napoleon's strategic genius and insane bias to action. Those three traits combined make him just incredibly singular. Even if you put in this amount of time trying to know your business from A to Z, without that deep technical intuition or sense of the fundamentals of the physics and the materials, you couldn't make the calls that he's making. The founder is the guardian of the company's soul."

Hey @MaxScore In Manako, when a regular user simply uploads a video or live feed: Will every video still be split and sent as tasks to the decentralized miners on Subnet 44 for the heavy computer vision work? Or will the AI agent layer eventually handle the analysis automatically?

Our data shows that 50% of @Tesla FSD + @Lemonade_Inc users let it drive autonomously 90-100% of the time. Impressive, @elonmusk!

@alc2022 Bittensor $TAO

What company should I do a deep dive on next?

"the market is oversold"

When AI connects with robotics, everything changes. Imagine this: You ask Claude to create a prototype of a product. That request goes straight to a factory filled with robots. The robots understand exactly what to build because they have AI brains. They produce it automatically. Then it gets shipped to you. Maybe even by drone. In theory, you could go from idea to a real product in one day. And over time, this will become cheap. This leads to a new model. Autonomous AI factories that build things on demand. No middlemen. No long back and forth. No confusion. Just a simple flow. Prompt → LLM → Factory → Delivery

I have actively tried to counter my short-term bearish bias on AI productivity gains. But I keep arriving at the same conclusion: AI productivity gains won't materialize short-term for the majority of companies. The gains are real, but they are happening at the individual level: Startups. Fast-moving tech companies. People already wired for experimentation. For established companies with legacy workflows? The technology is still too raw. We are in a phase where new models and features drop daily. Information overload is massive. On a personal level, we have not even settled on the most basic use cases yet. Companies are 10x harder. People say the AI timeline is different. I agree, but only to a degree. I believe Amara's Law still holds: "We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run." The cash flows will come. The productivity gains will come at scale. But the timeline priced into markets right now feels too aggressive. On a technical level, both Nasdaq and S&P 500 look weak. Consolidating at all-time highs. We all know the pattern: contraction → expansion. I believe the next expansion is to the downside. That said, if markets start falling, I will be scaling in heavily. The thesis is not bearish on AI. It is bearish on the timeline the market is pricing in.

@metanova_labs @alc2022

ArboNOVA: Patent–Molecule matching loop We’ve been experimenting with an agent that maps molecules → prior art using only open data + tools Benchmark: ~1500 molecules across ADHD-related patents (since 2012) In ~12 hours: 18 iterations of the loop → Best hit rate: 85.4% How this is usually done: Pharma intelligence teams + expensive proprietary databases + manual workflows + even conference attendance Early, but promising. Moving one step closer toward automating drug discovery and identifying which molecules are most strategic to advance in the wet lab. Based on @const_reborn (github.com/unconst/Arbos) and @karpathy autoresearch framework #Bittensor #SN68 #ralphloop #agents #DrugDiscovery #Desci #DeAI

Now, there is a website to translate from English to LinkedIn language. 😂🙊