NUKΞ

SITUATION DETECTED: Bloomington, Indiana's city council voted unanimously this week to ban any recursively self-improving entity capable of greatly outperforming all humans on all cognitive tasks from inside city limits, citing existential hazards and loss of human control.

China cracked recursive self-improvement 🤯 Tencent dropped 27B model that rewrites, retrains, and refines itself.. with zero human in the loop. It’s called "Terminal-RSI" and it solves one of the biggest problems with AI. They are incredibly lazy. Usually, an agent will make partial progress, claim it succeeded, and completely fail the final environment verifier check. this new pipeline forces agents to actually finish the job. here’s how it works: they built a recursive loop with two alternating phases.. - RSR (recursive self-rewrite): the model fine-tunes itself on its own successfully rewritten trajectories - RPR (recursive policy-refine): it then runs reinforcement learning (ppo) using a "critic" inherited directly from the previous stage to lock in those discovered behaviors - Loop: the training sequence leapfrogs itself: Base → RSR-1 → RPR-1 → RSR-2 → RPR-2 and the numbers are insane.. they applied this to a qwen3.8-27b base model. across the four "terminal-bench" suites, the agent's average success rate jumped from 20.68% all the way to 50.58%. on the absolute hardest benchmark suite (terminal-bench 4.0), the base model scored a flat 0.00%. after terminal-rsi? it hit 19.84% (putting it right on par with reference models like gemini 3.8 flash). the secret sauce? they fixed the reward system. instead of giving the ai "dense rewards" for just trying or passing semantic checks, they switched to a "success-prioritized reward" (spr). it mathematically separates strict success from partial credit. basically, the agent only gets the massive reward payout if the code actually passes the final terminal gate.. so it stops faking success. the gap between partial local repairs and strict terminal verification is closing fast. we are watching agents build their own training environments and level up autonomously.. the future is wild.

I know this is a predictable opinion from me but I don't agree with the strong emotional reaction against the anti-data-center movements that I've seen this week. As I see it, algorithmic progress means that whatever you would be able to do with more and bigger data centers in 2 years, you will be able to do with the data centers we have today in 4 years - probably sooner. So nothing remotely apocalyptically bad will happen from even extreme levels of success by anti-data-center people. Meanwhile we need all the time we can get to better harden the world against both already-visible cybersecurity issues and any bio and other risks soon to come. Science applications are bottlenecked by many things other than AI, so slowing AI down by one year will slow those down by much less than one year. Given this, a slowdown in compute hyper-scaling seems quite net good? Notice this is without any controversial "doomer arguments", RSI etc. Even if net-bad, it's certainly not worse than like 10+ other political things.

SpaceX has officially announced that it is currently targeting November 13th at 5:45 PM ET for Starship's 15th flight. Flight info: "This flight is planned to be the first to return both Starship and Super Heavy to the launch site for catch. If successful, this will mark the first full recovery of an orbital launch system and a breakthrough moment in making Starship a reliable, rapidly reusable rocket." • Flight will last more than 22 hours, completing approximately 14 orbits around Earth. • Starship will orbit at an altitude of approximately 275 km (171 miles). • First attempt to recover both Super Heavy and Starship at the launch site during the same mission. • First return-to-launch-site and catch attempt for a Super Heavy V3 booster. • Super Heavy will execute launch, ascent, stage separation, boostback burn, and return to Starbase for a tower catch. • Starship will deploy 26 Starlink V3 satellites into orbit before attempting its first return to Starbase and tower catch. • The 26 satellites will provide capacity equivalent to approximately 10 Falcon 9 launches carrying Starlink V2 Mini satellites. • Satellites will deploy antennas and solar arrays, establish radio and laser communications, and raise their orbits using onboard thrusters. • Satellites could begin serving customers within weeks of launch following successful on-orbit checkouts. • Three Starlink V3 satellites will carry cameras to inspect Starship's heatshield and transmit imagery to ground operators. • Starship's heatshield will feature improved tile retention, additional protection against plasma intrusion, and expanded use of curved tiles to reduce heating between tiles. • Starship will only attempt its return and catch if critical reentry, landing burn, and tower approach systems are confirmed healthy. • Multiple predetermined ocean deorbit locations and a passively safe suborbital trajectory provide contingency options. • A successful mission would mark the first complete recovery of an orbital launch system, a major milestone toward rapid reusability. More info: spacex.com/launches/starship…

No data centers on public lands. Period.

Hollywood Has Flipped on GenAI, Says 'Avatar' VFX House Weta FX Chief: 'If You Don’t Have an AI Strategy, Don’t Bother' variety.com/2026/film/market…

I am dying 😂😂😂☠️ Care to dance? 🤣✌️

US FAA says it will phase out floppy disks from air traffic control by 2028 reut.rs/4zsSsu2

It troubles me greatly that we do not yet see any sign of aliens

Dear Texas, Farewell, old friend. I’m sorry the blue wave swept across your once-great state. I guess all those Californians moving in didn’t work out too well after all. 🤷‍♂️ Oklahoma is open for business and still Ruby Red! We’re ready to welcome new companies and create more jobs for AMERICANS 🇺🇸 . We believe in low taxes, stable and limited regulation, and personal freedom. No socialism here, just good old-fashioned capitalism and prosperity. Our 39 federally recognized tribes are open for business as well, and ready to partner with the state and with you. I’m Oklahoma’s Lieutenant Governor-elect, and my DMs are open. Come join us! @elonmusk, call me. 📞

The governor-elect of Texas, Gina Hinojosa, has explicitly said that all currently permitted datacenters in Texas will be allowed to finish construction. This essentially covers all of the 2027 and likely well into 2028. She is going to pass laws that ensure that all *newly* permitted datacenters going forward for delivery in 2028/2029 will pay taxes, commit to no increase in power prices for consumers, don’t impact water availability for people, be constructed with union labor and give “Texans a seat at the table.” Sounds reasonable. She doesn’t want to lose a lot of good blue collar jobs and property taxes for small towns in rural Texas. Datacenter commentary starts at 6:30 in this interview. Also attaching her official campaign position from her website in the next post. Sidenote: After watching a bunch of her speeches and interviews, I do think she is a political talent and probably has a future in national politics. youtu.be/D-T6H3b8ZeE?is=mEWu…

SpaceX just confirmed a 6 GW supercompute campus in Arizona. That's almost half the 13 GW they guided for the end of 2027, now on a map. Do you understand how dumb the FUD is now? 747 acres in Cochise County, next to the Apache Generating Station. Hearing is 9 December. 6 GW is the full build-out, across 4 phases. Phase 1 is 1.5 GW. This is not another Brownsville. Different state. Different permit chain. The Texas freeze does not touch it. They designed it for grid power from the plant next door, through SSVEC and AEPCO. Gas is only backup. Not the own-turbine Colossus model. Dry-cooled. About 7 to 10 acre-feet a year at full build-out. The filing puts that at roughly 20 to 30 Arizona homes. No new pumping from the Willcox basin. $SPCX

America hates building things here. We don’t want the noise, we don’t want the pollution, we don’t want it in our way. Factories, data centers, whatever. That’s why we started just outsourcing production to other countries. It was much easier than dealing with the local environmental protesters. Now, as we build superintelligence, we have to ask ourselves if that’s going to be something we build abroad or if it’s going to be something we keep under our control. I think it would be a bad idea to put something so powerful outside US jurisdiction. We should be self-sufficient and build the factories and data centers we need here in the United States. Building here might be unpopular, but we have to find a way to do it. Maybe space is the answer. Maybe we find a way to share some of the earnings of the data centers with locals so they benefit too. I don’t know what the solution is, but we need to build data centers and we need them under our control.

Red will win the markets tomorrow