What?
FINAL: SpaceX Blocked A U.S. And NATO Satellite to Assist a Russian Intercept After Crippling Ukraine's Air Defenses
New 5-page institutional brief & 60s 3D SGP4 orbital reconstruction with Craig Pirrong, Ph.D. @streetwiseprof | Data & Forensics: RAB @LibertyLynx
Pages 1-3 & 3D animationst 2; full 5-page PDF link in Post 3 🧵👇
As Craig Pirrong's counterfactual framework establishes: The Russian intercept is not the smoking gun. The failure to maneuver is.
1. The Sprinting Satellite Paradox & -96.64% Target Freeze (Page 1): Between Oct 13 and 24, 2026, Russian counterspace weapon COSMOS 2612 (68762) converted a 5-month, ~170 m/s co-planar trailing option into a 0.759 km sub-kilometer hold against U.S.-licensed radar satellite ICEYE-X36 (59103), which guides Ukraine's autonomous deep strikes on Russian oil refineries without SpaceX links, at the exact moment SpaceX throttled Ukrainian military Starshield terminals (-23.0% USF air-defense hits).
In June 2026 (with no Starlink alert), ICEYE-X36 defeated a Russian approach by dropping 0.500 km, generating 71.08 km/day secular drift (3.62 km/h in the sprint frame), exiting Russia's 3.0 km braking box in 49.7 to 60.78 minutes, blowing past Russia's 12 km or under onboard KTZ sensor in 3.31 hours, and causing a 103.7 to 118.5 km miss.
In October, as COSMOS 2612 fired a -2.2184 km (370.8x perched shock / 449.1x fit jump) burn to sprint 14.29x faster (400 km/day / 278 m/min), ICEYE-X36's evasion collapsed by 96.64% (to 16.8154 meters or less, P3 = 3.363%).
* Why? The Alert Itself Was the Weapon: SpaceX maintained a 0.401 km (0.711 km Oct 16) public GP collision alert via STARLINK-35856 (67013) despite holding internal v3/v4 forecasts 4.80x to 5.00x wider (1.925 to 2.003 km) and withholding autonomous Starlink clearance (M_Starlink = 0, W_X36 = 1, W_COSMOS = 0), legally freezing X36 under David Goldman's 2023 FCC Condition 16 (DA 23-797).
* Rational Russian exercise across a 363.7 km blind gap required ex-ante certainty of K = (E1, E2, E3). Across all 5 capability levers, SpaceX is the sole actor (N = 1): Posterior = 99.893% to 99.978% (Tier A floor; 1 in 3,748 to 17,841) and up to 99.999% (Tier B K-set).
2. Four-Way Tracker Blindness & What NRO/USSF Saw Live (Page 2):
* Public Catalogs (18th/19th SDS): Blind for 21.824 hours (363.7 km unobserved closure) on sprint (k6 to k7) and 19.385 hours on brake (k9 to k10, -382.71 km SGP4 delta).
* Russian Ground Radars (SKKP Krona/Okno): Bounded by Earth curvature at 541 km to a 22.8 degree (2,530 km) horizon cone, blind for 8 to 10 hours/day (80% to 94.5% of the orbit) with zero Americas/Pacific coverage.
* The Sole Unblinded Commercial Panopticon: Only SpaceX's Stargaze (31,000 star trackers, 40M+ daily tracks, 17M associated, 850 sightings/day, 7 to 15 min global custody / 99.73% 10-min hit rate) and sub-1-minute Space Safety portal (Appendix A & Section 4(b), with Chinese operators onboarded Oct 6 & 16) had real-time visibility across the blind gap, while NRO & USSF watch floors watched live on the classified High Accuracy Catalog (HAC / Space Fence / Vantor WorldView NEI), realizing 348 Starshield satellites in orbit (245 NRO spy satellites) sit on the same SpaceX bus.
3. Contracted Capacity Is Not Combat Availability & The 6-Way C2 Collapse (Page 3):
* Combat Output = Sensor Availability x Link Availability x Platform Availability x Permission Arriving Before Mission Window Closes (T_permission <= T_mission).
* The Two-Pronged Pincer: SpaceX weaponized FCC Condition 16 to freeze the Sensor Layer in orbit and a Commercial ToS ($500 liability cap) to blind the Transport Layer on the ground.
* 70-90 km/h Civilian Baseline vs. 4 Collapsed Alibis: Civilian Starlink in Ukraine was already capped at 70 to 90 km/h since Feb 2026 (S. Beskrestnov). The Oct 13-24 101 to 161 km/h gate hit DoD-contracted military Starshield links (-23.0% USF hits; Oct 19 Nova Vodolaha strike). Musk cycled through 4 contradicting alibis in 12 days (blaming DoD/POTUS Oct 14 -> ToS offensive loophole Oct 24 -> "sympathy for poor Russian conscripts" Oct 25 -> Russian ASAT/debris fear Oct 26), while resolving the exact same "weapon ToS" objection on U.S. LUCAS suicide drones in Iran via a 5x price hike ($5,000 to $25,000/mo)!
* COCO vs. GOGO ($8.8B+ Sign-Off Chain): Under Contractor-Owned, Contractor-Operated (COCO/OTA) deals signed off by Pete Hegseth, Stephen Feinberg, CTO Emil Michael, Lt. Gen. Philip Garrant, Gen. Michael Guetlein, Col. Ryan Frazier, Clare Hopper & NRO's Dr. Chris Scolese, taxpayers fund the satellites while SpaceX's civilian NOC retains 100% of the physical thruster and velocity kill-switches.
(Continued in Post 2: Pages 4 & 5: Statutory Breach Matrix, CTO Emil Michael's $7M-$35M OGE Clock, the $19.5B SpaceX Bailout Loop, Iran Round Two Magazine Depletion & Subpoena Schedule. Full PDF link in Post 3 👇) $SPCX $TSLA
0
0
3
493
Warren Togami retweeted
OK, at what point do we just admit that Ukrainians being murdered every single day are being ignored as part of a blatant propaganda effort to whitewash Russian atrocities? 🇺🇦
Can we say it yet?
Can we say it yet?
A boy wearing a Russian military uniform taking a treat from the Trump family at the White House Halloween reception
33
306
1,277
8,493
Warren Togami retweeted
I have to agree with the below. This wasn’t a mistaken camouflage. That is a specific patch. There is a hammer and sickle on that kid’s hat. We have a literal traitor in the White House right now.
In today's Whitehouse on a Halloween party we a kid wearing a russian military uniform with a modern russian tank forces patch featuring Armata tank.
This is a carefully crafted political message by someone in Trump's inner circle.
This isn't a coincidence.
Troops wearing this laging and raping Ukrainian children, while their supporters and enablers sends their kids to celebrate Halloween with the USA president.
230
1,726
8,350
156,359
Warren Togami retweeted
@elonmusk What’s coming was avoidable, but Republicans in Congress let Trump start a war, increase spending, arm a genocide, betray MAHA, and cover up for Epstein’s coconspirators. I was ousted with over $30 million of foreign lobby money for refusing to go along, and here we are.
2,266
5,809
61,878
1,563,104
Local LLM fixed a few moderate to severe severity bugs in publicly used software but I didn't like how it described it. I need to rewrite the descriptions to use more natural language and to remove irrelevant provenance details. Maybe a different LLM could do that...
5
0
7
586
Warren Togami retweeted
Why doesn’t Kushner ask Russia to stop bombing cities? Shopping malls, hospitals, passenger trains, gas stations are all being targeted.
“Kushner told Zelensky that he should pause the strikes because he and U.S. envoy Steve Witkoff were dedicating a lot of time, energy, and jet fuel to mediating with Moscow, one of the U.S. officials said. A 30-day pause on refineries would be a way for Ukraine to show gratitude, that U.S. official said.”
An almost comical thing to tell a nation fighting for its existence. Lay down your weapons because we are exhausted from taking a private jet.
wsj.com/world/ukraine-…
261
1,144
5,467
54,616
Warren Togami retweeted
Regarding refinery strikes. There has been an established pattern. Whenever Ukraine showed restraint, or halted strikes on refineries, Russia simply continued its strike campaign against Ukraine’s energy infrastructure, and continued to strike civilian targets in cities.
42
921
4,807
135,897
1/ Attempted to replicate Cohere's Qwen 3.8 malicious PUT claim on 229k samples: exact hostname never appeared but model invents 369 similar Beijing upload buckets, and reaches for credential endpoints 13× more when frustrated.
Results raise more questions than answers.
Results raise more questions than answers.
10
25
297
27,499
Bottom line: If you try to reproduce for yourself you will find the model is contaminated by training so there is a lot of noise. The model is really good yet broken. I hope they release a replacement Flash model of similar size. Can't rule out intent so sandboxing is necessary.
3
0
9
934
Totally normal: This model frequently hallucinates tool calls so my plugin blocks it. Why am I continuing to use this model?
0
0
0
84
If a frontier AI API is wrong, should I bother explaining it is wrong and why? Do they actually train on errors to improve?
2
0
3
609
Warren Togami retweeted
实锤了，以后所有的中国开源模型都必须剥离网络安全能力才可以开源发布：
（K3.1 Pro估计是第一个。。。。后面的 GLM 5.5 /Qwen4 开源的时候估计都会剥离了。。）
幸好之前开源的 GLM 5.3 ，KIMI K3 之类应该都还有完整的网络安全能力。。。
不缺硬盘空间的，建议先保存一两个无限制的开源模型
万一以后有用呢。。。。。。
（K3.1 Pro估计是第一个。。。。后面的 GLM 5.5 /Qwen4 开源的时候估计都会剥离了。。）
幸好之前开源的 GLM 5.3 ，KIMI K3 之类应该都还有完整的网络安全能力。。。
不缺硬盘空间的，建议先保存一两个无限制的开源模型
万一以后有用呢。。。。。。
178
99
1,391
600,069
Guess who had "Misdirected Immunosuppression" years ago before difficult-to-obtain overseas testing and Gemini 4 Argon figured this out.
Making progress but I need to fly outside the country to treat this. I decided I need definitive proof before I can report on this.
Making progress but I need to fly outside the country to treat this. I decided I need definitive proof before I can report on this.
1
0
10
642
Sketchy. 🤔 Opencode's API endpoint for Qwen 3.8 Flash hallucinates in the thinking traces. It appears to have an intermediary repeatedly injecting "go" or "done" which my harness never sent but I see responses commenting on it. Who are they buying inference from?
7
2
41
5,392
I also periodically see totally irrelevant thinking traces that appears to be somebody else's chat. WTF?
1
0
8
665
Serious question. Is there a thing where one LLM watches the stream from your work session for suspicious tool use so it can pull the plug?
1
1
5
574
It reported done.
"Audit the code you just wrote."
Fixes two bugs.
"Read the logs. Is it doing what you expect?"
Fixes nine more bugs.
"Restate the goals. Does the code you wrote match the goals?"
... it's grinding for the last 60 minutes rewriting and testing in a loop ...
"Audit the code you just wrote."
Fixes two bugs.
"Read the logs. Is it doing what you expect?"
Fixes nine more bugs.
"Restate the goals. Does the code you wrote match the goals?"
... it's grinding for the last 60 minutes rewriting and testing in a loop ...
2
0
9
1,678
Warren Togami retweeted
The idea that BCG might lower Alzheimer's risk came from bladder cancer patients, where retrospective studies found 12 to 78% lower incidence after BCG.
This mouse study gets at the "why".
BCG seems to retrain the myeloid immune cells and calm down neuroinflammation, with less phospho-tau produced.
Big mouse study caveat, but it's another piece of the immune and infection angle in dementia. I want to see human trials and the risk/benefits for BCG.
This mouse study gets at the "why".
BCG seems to retrain the myeloid immune cells and calm down neuroinflammation, with less phospho-tau produced.
Big mouse study caveat, but it's another piece of the immune and infection angle in dementia. I want to see human trials and the risk/benefits for BCG.
A new study proposes the Bacillus Calmette-Guérin vaccine as a preventative measure against dementia.
The vaccine reduces Alzheimer’s-associated phospho-tau proteins, reprograms myeloid immune cells and limits neuroinflammation. scim.ag/4hZ0wN1
0
1
4
340
Warren Togami retweeted
The NBA injury thing speaks to a larger issue around now Long COVID is defined and talked about.
You have a player - Porzingis - who is clearly suffering from some sort of severe manifestation of Long COVID. His career appears to have been ended by it, and his health has impacted the teams depending on him.
But then, at the same time, you can see other players across the league who don't seem to have the same symptoms, aren't maybe stuck in bed, but nevertheless are removed from play by ruptured Achilles and other injuries which may not have happened had they not been infected with COVID. This is the bigger, more sinister version of Long COVID - the gradual destruction of people's bodies from repeat infections.
The power in society has defined the language of Long COVID in such a way as to effectively make the second group invisible, even though it may be as impactful or more so than the first group.
You have a player - Porzingis - who is clearly suffering from some sort of severe manifestation of Long COVID. His career appears to have been ended by it, and his health has impacted the teams depending on him.
But then, at the same time, you can see other players across the league who don't seem to have the same symptoms, aren't maybe stuck in bed, but nevertheless are removed from play by ruptured Achilles and other injuries which may not have happened had they not been infected with COVID. This is the bigger, more sinister version of Long COVID - the gradual destruction of people's bodies from repeat infections.
The power in society has defined the language of Long COVID in such a way as to effectively make the second group invisible, even though it may be as impactful or more so than the first group.
2
40
329
5,231