Terrence

(MBA/AR-15/TX) "I'm a people person, damnit!"

[Terrence @tbonetmb]

Hey, what the hell happened to my Twitter egg?

[DogeDesigner @cb_doge]

Incredible footage of the booster from today’s launch!

[Sir Doge of the Coin ⚔️ @dogeofficialceo]

Super Heavy Booster 22 return 🚀

[Terrence @tbonetmb]

@MattKilby23 Not just SpaceX company; go to Isla Blanco park (north of the launch site), and you'll see a few thousand people, all watching in awe. I've seen those old photos of the Saturn V launches for the Apollo program; I feel like this is a throwback to that special age of Americana.

[Terrence @tbonetmb]

@Silver_Hornet @MikeBeauvais lol. Miss that guy.

[Everyday Astronaut @Erdayastronaut]

Starship in orbit visible from Starbase!!! Mind blown!!!

[Zack Golden @CSI_Starbase]

Jesus. This is the best iPhone tracking footage I’ve ever seen. The new from Mexico is ridiculous.

[Terrence @tbonetmb]

@AJamesMcCarthy So jealous

[Terrence @tbonetmb]

@Maximo8136151 @CosmicNomad44 lol - my thought too. SpaceX is probably functioning fine; but the entire US government runs on Excel and Outlook....

[Terrence @tbonetmb]

@CosmicNomad44 @CatboyRocketry Kinda wondering about that. I know they've gotten it 24 hours prior before. Anyone know anything about the hold-up?

[Terrence @tbonetmb]

@wholemars So jealous

[Terrence @tbonetmb]

@wholemars No rental cars available in Brownsville as of last night - at least from online booking.

[Terrence @tbonetmb]

@_MaxQ_ @NASASpaceflight f*cking awesome.

[Terrence @tbonetmb]

I'm so jealous of every one heading down to Starbase, TX.
I was hoping to go; but had an important work commitment yesterday that complicated scheduling.
Then was hoping to throw the wife and kids in the van and head down late yesterday, but the wife wasn't up for that adventure. Le sigh.

[Terrence @tbonetmb]

@LanceWoolie So jealous….

[Terrence @tbonetmb]

@battleangelviv Yup. Very envious.
I live about 600 miles away; but a work obligation and an angry wife stood in the way of me making the trek. Pretty bummed.

[바울이 @kaleumeleu37968]

여섯번째 생존법.

재난이나 전시 상황에 도시를 탈출(Bug Out)하려던 사람들의 80%는 목적지에 도착하기도 전에 길 위에서 죽습니다.

영화 속 주인공들은 차를 타고 유유히 교외로 빠져나가 시골 은신처에 도착하지만, 현실의 피난길은 가장 처참한 약탈과 조난의 무덤입니다.

이번 사례는 우크라이나와 러시아를 참고했습니다.

문명이 붕괴하는 순간, 도시를 벗어나기 위해 반드시 알아야 할 5대 피난·탈출 생존 철칙입니다.

1. 재난 터지자마자 차를 몰고 고속도로로 나가지 마라

재난 사이렌이 울리는 순간, 수백만 명이 동시에 차를 끌고 나옵니다. 도심 주요 간선도로와 고속도로는 1시간 안에 '움직이지 않는 거대한 철창'으로 변합니다.
- 도로가 막히면 유턴도, 후진도 불가능합니다.
- 히터를 틀다 연료가 바닥나면, 영하의 도로 한가운데에 버려진 금속 관 속에 갇히는 꼴이 됩니다.
- 차량 탈출은 재난 경보 후 '단 1~2시간 이내'에 도시 외곽을 완전히 통과할 수 있을 때만 유효합니다. 그 타이밍을 놓쳤다면 차라리 튼튼한 오토바이, 산악자전거, 또는 도보가 10배 안전합니다.

2. 밀리터리 전술 배낭을 메는 순간 당신은 1순위 표적이 된다

카키색, 카모플라주 무늬의 몰리(MOLLE) 전술 배낭에 전투화를 신고 나서는 것은, 약탈자들에게 "내 가방에 비싼 생존 장비와 비상식량이 가득하다"고 광고하는 자살행위입니다.
- 극한의 혼란 속에서는 누구의 눈에도 띄지 않는 '회색 인간(Gray Man)'이 되어야 합니다.
- 낡고 평범한 학생용 백팩, 허름한 스포츠 가방, 때 묻은 비닐 쇼핑백에 장비를 나누어 담으세요.
- 겉모습은 누구보다 가난하고 절박해 보여야 약탈자의 표적에서 벗어납니다.

3. '시골 친척 집'이라는 막연한 탈출은 노숙자로 가는 지름길이다

미리 연료, 식량, 방어 수단을 완벽하게 비축해 둔 '검증된 베이스캠프'가 없다면 무작정 도시를 떠나지 마세요.
- 평소 연락도 없던 시골 친척이나 지인의 집은 당신의 피난처가 되어주지 않습니다. 그곳 사람들도 이미 자원 고갈과 침입자 공포로 극도로 예민해져 있습니다.
- 연고도 없는 시골로 향하는 것은 스스로 자원이 고갈된 채 길거리 노숙자가 되겠다는 선언입니다.
- 확실한 목적지와 도달 경로, 예비 루트(Plan B)가 없다면 무리하게 길을 나서지 마세요.

4. 가장 강력한 탈출 무기는 총이 아니라 '신발과 양말'이다

피난길에서 당신을 살리는 것은 고가의 장비가 아니라 당신의 '발'입니다. 도로가 마비되면 결국 수십 킬로미터를 걸어야 합니다.
- 평소 신지 않던 새 등산화나 군화를 신으면 5km도 못 가 발뒤꿈치가 까지고 물집이 터져 그 자리에 주저앉게 됩니다.
- 완전히 발에 길들여진 편안한 트레킹화나 러닝화를 신으세요.
- 그리고 반드시 '여분의 울(양모) 양말 3켤레'와 물집 방지용 바셀린, 테이프를 챙기세요. 발이 망가지는 순간 당신은 이동력을 잃고 그 자리에서 낙오자가 됩니다.

5. 골든타임 2시간을 놓쳤다면, 탈출을 포기하고 실내를 요새화하라

재난 발생 직후 2시간이 지나면 도로는 패닉에 빠진 군중, 버려진 차량, 경찰의 바리케이드로 아수라장이 됩니다. 이때 뒤늦게 짐을 싸서 거리로 뛰쳐나가는 것은 무법지대 한복판에 맨몸으로 뛰어드는 짓입니다.
- 초반 골든타임을 놓쳤다면, 무리한 탈출을 포기하고 즉시 집 안으로 후퇴(Bug In)하세요.
- 문과 창문을 굳게 걸어 잠그고, 물을 최대한 받아두며, 초기 72시간의 대혼란과 군중 패닉이 가라앉을 때까지 조용히 숨죽이고 버텨야 합니다.
- 탈출은 도로의 혼란이 정리되고 군경의 통제가 시작되거나, 혼란이 진정된 이후에 시도해도 늦지 않습니다.

최고의 탈출은 영웅처럼 길을 뚫는 것이 아니라, 군중의 패닉에 휩쓸리지 않는 냉정함입니다. 가족의 안전을 위해 이 글을 북마크해 두세요.

[Valerie Stivers @valerie_reads]

For me, having a baby was the beginning of the end of my marriage. My ex and I had set things up to be fairly trad. He made the money, and I planned to stay home with the children. I didn't, however, realize that his vision of that was actual-zero childcare, to the best of his ability. He spent most weekends watching TV in another room with the door closed, while I ran around to parks and playgrounds, etc. Eventually we had no shared life.

@jmrphy for @unherd recommends this. I am not so sure.

Nor am I totally unsure. I was a stay-at-home-mom, but also had fairly feminist expectations, and I thought that my ex *should* have walked in the door from work and started helping me make toddler-dinner. After all, we'd both worked all day, and his work was a lot more fun. More than that, I wanted someone to share the experience with ... him.

We also did have other, deeper problems. But mainly what happened is that I started resenting him and stopped caring about his wants and needs in the relationship (because he was a lazy-ass bum sitting there waiting for his adult-dinner, too). And I devoted my considerable powers to trying to force him to participate, which he did, but so resentfully in his own part that it just wasn't worth it. eventually I started arranging things to discourage his participation. I can think of summer mornings where I was scuttling around trying to get me and the kids out of the house before he woke up....

Also I tend to forget that for the first few years, I tried to treat him like the luxury-king-of-finance that he wanted to be, and *tried* to do it all myself without inconveniencing him, and I just wasn't able to pull that off, logistically or emotionally.

I think we were probably just doomed for other reasons, but, with great hindsight, my resentment and feminism didn't help us. And his view of what a father was supposed to do was pretty impoverished. It's possible that if both of us had moved a few clicks towards the other's position, we could have made it. Or if we'd even just done the things we were already doing without the resentment, perhaps we could have grown together into the job.

https://t.co/BAP5YhIh4M

[Terrence @tbonetmb]

@MrSeanGoodall @niccruzpatane @SpaceX Yeah; there’s some epic BBQ in that stretch near San Marcos and New Braunfels.

[Terrence @tbonetmb]

@niccruzpatane @SpaceX Wish I could be down there too. Between work obligations and angry wife mood swings, it wasn’t in the cards for me.

[will depue @willdepue]

I have a major result to share.

I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China.

This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model.

Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves.

I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems.

I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used.

It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20.

These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs.

I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at https://t.co/YOtzND0Jqk.

There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.