Good morning! I've just published the first article on the AUTOWAR25 website.
The Ocean as the Proving Ground for the Next Architecture of Command - Machine Warfare, the Three Tyrannies, and the Limits of Continuous Human Control
"The resulting challenge is not simply how to accelerate human decisions, but how to preserve command authority when human judgment cannot accompany every operational decision."
We are not running out of ways to make machines more capable. We are running out of ways to keep the human continuously present.
The ocean may be where we learn to keep command authority without continuous human control.
(10-minute read) autowar25.com/articles/the-o…
What's remarkable about this video of a Ukrainian long-range drone strike on Russian infrastructure is the slow plodding pace of the drone. It indicates an air defense posture that is exhausted or just incapable over such a large area – likely a mixture of both.
If I'm sitting on the first island chain against China, I'm building tens of thousands of these things and widely distributing them in ready storage.
Could China defend against 10, 20, 50,000 of these, delivering 1-5 million pounds of HE, all launched at coastal infrastructure in a matter of days? Does China take that bet?
x.com/Osinttechnical/status/…
Ukrainian attack drones hit a major Russian oil depot in Rybinsk this morning, setting multiple tanks ablaze.
Seen here, a Ukrainian attack drone flies towards the already burning Russian depot, descends, and then smashes into another oil tank.
If it happens, what will first contact be like? Voyager is interesting.
A dying metal object, launched decades ago, now a light-day out and still going. Most people picture first contact as a signal or a ship. But the numbers say something quieter is more likely: one of our old artifacts, or someone else’s, drifting into the path of another. No power, no radio, just a piece of hardware that has been moving for a very long time.
That sounds straightforward until you ask what would actually have to be true for anyone to notice.
First, the thing has to still be recognizable as made. After a few hundred thousand years the plutonium is long gone. The electronics are gone. What’s left is a cold lump of metal and composite that has been sandblasted by dust and cracked by radiation. A big dish or a long boom might still stick out, but the details that say “someone built this” are small and they wear away. At the speeds these objects move—tens of kilometres a second—you don’t get a second look. If you haven’t already spent a long time cataloguing every ordinary rock and chunk of ice on a similar path, that worn-out probe just looks like another dark irregular body.
Second, someone has to be in a position to see it. At interstellar distances even a large structure is faint. You need either a lucky pass by another probe that is still working, or a survey that has been watching the same volume of space for a long time. A one-time detection at closest approach doesn’t give you time to turn around. Relative speed closes the window in hours or days.
Third, the people (or machines) who find it have to still know what “alien” would look like. After a million years the baseline of normal interstellar junk has changed. Better instruments, more objects, tighter models. The same leftover probe can be obviously artificial under one set of data and just another outlier under a later set. There is no permanent label on the object that survives the trip.
None of this makes the idea impossible. It just means the interesting (and important) part is not the moment of contact. It is everything that has to be in place beforehand: a long, continuous record of ordinary objects so that anything that doesn’t fit stands out clearly, and a way of keeping those records from being rewritten every time someone has a new theory. Voyager itself is already a test case. In another forty thousand years it will pass a few light-years from a small star. By then it will be a silent piece of debris. Whether anyone can still tell it was made depends less on the probe and more on how carefully the ordinary background has been mapped in the meantime.
General observation about using Grok Bot — running tasks uses almost no tokens. Editing code is a massive use of tokens.
The amount of work that you perceive has nothing to do with the actual work being done.
I know… duh. I just didn’t have a measure before this month.
AUTOWAR25 No. 35A · Morning Edition, Wednesday, Nov. 18. The top 25 autonomous-warfare stories, ranked. 🧵
The Air Force starts building training for its new point-defense career field to protect bases from drones. Germany and Ukraine set up a joint venture to build ground robots in Germany, and Denmark gets access to Ukraine's battlefield-AI dataset. Links in every post below.
Such a good idea.
One-by-one, a bunch of Type-1 drones are flown in and distributed throughout a large structure. When they get enough, they are all command detonated in the same moment.
You really never know what they’ll come up with.
Ukrainian FPV drones gather in a group inside a high-rise building occupied by Russian infantry in Kostiantynivka, after which they are remotely detonated, bringing down part of the building and burying the occupiers under the rubble.
🚨Texas: “O Allah, deal with the oppressive aggressors. O Allah, take revenge on them. Send down Your wrath upon them and unleash upon them the sword of Your vengeance.
Count them by number, kill them one after another, and do not leave a single one of them.”
Uncovered by @Theswampmonitor
John Wick, Jason Bourne, James Bond, and Rambo aren’t available. You’ve been kidnapped and you get one movie character to come rescue you. No superpowers. Who are you trusting?
John Wick, Jason Bourne, James Bond, and Rambo aren’t available. You’ve been kidnapped and you get one movie character to come rescue you. No superpowers. Who are you trusting?
Anyone seen @JoshuaSteinman? Does he need oxygen?
His business is industrial cyber security. I’ve got to assume that the customers are coming out of the woodwork right now…
Investors too.
Make your mark sir.
This is your time.
My July idea still holds: 25,000 medium/long-range one-way attack UAVs (about $2 billion total, roughly $80,000 each) pre-positioned in hardened, dispersed storage across Taiwan, Japan, the Philippines, and South Korea would give the First Island Chain a scalable dual-role capability similar to Ukraine’s long-range infrastructure strikes.
Details:
The same class of airframe that has already demonstrated multi-thousand-km strikes on fixed energy and logistics targets can be configured for anti-ship (seeker + warhead) or anti-infrastructure (larger payload).
Distributed hardened storage next to minimal asphalt or road sections for takeoff, maintained by local reserve or guard forces, reduces reliance on vulnerable airfields and allows rapid local generation.
At this cost and scale the inventory is expendable at rates traditional cruise missiles cannot match, forcing the PLA to defend ships, ports, airfields, and logistics nodes from multiple axes simultaneously.
Pair it with the surface side. The recent recommendation that Taiwan alone stockpile about 5,500 USVs in hardened underground storage is the maritime counterpart.
Together, massed air and surface one-way attack systems create overlapping, distributed threats that raise the cost and uncertainty of any blockade or invasion attempt without requiring continuous high-end presence.
#11 A Ukrainian think tank says Taiwan should stockpile about 5,500 drone boats in hardened underground storage before any conflict
Research · Policy · USV, maritime strike · Industry/Research
Serpens Center's Catarina Buchatskiy told an FDD forum in Washington that Taiwan needs about 5,500 USVs, citing its "Black Sea Blueprint for Taiwan" report, Focus Taiwan reports.
• ~5,500 USVs, assuming 75% losses • Hardened, dispersed underground storage • Target transports and supply ships, not frontline warships
Via @Focus_Taiwanfocustaiwan.tw/politics/2026…