If you’re interested in some deep academic thought on the topic of AI in warfare, I highly recommend Berkeley’s Stuart Russell’s two-part lecture on the topic, which you can find on the CBC Radio program, Ideas:
"This Terminator picture is wrong for so many reasons. First of all, the Terminators fire a lot of bullets that miss their targets. Why do they do that? Secondly, it makes people think that autonomous weapons are science fiction. They are not. You can buy them today," Russell explained.
"Third, it makes people think that the problem is SkyNet, the global software system that controls the Terminators. It becomes conscious, it hates humans, and it tries to kill us all."
Good, would be nice to see spacex style startups start to win contracts for the military. Reduce cost, break up the boeing / lockheed / northrop / etc. lock on hardware contracts and optimally produce new dynamic ideas.
They have always been lots of startups and small companies in defense, but very often focusing on one very specific thing leveraged by a main contractor that interacts with the government.
For example a company of a dozen people that make just a laser sensor which is then integrated into defense products.
What would be nice is more like defense contract interaction as a service. How to navigate the defense contract acquisition and compliance environment is hard and boring and part of why defense systems actually work and why they’re so expensive.
Military contracting will always be frightfully hard.
Generals have to make judgement calls about what will work in the next conflict - talk about incomplete information - while the future enemy’s preparations and even identity is unknown. Then this has to be translated into something that can slot into existing military systems.
To say nothing of political and career considerations, moving budgets, revolutionary new ideas that will come along, training and culture.
And if they get it wrong, the bad guy wins the next war.
Yes exactly. There's entire classes of contracts reserved exclusively for small companies and startups even. Many exist only because of the constant pipeline of SBIR/STTR money.
One of my best buddies is a mid-level exec at Boeing/Lockheed/Northrop. I just met him a week ago and we had a discussion about the utter waste of Tax Payer dollars in these companies. He was getting compensation beyond his contributions (par for the course in large corporations), but his job was to make sure billable hours continue to exceed the budget - like a boiling frog, projects continue to get delayed bit-by-bit and after 10 years you realize that your program is 3x over budget (F-35 Program?).
I want SV/Texas enterpreneurship to completely dismantle the MIC and start afresh. Even if you disagree with defense applications and its ramifications, simply from ROI perspective (tax payer dollars), instead of increasing military budget, we should be asking how can we do more with current budget. Or even cut down the budget and expect 2x more productivity from these fat military contractors with moats like no other.
> Yeah, that's just what we need, Tech company's approach to defense.
> "Hey the missile doesn't work in v1, but we still shipped it"
Yes, that’s how we won WWII: we shipped prototypes based on the new ideas, and improved them based on the lessons from the field. In war, stakes are higher, and you can’t always afford to delay for a long time to ship perfect product: by the time it arrives, the war might already be lost.
This is one end of the extreme, the other is spaceX.
New tech companies are very capable of iterating quickly and producing well tested reliable hardware far better than existing versions of it.
An extreme but absolutely relevant and potentially devastating example of the current system is the F-35.
They did not "iterate," SpaceX, from the start, had a very senior engineering team which they got on a wave of layoffs in aerospace undustry, and NASA basically doing their V1 work.
SpaceX has absolute massively iterated in every single thing they do, what the heck are you talking about? The Falcon 9 more than doubled it's payload from v1.0 to FT Block 5, and gained first stage landing/reusability through a host of iterative improvements. Starship's development has been all about hardware rich iteration with plans to enable it right down to the basic material level. They've been big on MVPs and then iterating from there.
Not really a convincing argument. Had most of those engineers worked on reusable rockets before? It's like saying stripe, netflix, uber, or Facebook did nothing new because their software engineers had written software before. No one is suggesting that the company needs to compromise of novices in the field, only that they are a new aggressive organization that seeks to upset the status qou.
The discipline for program management on the government side is shit and as a result contractors can be too. Some in a malicious or wasteful way. SV startups don’t operate with that mindset (during their scrappy phase). It’s just two ends of the spectrum. There is opportunity it’s just not going to produce these headlines.
The year is 2030. The government of Tel Aviv has outsourced site-management of the Gaza relocation camp to Anduril Industries, which provides 24/7 full-spectrum security guarantees from their command center in Orange County, California. Nothing will get in -- or out.
America's allies are now safe from malevolent authoritarian actors.
Israel doesn't need Anduril. They have a large domestic drone industry with decades of experience and a strong export sector. Israeli drones gave Azerbaijan a decisive edge in that country's recent armed conflict with Armenia. By some estimates, Israel is the world's largest exporter of military drones:
These "anti-drone" drone-launching turrets will be briefly successful, until opponents code trivial evasive maneuvers (randomly perturb velocity) into their cheap attack drones. Then some genius at Anduril will say "what if our 'drone' was really small and got catapulted really fast to overcome any meaningful evasive capabilities of its target" and then you have autonomous rifle turrets, effective against a wide variety of targets.
The US military could probably use a few dozen Andruils to modernize it and make it truly effective at its core mission.
It does seem like a failing of Silicon Valley that it doesn't produce more military tech startups. Palmer Luckey had to get megarich first before it was possible to get Anduril funded. Same thing Elon Musk had to do to create SpaceX.
There's obviously a VC funding gap, and clearly "the VR guy" and the "the PayPal guy" can do a lot more than most people believed. And so can a lot of other people who understand modern software and hardware.
The normal customer/client relationship is completely twisted in regards to the military industrial complex.
A normal relationship would be that you have 5 people who make guns who all show their gun to the army and the army picks the best gun. Instead the army has the 5 companies "propose" a gun, and they choose who gets to TRY To make that gun. And if they fail its sunk cost fallacy all the way down. Also part of why they choose a company to try and make that gun is because they always choose that company and they have the most "experience".
The argument for this state of affairs is that the military industry can no longer be spun up on demand and basically needs to be subsidized so it can continue exist in peace time. The reality is that these industries fuel continual low level conflicts, waste massive amounts of money and destroy the country.
The only real solve is that the they need a new branch of the military that handles production and R&D in house, and also mans and owns the factories, and to stop giving contracts to parasitic companies.
> The only real solve is that the they need a new branch of the military that handles production and R&D in house, and also mans and owns the factories, and to stop giving contracts to parasitic companies.
Might be a good idea, but the military doesn't pay enough (and can't because of regulations) or have the prestige to do this. Part of why the contractors exist is so the military can pay more.
The nice/bad thing about the military is they can literally train people from the ground up in whatever particular skill sets they need. No matter how obscure. The military pays for people to get medical degrees and all kinds of advanced degrees already.
In fact military training is the primary pipeline for people to enter industries with esoteric skill sets. Adding to those skill sets stuff like "chip production" isnt that big of a reach.
The brutal reality is that the modern American military is a giant jobs program to begin with. Having a "factory corps" that has "soldiers" who are basically assembly line workers with military retirements and benefits might actually be a plus for alot of people.
I'm sure they could do production just fine. It's the R&D that is hard. You can't just train random people from the ground up to be great engineers. The people who are capable of that choose by-and-large to make lots more money in the private sector. Even if you could, good luck making them stay when they can easily make 3x or more by leaving.
In what jobs? I don't think that's true. Or it's only true if they make a career change. The jobs it is true for, like pilots, do have a problem with retention and they give large bonuses to keep people in.
If you have a sufficiently high security clearance you will find jobs based on basically that detail alone. Many of which pay much more than what those persons currently make (and yes they would represent a career change)
And yes retention is a constant problem for people in specialized training. But even in "grunt" jobs, if individuals are willing to work for a PMC they can make more money. The military trains electrical engineers, network engineers, security engineers, (even a few software engineers), medical doctors, materials scientists, psychologists, psychiatrists and more.
The officer corps is filled with people with advanced degrees and training. Most of them are in because of the benefits package they are guaranteed to get and pension. I once met a person who got their medical degree from the military and managed to retire debt free in their late 40s while pulling a military pension and was opening their own medical practice basically free of the debt worries that plagued most doctors (and was able to use their veteran status to get some nice loans to set up their business too).
This shocks people but some people actually ARE motivated by a sense of patriotism as well, and in the military it becomes possible to work on things that are basically impossible to do in the "real world". Not everyone is universally motivated by wealth so long as they hit a certain degree where they feel safe and secure in their livelihood and future.
SpaceX showed a way through the bureaucracy and corruption. By building technology that was dramatically better and cheaper, it became almost impossible for NASA to avoid awarding them contracts.
This seems to be Anduril's approach as well. Whether they can execute as well as SpaceX has done remains to be seen.
You could argue this only happened because the technology effectively stagnated for so long that private sector actors were able to not only catch up but get ahead. And even then it only caught up because that particular technology was a passion project for a billionaire, and they werent necessarily optimizing towards wealth extraction in the space.
Personally I dont think its a great idea for a country to let their military technology stagnant if military supremacy is a core part of your international doctrine like the United States.
Also that it's a monopsony - a market with only one buyer. Who's going to develop a new weapon in a competitive field with 2nd place being first loser? That's a lot of money to risk.
Beginning of WW1 era worked kind of liked i am talking about, companies would make prototype guns, and the army would test them all out and choose the one they liked the best. The companies that lost the bids might and did go out of business. But relative cost to make a prototype gun was relatively low. Im sure there was quite a bit of good old boy network going on, but at least we have evidence of the different prototypes being competed and assessed against.
But something like modern fighter jets are just too expensive for a company to have a working prototype first. But I would argue thats a sign that a nation shouldnt be investing so heavily in things like F35s to begin with. I think most estimates have an actual peer level shooting war result in most air plane fleets being wiped out pretty quickly by missiles and destroying the bases. Why does Russian need an F35 equivalent when they can produce more than enough SAM missiles to destroy it? And the time to actually replace an F35 that gets destroyed is what? A year or more? Time to make missiles isnt a year.
The nation that has designs for a "good enough" plane that can go from factory sky to in 1 to 3 months will win that conflict, and the United States doesnt explore that style at all. Even though its the exact method that helped them win their last major conflict!
Of course there are lots of military contractors. There were lots of space-related government contractors before SpaceX too.
SpaceX and Anduril seem to be unique in that they're following the Silicon Valley tech startup playbook and not the typical US government contractor playbook.
While I would be thrilled to not need any counter-drone military spending at all (there should be an international initiative to ban automated military weapons), if have to spend billions to counter them... I feel safer when the contracts aren't going to Boeing.
The Houthis improvise with existing technology but don't have any real manufacturing capacity. Restrictions on the underlying drone tech would be effective for now.
Longer-term, as the technology becomes easier to DIY... if it became unavoidable, it would not be fun, but I am confident the right coalition of the willing could align the Houthis with international weaponry guidelines.
>Restrictions on the underlying drone tech would be effective for now.
RC planes have been around for decades and are mechanically very simple. High-performance quadcopters and their component parts and widely available on sites like Ali Express. Everything that’s difficult to DIY in a modern UAV- microchips, gyros, cameras, radios- are necessary component of any regular smartphone. This genie isn’t going back in the bottle.
Wasn't there this crazy bloke from New Zeeland that made his own DIY cruise missile? Just for fun, apparently, but he got into trouble with the authorities when they found out?
Houthis get many of their weapons from Iran. No coalition of the willing will stop Iran from selling autonomous weapons short of invading and occupying the country. Is that what you want?
Does anyone know if Pentagon's procurement process has changed since the days of "The Pentagon Wars" [0]?
I initially jumped into this whole topic because I was terrible at estimating and thought it was a uniquely personal failure. I started reading into it an US Military has a history of projects that are incredibly late and massively over budget:
- The joint strike fighter program, of which F35 [1] is the result. 10 years late and double the cost (projected life time cost of nearly 2 TRILLION, with a T).
- F22 raptor, initially 750 were planned for but only 170 were made [2].
- C5 is known for massive cost overruns ($2 billion) [3]
Burton, the author of the pentagon wars attributes it to the procurement process itself. The process does not incentivize delivery. The contracts goals is to get in with the pentagon by underbidding or not clearly defining the requirements and then having the issues be fixed in "operational testing", basically at the point right before front line troops get the hardware. In developmental tests contractors faked stuff a lot:
> The Air Force’s sensor fuzed weapon concept of the early 1980s was advertised as a new, high-tech antiarmor weapon that would home in on the heat of a Soviet tank engine.
Fourteen tanks had been arranged in a tight circle, nose to tail like a wagon train in an old western movie. The antiarmor weapons were suspended on a tall crane high above the circle of tanks. When dropped from the crane, the weapons could hit a tank even without the high-tech sensors guiding them. To make matters worse, the weapons did not home in on the heat from the tank engines because the tanks did not have their engines running. In fact, the weapons had homed in on electric hot plates that the program manager had placed on top of each tank. The hot plates were heated to a temperature four times the threshold temperature for the infrared sensors in the weapons, thereby guaranteeing success of the test.
and
> The Air Force’s third-generation laser-guided bomb, PAVEWAY III, scored fourteen direct hits out of sixteen launches in developmental tests conducted by the program manager.* During the operational tests, thirty-nine launches yielded only twenty hits and nineteen failures. Operational testing was suspended seven times because the system did not work well enough to run a test. When it worked, it usually hit the target. When it did not work exactly right (which was half the time), the average miss distance was five miles.1 Yet, PAVEWAY III was approved for production largely on the recommendation of the new commander of the Operational Test and Evaluation Command. It was his view that the “bugs had been worked out,” and he expressed this same view after each of the seven interruptions. This may sound strange, but the new commander was General Richard Phillips, whose former position was that of advocate for Air Force weapons systems, including PAVEWAY III.
Finally:
> The secretary’s chief tester continued to be subordinate to and work directly for the chief developer. Under this arrangement, the chief tester’s views on the adequacy of testing and the implications of test results were stifled by the chief developer.
I'm just stoked we live in a time where companies are named after stuff from Lord of the Rings. It does seem like it's all war stuff though (aptly fitting?)
Anduril - Aragon's sword after it was reforged, also called "Flame of the West".
Palantir - One of the seeing stones, originally made by men but corrupted by Sauron to drive people to despair who used it. Ominous TBH for an intelligence company.
I agree, it grosses me out. Although I always found it appropriate that Palantir is named after an object tainted by Sauron, which corrupts those who look into it.
The palantir lets the user see anywhere and communicate telepathically with other palantir users. Sauron didn't taint the palantir. He would simply overpower anybody who recklessly used one. Aragorn was able to use one and survive Sauron's mental attack.
Granted..in the stories, the palantir caused all sorts of trouble (backdoor for Sauron to infiltrate the corporate networks..) so maybe the name is apt...
he gives the impression of anti-war but also anti-appeasement. The core of LOTR is about dealing with the threat once and for all rather than kicking the can down the road for future generations to deal with
If I’m not mistaken, the palantir was not accurate in its clairvoyance, right? Wasn’t it showing the black flag ships approaching Minas Tirith and caused the head of the city to flee when really it was Aragorn or the elves coming to save the city? Maybe I’m misremembering, it’s been a long time.
There's an area with great backcountry terrain for skiing and hiking not too far from Vancouver, it would make a great place to build their headquarters [0] (please don't...).
This is what happens when students (UBC) get the chance to name peaks... (There's another range not too far with Mt. Brew, Mt. Keg, etc...)
Copyright doesn't apply to created proper nouns. They're too short. Maybe some extreme example might be considered copyrightable, but generally it's a work or creative design, something that has meaning in itself, that's copyrightable. A proper noun without context means next to nothing.
What is copyrighted is a character, or potentially a detailed fictional entity behind the proper noun. If you make a story about a character with similar identifiable traits to Harry Potter, for instance, that violates Rowling's copyright on the character. Even if your character isn't named Harry Potter, but using the same name would make the case super easy to win.
They could be registered as trademarks, but then the estate would have to show that the use of the trademark was confusing or deceptive. I don't see how it could be, unless someone believes we live in Middle Earth, or unless Tolkien's estate used those names for companies (which would have to do things, not just be paper or shell companies) or products. They'd still have to show that the other companies' use of the names was confusing or deceptive; if the estate's companies or products weren't in the tech space, that might be difficult.
I've always thought Palantir's name was particularly good.
The seeing stone lets you see things, but is easy to misinterpret if you're not careful. It is itself neutral, but can be corrupted for evil. Lots of power, but hard to wield. It's a tool and important for those who use it to understand the risks and account for them - to try to wield its power for good.
MEE specifically owns the rights to adaptations. Just using a name would not be related as far as I understand it.
If there is a "trademark" issue, I would think they'd need to work directly with the Tolkien Estate, who still own the original works. (Didn't original D&D use "hobbits" and have to change it to "halflings" because of the Estate?)
Not because trademarks, none are. But because courts have ruled if you use enough elements, esp characters you somehow are infringing copyright. Or rather, it is easier to cave and change your product than endure the risk and expense of long court battle.
Varda - "mightiest of the Valier and Queen of the Valar, was an Ainu and Aratar, responsible for situating the stars in the heavens above Arda". Company that works on space manufacturing.
As a former infantryman, the officers I knew were about evenly split between these types and football players who clearly had a lot of help with their coursework.
And me being in electronics industry, working on 5 drone projects so far, with one more fixed wing in making.
These are exactly same parts of heli drone sold by one the biggest RC toy makers in China — Fly Wing (also apparently some cooler looking SAB, and MKS parts) lobotomised, and wrapped in a sci-fi looking sheet metal skin.
A tremendous achievement in an incredibly short period of time. Congrats to the absolutely stellar Anduril team and to America and her allies for getting better technology to save lives.
https://www.cbc.ca/radio/ideas/quit-using-the-terminator-as-...
"This Terminator picture is wrong for so many reasons. First of all, the Terminators fire a lot of bullets that miss their targets. Why do they do that? Secondly, it makes people think that autonomous weapons are science fiction. They are not. You can buy them today," Russell explained.
"Third, it makes people think that the problem is SkyNet, the global software system that controls the Terminators. It becomes conscious, it hates humans, and it tries to kill us all."