- Palantir was incredible technology during the wars in Afghanistan and Iraq for putting the proverbial warheads on foreheads of insurgents with terrible SIGINT practices and a lot of generated data. You could build and analyze graphs of insurgent networks that were tangibly powerful
- After that, in my mind what was very similar tech was sold to US domestic police, corporate insider threat teams, whatever. As I recall it had uneven adoption due to expense
- Now in 2025, that same tech is slated to have broad access to American citizen data under an entirely trustable and stable executive branch.
With those face value facts, a capable technical mind like those in hackernews could draw logical conclusions.
To put a pin in it - threat modeling for what you say and do online as this era progresses is interesting to consider. Now with tech like this, your threat model is now you + your friends. Who’s the “radical” in your friend group, and is the group chat on unencrypted systems? Consider what your graph would be, and how much do you trust tech like this ran by either the current team or the other team.
Don't forget that Lavender AI, the "cool system" that automatically targets all Hamas fighters (with probably 1000% civilian casualties because it destroyed all hospitals, churches, mosques and schools along the way) was developed by Palantir.
The irony is that this really bad SIGINT graph flags also relatives, e.g. cousins of cousins of fighters, just because they had e.g. family events where they attended together, even though all other intelligence data would point to the contrary. The documentary that got banned from BBC highlights this with a lot of stories where e.g. hospital workers were specifically targeted because a distant relative was associated with hamas.
Palantir had a video on YouTube where they were even bragging about this graph, though not under its now-leaked codename.
Exactly what I remember reading all the time about the Afghanistan war - the OP calls the Palantir tool 'incredible technology' but the one thing I remember seeing reported time and time and time again during that war was reports of strikes against civilians accidentally being targeted because their mobile phones had been nearby to insurgents (like maybe having visited the same mosque or the same family gathering)...
Palantir is extremely bad, but this not making the point you want to make. Hamas infamously wears civilian clothes during combat and operates out of civilian structures in civilian zones. We ought to oppose the destruction of democracy and the arrival of dystopia without defending terrorism.
> destroyed all hospitals, churches, mosques and schools along the way
Hamas doesn't have an air force, and they're facing an enemy with a good air force and good intelligence. They can't have nice things like "zoning", if they stored rockets away from civilians they would all get destroyed.
So, in order to have rockets, they must be stored under targets that would cause political trouble if they were to be hit: hospitals, churches, mosques and schools.
It's funny (horrifying) coming from an advertising background - this sounds exactly like the tech that allows advertisers to take credit for conversions that are wildly unrelated to ads, in reality.
The comment throws around impossible numbers and unverified claims as if they were facts. “1000% civilian casualties” is nonsense mathematically, and nothing backs it up. There is no public evidence the BBC “banned” the documentary, so that looks like hearsay.
Lavender AI is one lead-generation tool among several, with human analysts and other intelligence sources involved. It does not “automatically destroy” hospitals, churches or schools, that is an unfounded exaggeration. Many such sites remain standing, and damage in a war zone is not proof of an AI-driven targeting policy.
Social network mapping is used by counterterrorism agencies everywhere. Being on a graph does not mean you are on a kill list. Without solid data on actual misidentifications, repeating anecdotes does not prove the system works the way you claim.
Don't forget hamas, hiding behind hospitals, churches, mosques and schools makes those valid targets. And it could quit islamo-facism any day of the week!
That "incredible" tech didn't seem to help all that much in Afghanistan. Not only did the US lose, I never got the sense we were even particularly close to winning, even if we'd stayed there for another 20 years and trillion dollars. In terms of tangible wins, what was Palantir's "incredible" tech actually delivering?
It delivered two things, and the easy response to your fair point is tactical tools — a rifle, great software — don’t win wars on their own.
1) Palantir was the first breath of fresh air that brought actually good tech with modern tech support practices to the warfighter, and by extension put the big
defense contractors on notice. I personally believe this impact was tremendously important as there were real safety connotations involved, and anyone with a family member downrange could appreciate this.
2) Palantir was great targeting software that worked like modern tech vs a custom Linux distro with a GUI from 1970 and required 5 months of finagling to get vendor support for.
So Palantir just brought standard 2010’s tech to soldiers betting their safety on it. This was incredible although ordinary.
On the contrary, the US-led coalition achieved military victory in Afghanistan in under 60 days. Which is an incredible feat. Though what that coalition failed to achieve, and where people try to adjust the definition of tactical victory, was the nation-building goal of creating a functional, independent Afghanistan government. The counterinsurgency aspect was the process of protecting that fledgling "nation".
The very uncomfortable truth here is that Israel is demonstrating how to effectively destroy insurgencies in Gaza and Lebanon. You cannot pussyfoot with nasty, brutal tactics and expect to accomplish anything. This was a lesson the west learned in the world wars, and we seem to have collectively forgot it again.
there's lots of near admission by numerous service members (retired now) who go on those war/special forces podcasts and admit they had their hands tied and told to stall. the 'brass' didn't want the war to end - unclear if it was the presidents of the time or the generals but we accomplished the mission those in charge wanted (forever war).
It sounds like you’re judging by the political outcomes, and frankly the tactical effectiveness is pretty far disconnected from that. It’s like saying a Chef’s knife must be dull because the meal tastes bad.
Yea, but with a cool modern spin on it, it's not imperialism it's fighting terrorism!
Still not clear on how murdering the poorest people in the world with the least reach of anyone on the planet with devastating weapons like the R-9X missile, which has been described as a "samurai sword warhead" because it uses bladed protrusions to penetrate and then mangle it's victims, has solved anything. Other than putting loads of money in Palantir and Raytheon's pockets.
I think these days though they just sell political influence since Trumps brand of populism has made war _slightly_ less popular than it was 10 years ago.
> ...incredible technology during the wars in Afghanistan and Iraq
> - Now in 2025, that same tech is slated to have broad access to American citizen data
Speaking of which, only loosely-related, but is there any indication of where the 'recent' leak of British special forces, contractors and/or informants (?) happened? (est. 2022, discovered later, now in news)
News reported it was data getting passed around as a spreadsheet attached to emails. Ironically it would have been possible to build a case management tool with rbac on Palantir Foundry and avoid that screwup.
Military integration with law enforcement -> military tech licenses -> focus on cities -> cities have troves of SIGINT
Unencrypted group chat -> one friend hates one party -> another friend loves to talk about illegal habits -> tool hoovering it all up -> illegal habits friend is the pretext to look at politics friend
Clear as day, as this is what caused a bad time for insurgents in an actual war. Makes a lot of sense to apply it domestically! Tread on me.
Apparently crime only happens in big cities. It's weird, because where I grew up in rural Missouri, every-other dipshit was a meth junkie, robbing houses to support their addiction. But, well, maybe I just invented all that with my crazy imagination.
I heard in reality there was no "tech" but very good PR for consultants ("forward deployed engineers" I think their term was) doing standard data engineering?
My take is that nearly all military technology, particularly mass surveillance tech, will be applied to the civilian sector eventually. That's why everyone using China as the baseline for human rights worldwide is so terrifying.
As far as Palantir goes, they're just another in a long line of corporations that sell such services to the government. Ever heard of RAND Corp? They undertook all kinds of weird studies during the Vietnam era.
So please, spare me the concealed distaste for the current Administration, as if the previous one was some kind of wonder for America! The left is, on the whole, far more totalitarian than the right in the US. I sleep better knowing our borders are secure, racist or not. I also sleep better knowing that criminals are getting locked up once again--that's an important aspect of even having and maintaining civilization! It's such a simple concept, yet many of you ignore it or misunderstand it or something.
Anyways, if Palantir was anything special, the war in Afghanistan would have turned out much differently. They failed to predict that the Afghan government and military would completely collapse the instant Uncle Sam stopped paying the bills there. It was all a giant scam.
Palantir is a tech platform that consumes data from their clients in return for providing high level data-driven insights. They assign FDEs (or consultants) to really learn the details of a customers data. Foundry allows them to get single pane view of the data in an org and they actually have both the tech and engineering skills to do the dirty data cleaning jobs.
For an extravagant fee, you give them your data, they clean it for you, and then those same FDEs can tell you interesting things that you should have known, had you actually done proper data architecture in the first place.
Another thing to add is (according to a coworker who worked with them at previous big company) that they present solutions in a way that makes it hard for the customer to use the solution outside their ecosystem.
They're supposedly very aggressive on that point, so once you integrate their solution to your pipeline you're pretty much stuck with them for the foreseeable future.
Many safety/mission critical companies can get really bogged in by this, with too many administrative hoops to detach. which is probably why they're focusing on that industry.
My coworker liked to describe them as a parasite that creates a symbiotic relationship with their host.
Yes but if you don't have enough budget to pay for their engineering time, they also provide good UI to do data science. It's like a fancier version of Excel for data wrangling: imagine Excel but your data is not necessarily tabular; it may be a graph; it may contain images and multimedia, etc.
I once interviewed at Palantir and at the same they gave a demo of their software to every candidate.
Worked for a similar company with similar clients at the time. Making the data usefull in innovative ways was a big part of it so in a way sure part of it data science related. At the same time I’d say it’s broader than that.
>For an extravagant fee, you give them your data, they clean it for you, and then those same FDEs can tell you interesting things that you should have known, had you actually done proper data architecture in the first place.
AFAIK, this is the most succinct description of Palantir I've read. A looser-fitting analogy is they come in, replace whatever the hell you were trying to use SAP for with actually competent software. Most "FDEs" can't explain what the company does because what they did was work at $CLIENT for 18 months ripping apart all their internal software with Palantir building blocks.
It sounds like fundamentally SAP and Palantir target different use cases though? While SAP has OLAP functions, their bread and butter is highly domain-specific and transactional.
I will not allow Palantir to extend their reality distortion field to me. They are consultants. They are also engineers. Other places call them FEs. But they didn't invent some new class of engineering, they just rebranded one.
As opposed to the more commonly known 'Reverse Deployed Engineer', who sits behind the product manager who can deal with the goddamn customers so the engineers don't have to.
They take an exorbitant fee to clean up the mess government created when they outsourced their tech infrastructure to private sector companies preying on dumb government money.
That’s the thing with government: They always believe you can drown out problems with taxpayer money. They don’t get that what solves problems is never money, but competence, hard work, and having skin in the game.
At least in my country the reason is that the politicians force them to outsource in various ways like not letting them pay their employees market rate salaries.
It is not that they believe more money will solve the problem. It is often cost cutting which makes things this expensive.
> That’s the thing with government: They always believe you can drown out problems with taxpayer money.
They know they can’t drown out the problems, nor do most of them want to. The privatization of government work is just a dog and pony show that lets rich assholes give taxpayer money to other rich assholes.
Not to say the left doesn’t do this too (assuming the US political speak, “left” referring to democrats is really just barely right of center), but part of the conservative playbook has always been to rip apart the federal government (or the parts that they don’t like, such as providing social services). The easiest way is to tank a group by hiring a private company to do a shit job and then saying “see how bad they did? We should just axe food stamps.”
I interviewed at Palantir London about 10 years ago.
I am based in Europe and one of the younger interviewers let-slip that they will all be working during the local public holiday. lols. No thanks.
Also, I grew up in a mixed ethnic environment. For the last few decades there has been a focus on trying to make society more inclusive. Such that my school exam papers would have questions like "Susan has 6 apples and gets 6 more. How many does she have" or "Rohit is travelling at 50mph ...." So a variety of names and genders etc to reflect the people who live here.
Well, my Palantir interview information was about "networks of people that need to be tracked"... all Muhammads, Omars etc. These names were my school colleagues and friends, so this didn't sit well with me (just to be clear, I didn't want to work somewhere that seemed to be making software to track entire groups of people).
They really should have sanitised their material and made it about helping Susan and Rohit track financial crime or some such. Instead I got vibes of that tv show Homeland.
Cynically, it sounds like their interview process worked pretty well for selecting people who were quite happy to work public holidays and had no issue with tracking "people that need to be tracked"
I don't think cynicism is even needed. Public holidays are surprising - there's no legal way in Europe[1] to force someone to work during a holiday. And trying is a lawsuit waiting to happen. But tracking Muhammads? If someone has moral problems with that, they better resign before being employed. Because at best they will resign soon after joining, at worst they will become a whistleblower.
[1] I know Europe is big etc, but I used to work in UK on particular and everyone took bank holidays seriously.
> Well, my Palantir interview information was about "networks of people that need to be tracked"... all Muhammads, Omars etc.
I can't begin to describe how sick this makes me feel. At the same time, I'm happy that your upbringing resulted in you making the right choice of not pursuing such a horrible company.
I on the other hand was contacted by Helsing, which at the time sounded cool to me (????) (secret agent vibes???). The recruiter however failed to appear at the screening call 2 or 3 times, which I interpreted as a strategy on their part to select for persistence/yes-men (Occam's razor on the other hand is that the recruiter was overworked and/or shitty).
I told the recruiter off.
Only after a while I realized that it really, really was not a good thing that I even considered working for a company that in the end just kills people for money (I know, european defense yadda yadda). I have enough trouble sleeping as it is.
I'm curious what names did you expect for an intelligence op in Afghanistan or Iran?
Generally the police take care of things like financial crimes, so not sure why you expected this from a service provider to US three letter agencies
If they can't explain what the product is, it is because they are effectively selling custom high margin consulting services anchored on a data pipeline product. This is not too dissimilar from Oracle, et al, that sell "solutions," "support," and "services" anchored on a not-so-special database product.
But that's not sexy for recruitment and the VC investment math does not/did not like to hear the word consulting, so they lipstick the pig and sell depth-first search as some secret voodoo magic.
This was actually surprisingly clear. This, and htrp's comment are much clearer than the entire noise article.
They make dashboards and apps for killing people. With a lot of technical jargon like "integrating disparate weapons and sensor systems for a kill chain".
Somebody in America says "we want to kill somebody" -> satellite gives real-time imagery on location -> weapons systems available nearby are recommended -> user clicks orders and telemetry go out to field operators and ex: drone systems -> predator fires up and flies to location and bombs target -> real-time imagery confirms explosion and results.
Sure war is bad and killing people is bad, but can we stop acting like it's a choice ? Unfortunately, wars will happen as long as humans exist and it's much better to be on the winning side. So yeah, there are a lot of people building dashboards for killing people and it's not necessarily bad. I would even argue that it's much better than a lot of people whose work is to make kids and adults addicted to screens.
Is it that surprising? Ignoring war being good or bad, you would assume there needs to be some method to the madness. I assume before computers this meant a central com center that kept track of everything using humans and chalkboards or tables.
Do you really expect journalists to do that? What's next - expecting them to travel to countries they're reporting on? It's not the 90s for gott's sake
Recently, I have been increasingly associating Palantir with the 'Samaritan' from Person of Interest, an evil entity monitoring everyone in the digital world, collecting data, and selling it to authoritarian regimes.
They also have one of the most profitable business models the world has ever seen.
Their RPE (revenue per employee) is roughly $1mm and growing at a 50% YoY rate...
They heavily use technology as leverage for insane margin growth. 90% rule of 40 as well.
Have you looked at companies such as Jane's? Overall bigger market cap and RPE orders of magintude higher - don't fall for hype like everyone else, or at least check the numbers before saying 'most profitable', they are not even close.
Readit News
- Palantir was incredible technology during the wars in Afghanistan and Iraq for putting the proverbial warheads on foreheads of insurgents with terrible SIGINT practices and a lot of generated data. You could build and analyze graphs of insurgent networks that were tangibly powerful
- After that, in my mind what was very similar tech was sold to US domestic police, corporate insider threat teams, whatever. As I recall it had uneven adoption due to expense
- Now in 2025, that same tech is slated to have broad access to American citizen data under an entirely trustable and stable executive branch.
With those face value facts, a capable technical mind like those in hackernews could draw logical conclusions.
To put a pin in it - threat modeling for what you say and do online as this era progresses is interesting to consider. Now with tech like this, your threat model is now you + your friends. Who’s the “radical” in your friend group, and is the group chat on unencrypted systems? Consider what your graph would be, and how much do you trust tech like this ran by either the current team or the other team.
The irony is that this really bad SIGINT graph flags also relatives, e.g. cousins of cousins of fighters, just because they had e.g. family events where they attended together, even though all other intelligence data would point to the contrary. The documentary that got banned from BBC highlights this with a lot of stories where e.g. hospital workers were specifically targeted because a distant relative was associated with hamas.
Palantir had a video on YouTube where they were even bragging about this graph, though not under its now-leaked codename.
Hamas doesn't have an air force, and they're facing an enemy with a good air force and good intelligence. They can't have nice things like "zoning", if they stored rockets away from civilians they would all get destroyed.
So, in order to have rockets, they must be stored under targets that would cause political trouble if they were to be hit: hospitals, churches, mosques and schools.
Lavender AI is one lead-generation tool among several, with human analysts and other intelligence sources involved. It does not “automatically destroy” hospitals, churches or schools, that is an unfounded exaggeration. Many such sites remain standing, and damage in a war zone is not proof of an AI-driven targeting policy.
Social network mapping is used by counterterrorism agencies everywhere. Being on a graph does not mean you are on a kill list. Without solid data on actual misidentifications, repeating anecdotes does not prove the system works the way you claim.
Dead Comment
Dead Comment
1) Palantir was the first breath of fresh air that brought actually good tech with modern tech support practices to the warfighter, and by extension put the big defense contractors on notice. I personally believe this impact was tremendously important as there were real safety connotations involved, and anyone with a family member downrange could appreciate this.
2) Palantir was great targeting software that worked like modern tech vs a custom Linux distro with a GUI from 1970 and required 5 months of finagling to get vendor support for.
So Palantir just brought standard 2010’s tech to soldiers betting their safety on it. This was incredible although ordinary.
The very uncomfortable truth here is that Israel is demonstrating how to effectively destroy insurgencies in Gaza and Lebanon. You cannot pussyfoot with nasty, brutal tactics and expect to accomplish anything. This was a lesson the west learned in the world wars, and we seem to have collectively forgot it again.
Still not clear on how murdering the poorest people in the world with the least reach of anyone on the planet with devastating weapons like the R-9X missile, which has been described as a "samurai sword warhead" because it uses bladed protrusions to penetrate and then mangle it's victims, has solved anything. Other than putting loads of money in Palantir and Raytheon's pockets.
I think these days though they just sell political influence since Trumps brand of populism has made war _slightly_ less popular than it was 10 years ago.
> - Now in 2025, that same tech is slated to have broad access to American citizen data
Speaking of which, only loosely-related, but is there any indication of where the 'recent' leak of British special forces, contractors and/or informants (?) happened? (est. 2022, discovered later, now in news)
Unencrypted group chat -> one friend hates one party -> another friend loves to talk about illegal habits -> tool hoovering it all up -> illegal habits friend is the pretext to look at politics friend
Clear as day, as this is what caused a bad time for insurgents in an actual war. Makes a lot of sense to apply it domestically! Tread on me.
Who the fuck talks like this, seriously?
Dead Comment
As far as Palantir goes, they're just another in a long line of corporations that sell such services to the government. Ever heard of RAND Corp? They undertook all kinds of weird studies during the Vietnam era.
So please, spare me the concealed distaste for the current Administration, as if the previous one was some kind of wonder for America! The left is, on the whole, far more totalitarian than the right in the US. I sleep better knowing our borders are secure, racist or not. I also sleep better knowing that criminals are getting locked up once again--that's an important aspect of even having and maintaining civilization! It's such a simple concept, yet many of you ignore it or misunderstand it or something.
Anyways, if Palantir was anything special, the war in Afghanistan would have turned out much differently. They failed to predict that the Afghan government and military would completely collapse the instant Uncle Sam stopped paying the bills there. It was all a giant scam.
For an extravagant fee, you give them your data, they clean it for you, and then those same FDEs can tell you interesting things that you should have known, had you actually done proper data architecture in the first place.
They're supposedly very aggressive on that point, so once you integrate their solution to your pipeline you're pretty much stuck with them for the foreseeable future.
Many safety/mission critical companies can get really bogged in by this, with too many administrative hoops to detach. which is probably why they're focusing on that industry.
My coworker liked to describe them as a parasite that creates a symbiotic relationship with their host.
I once interviewed at Palantir and at the same they gave a demo of their software to every candidate.
Worked for a similar company with similar clients at the time. Making the data usefull in innovative ways was a big part of it so in a way sure part of it data science related. At the same time I’d say it’s broader than that.
AFAIK, this is the most succinct description of Palantir I've read. A looser-fitting analogy is they come in, replace whatever the hell you were trying to use SAP for with actually competent software. Most "FDEs" can't explain what the company does because what they did was work at $CLIENT for 18 months ripping apart all their internal software with Palantir building blocks.
I will not allow Palantir to extend their reality distortion field to me. They are consultants. They are also engineers. Other places call them FEs. But they didn't invent some new class of engineering, they just rebranded one.
so beautiful.
Deleted Comment
That’s the thing with government: They always believe you can drown out problems with taxpayer money. They don’t get that what solves problems is never money, but competence, hard work, and having skin in the game.
It is not that they believe more money will solve the problem. It is often cost cutting which makes things this expensive.
They know they can’t drown out the problems, nor do most of them want to. The privatization of government work is just a dog and pony show that lets rich assholes give taxpayer money to other rich assholes.
Not to say the left doesn’t do this too (assuming the US political speak, “left” referring to democrats is really just barely right of center), but part of the conservative playbook has always been to rip apart the federal government (or the parts that they don’t like, such as providing social services). The easiest way is to tank a group by hiring a private company to do a shit job and then saying “see how bad they did? We should just axe food stamps.”
I am based in Europe and one of the younger interviewers let-slip that they will all be working during the local public holiday. lols. No thanks.
Also, I grew up in a mixed ethnic environment. For the last few decades there has been a focus on trying to make society more inclusive. Such that my school exam papers would have questions like "Susan has 6 apples and gets 6 more. How many does she have" or "Rohit is travelling at 50mph ...." So a variety of names and genders etc to reflect the people who live here.
Well, my Palantir interview information was about "networks of people that need to be tracked"... all Muhammads, Omars etc. These names were my school colleagues and friends, so this didn't sit well with me (just to be clear, I didn't want to work somewhere that seemed to be making software to track entire groups of people).
They really should have sanitised their material and made it about helping Susan and Rohit track financial crime or some such. Instead I got vibes of that tv show Homeland.
[1] I know Europe is big etc, but I used to work in UK on particular and everyone took bank holidays seriously.
I can't begin to describe how sick this makes me feel. At the same time, I'm happy that your upbringing resulted in you making the right choice of not pursuing such a horrible company.
I on the other hand was contacted by Helsing, which at the time sounded cool to me (????) (secret agent vibes???). The recruiter however failed to appear at the screening call 2 or 3 times, which I interpreted as a strategy on their part to select for persistence/yes-men (Occam's razor on the other hand is that the recruiter was overworked and/or shitty).
I told the recruiter off.
Only after a while I realized that it really, really was not a good thing that I even considered working for a company that in the end just kills people for money (I know, european defense yadda yadda). I have enough trouble sleeping as it is.
Dead Comment
No they shouldn't. I prefer seeing the beast as it is, not as how it would like to be seen.
But that's not sexy for recruitment and the VC investment math does not/did not like to hear the word consulting, so they lipstick the pig and sell depth-first search as some secret voodoo magic.
They have a fucking kanban board for bombing people.
They make dashboards and apps for killing people. With a lot of technical jargon like "integrating disparate weapons and sensor systems for a kill chain".
Somebody in America says "we want to kill somebody" -> satellite gives real-time imagery on location -> weapons systems available nearby are recommended -> user clicks orders and telemetry go out to field operators and ex: drone systems -> predator fires up and flies to location and bombs target -> real-time imagery confirms explosion and results.
Sure war is bad and killing people is bad, but can we stop acting like it's a choice ? Unfortunately, wars will happen as long as humans exist and it's much better to be on the winning side. So yeah, there are a lot of people building dashboards for killing people and it's not necessarily bad. I would even argue that it's much better than a lot of people whose work is to make kids and adults addicted to screens.
Deleted Comment
In the same way, do people who do things to paint targets on their foreheads read this stuff, and think twice?
A lot of criminals are rational.
Do you really expect journalists to do that? What's next - expecting them to travel to countries they're reporting on? It's not the 90s for gott's sake
It's such a disgusting modern day leviathan, I roll my eyes to the back of my head when people casually say you should buy their stock
They also have one of the most profitable business models the world has ever seen. Their RPE (revenue per employee) is roughly $1mm and growing at a 50% YoY rate...
They heavily use technology as leverage for insane margin growth. 90% rule of 40 as well.
Meanwhile OnlyFans is at something like $30mm per employee, which is wild.
so yeah not the top of chain
Is their profitable business model based on the fact that they're good at enabling & profiting from authoritarianism and corruption?
The business model was completely stagnating before LLMs.
They are cashing in on the rush for large firms to wrangle data for LLMs but the entire concept of large firms and FDE has obvious scaling issues.