Tech billionaire Peter Thiel was an FBI informant

[liminalOyster]

Exclusive: Tech billionaire Peter Thiel was an FBI informant

Peter Thiel has worn many hats over the years: Silicon Valley founder, Trump megadonor, cryptocurrency booster, democracy skeptic.

But there is yet another facet to Thiel, one that has remained secret until now: FBI informant.

In the summer of 2021, Insider has learned, Thiel began providing information as a "confidential human source," or CHS, to Johnathan Buma, a Los Angeles-based FBI agent who specializes in investigating political corruption and foreign-influence campaigns.

Charles Johnson, a longtime associate of Thiel's and a notorious figure in the far-right movement that Thiel has subsidized for a decade, told Insider in a statement that he helped recruit the billionaire as an informant by introducing him to Buma.

A source with knowledge of Thiel's relationship to the FBI, whose identity is known to Insider but who insisted on anonymity, corroborated Johnson's account, telling Insider that Johnson brokered a relationship between Thiel and Buma. Insider was able to confirm through an additional source that the FBI added Thiel to its formal roster of registered informants.

Another source close to Thiel told Insider that while they could not confirm that Thiel was a CHS, Thiel did speak to Buma occasionally. The source said that any assistance Thiel might have provided to the FBI should be understood as part of Thiel's gradual distancing of himself from Trump and the broader MAGA movement, which has vigorously criticized the FBI and other federal law-enforcement agencies.

Valuable information on a recurring basis
The FBI maintains a vast network of informants to keep tabs on organized crime, terrorist threats, extremist groups, and other criminal and intelligence targets. These sources, according to the bureau's Confidential Human Source Policy Guide, are more than casual tipsters.

Confidential human sources enter "into a relationship with the FBI, and that relationship will forever affect the life of that individual," the guide says. "[They] will be either an 'FBI source' or a 'former FBI source' and, in turn, his or her conduct or misconduct will reflect upon the FBI." As such, the process for recruiting and maintaining such sources is highly regulated, requiring multiple layers of approval. Only people who are able to provide "valuable information … on a recurring basis" are granted CHS status, according to the policy.

Mike Pence, Donald Trump, Peter Thiel
Then President-elect Donald Trump shakes hands with Peter Thiel during a meeting at Trump Tower in December 2016. Years later, Thiel would become a confidential human source for the FBI. He was directed not to report on his interactions with Trump or other US political figures, an associate said. Drew Angerer/Getty Images
As a CHS, Thiel was assigned a code name and an internal serial number to track his reporting. The information he passed on about foreign contacts and Silicon Valley intrigue was reviewed and "validated," or cross-checked against other sources, by his case agents and their colleagues.

Thiel did not respond to multiple requests for comment. A Thiel spokesperson stopped responding to inquiries after being told that Insider was reporting on Thiel's relationship with the FBI.

The FBI's national press office and Scott Horton, an attorney who represents Johnathan Buma, both declined to comment.

No reporting on political ties
Thiel is a citizen of Germany, the United States, and New Zealand; as of last year, he was reportedly in the process of acquiring yet another passport, from Malta. In 2016, he donated $1.25 million to Trump's campaign and endorsed him from the stage at the Republican National Convention. After Trump won, Thiel served on his transition team.

Johnson, who said that he was also an informant for Buma, told Insider that he believes that Thiel's reporting to the FBI was largely limited to foreign contacts and attempts by foreign governments to penetrate Silicon Valley. Thiel has publicly called on the FBI to investigate Google's ties to the Chinese government.

Thiel, Johnson said, was directed by the FBI not to report on his interactions with Donald Trump or other US political figures.

Many of the politicians that Thiel has bankrolled — including Trump, Sen. J.D. Vance of Ohio, and Blake Masters, a former Arizona senatorial candidate — have repeatedly attacked the bureau and its leadership in public. In 2022, Vance, whose campaign and affiliated PACs received a total of $15 million from Thiel, falsely claimed that the FBI had illegally wiretapped Trump's phone. Vance accused the bureau of "harassing faithful Christians" and pledged to block all of President Joe Biden's Department of Justice nominees in retaliation for the Trump prosecutions. (Most but not all of Thiel's donations occurred in early 2021, before the launch of Vance's campaign and the many false claims he made over the course of the race.)

Senator J.D. Vance
While serving as an FBI informant, Thiel bankrolled the campaign of now-Senator J.D. Vance of Ohio. Vance has made a number of false and disparaging claims about the FBI. Anna Moneymaker/Getty
Masters, whose campaign received $20 million of support from Thiel, has endorsed the false conspiracy theory that undercover FBI agents fomented the January 6, 2021, insurrection and accused the FBI agents who executed a search warrant on Mar-a-Lago of "going after President Trump because they hate him." (As with Vance, most but not all of Thiel's donations to Masters occurred before Masters made his most inflammatory and baseless statements during the later stages of his campaign.)

Thiel is reportedly planning to sit on the sidelines of the 2024 election.

Neither Trump, Masters, nor Vance responded to requests for comment.

Doing business with the FBI
Some of Thiel's business interests rely on the FBI and other government agencies as potential revenue sources. He retains a 10% stake in Palantir, a data company that has sold more than a billion dollars of software and related services to the federal government, including the Pentagon, the CIA, the National Security Agency, and the FBI. A $250 million contract with the US Army in September adds to the evidence that Palantir is essentially "a government service provider," a financial analyst said.

Thiel also backed Boldend, a spyware company marketing itself as an American competitor to the Israeli NSO Group, Forbes reported last year. NSO's products have been bought and tested by the FBI.

(Mithril Capital, another entity Thiel cofounded, was reportedly a subject of FBI interest in 2019, although nothing appears to have come of the inquiry. Mithril's co-founder did not immediately respond to a request for comment. The FBI did not respond to a question regarding the Mithril matter.)

Johnson, who revealed Thiel's FBI ties, is a tech investor and far-right agitator with long-standing ties to both Thiel and to the network of MAGA political operators surrounding Trump. He claims to have had a hand in founding Clearview, a facial-recognition startup, and Traitwell, a genomics company. According to Forbes, he worked with Thiel to help vet and select senior staffers for the Trump transition in 2017.

Johnson claimed to be an FBI informant in a lawsuit he filed against Clearview's founders. He told Insider he recruited Thiel to serve as a CHS and introduced him to Buma, the FBI special agent who was Johnson's handler.

Charles Johnson with Matt Gaetz and Ginger Gaetz
Charles Johnson (left) with Rep. Matt Gaetz (right) and Gaetz's now-wife Ginger Gaetz (center), in 2021. Johnson told Insider that he brokered an introduction between Thiel and the FBI special agent who would become his handler. Charles Johnson
It's unclear whether Thiel remains a CHS for the FBI. Johnson told Insider that he believes that the relationship has been severed but declined to offer details; Buma wrote in his statement to the Senate Judiciary Committee that he was ordered to cease contact with all his sources in late 2022.

Johnson's ties to Trump and Thiel are well-documented. But he is also a self-identified "troll" with a history of spreading false information and smearing his rivals. In this instance, his claims are corroborated by two additional sources, as well as supported in part by a third who says Thiel and Buma spoke occasionally.

'Join up or get crushed'
Buma came forward in August as a whistleblower, alleging that the FBI under Trump and Attorney General Bill Barr shut down his efforts to determine whether Rudy Giuliani had been compromised by a Russian asset. Insider was the first news organization to report on his claims.

In a statement prepared for the Senate Judiciary Committee, Buma said that FBI headquarters had closed his most valuable human sources, including one code-named "Genius," who had reported on far-right figures involved with planning the attack on the Capitol on January 6, 2021. Johnson told Insider that "Genius" was his CHS code name. Insider was able to confirm Johnson's identity as "Genius" with two additional sources. The statement does not mention Thiel.

Peter Thiel at the RNC
Thiel speaking at the Republican National Convention in 2016. He donated $1.25 million to Trump's campaign in 2016. He backed MAGA-aligned congressional candidates in 2020 and 2022. He has said he will be staying on the sidelines of the 2024 presidential race. Alex Wong/Getty
In a written statement to Insider, Johnson said he was stepping forward as a CHS to support Buma's effort as a whistleblower to bring about what Buma believes to be necessary reforms in how the FBI handles informants. Johnson said that he was exposing Thiel's work as a CHS as retribution for what Johnson perceives to be bad decision-making by the Founders Fund, Thiel's venture-capital firm.

Johnson also told Insider he felt betrayed that Thiel did not invest in Johnson's own startups, which he had expected Thiel to do in exchange for introducing him to Buma. Johnson said that he told Thiel that by offering the FBI a window into his contacts with foreign governments, he could demonstrate his loyalty to the United States.

He described Thiel's motivation for working with Buma as a kind of hedge in an environment where extravagant wealth no longer affords the safety it used to. He pointed to ProPublica's reporting on Thiel's income-tax avoidance and the death of Jeffrey Epstein, who had reportedly scheduled several meetings with Thiel.

"I told him to join up or get crushed," Johnson said.

The FBI's recruitment of Thiel as a CHS puts him among the bureau's most prominent assets, but he is not alone among right-wing figures who have collaborated with the bureau. Trump himself offered to help the FBI fight organized crime in Atlantic City in the early 1980s. Truth Social, the Trump-owned social-media company, has quietly tipped the FBI off to users who threaten violence, even as it seeks to cash in on their anger. At least two of the rioters who showed up to storm the Capitol on January 6 were also FBI informants, as was Enrique Tarrio, the Proud Boys leader who was sentenced to 22 years in prison for seditious conspiracy and other felonies stemming from the Capitol breach. Tarrio reportedly served as a source for federal and local law enforcement, assisting with the prosecution of more than a dozen people.

Mattathias Schwartz is chief national security correspondent at Insider. He can be reached at schwartz79@protonmail.com.

https://www.businessinsider.com/peter-t ... us-2023-10

[Belligerent Savant]

.
Likely a stooge for other letter agencies as well...

A couple breadcrumbs (though I believe this forum has historical content on this):

https://theintercept.com/2017/02/22/how ... ole-world/

HOW PETER THIEL’S PALANTIR HELPED THE NSA SPY ON THE WHOLE WORLD
Documents provided by NSA whistleblower Edward Snowden reveal Palantir’s role in creating the U.S. government’s international spy machine.

Sam Biddle
February 22 2017, 6:06 a.m.

Palantir Technologies is a public American company that specializes in big data analytics. Headquartered in Denver, Colorado, it was founded by Peter Thiel,[3] Nathan Gettings, Joe Lonsdale, Stephen Cohen, and Alex Karp in 2003.
...

2013–2016: Additional funding

"[As of 2013] the U.S. spy agencies also employed Palantir to connect databases across departments. Before this, most of the databases used by the CIA and FBI were siloed, forcing users to search each database individually. Now everything is linked together using Palantir."
— TechCrunch in January 2015[28]

A document leaked to TechCrunch revealed that Palantir's clients as of 2013 included at least twelve groups within the U.S. government, including the CIA, the DHS, the NSA, the FBI, the CDC, the Marine Corps, the Air Force, the Special Operations Command, the United States Military Academy, the Joint Improvised-Threat Defeat Organization and Allies, the Recovery Accountability and Transparency Board and the National Center for Missing and Exploited Children. However, at the time, the United States Army continued to use its own data analysis tool.[28] Also, according to TechCrunch, the U.S. spy agencies such as the CIA and FBI were linked for the first time with Palantir software, as their databases had previously been "siloed."[28]

...

https://en.wikipedia.org/wiki/Palantir_Technologies

(it shouldn't be assumed that this tech didn't have funding from certain agencies from its inception)

etc.

[drstrangelove]

i'm in two minds about the benefits of deifying mass surveillance institutions. i mean, what power does an omnipotent god have if no one believes in them? sure, if they are to use the information kinetically to inform policy decisions that remove liberties etc. , but that requires the mass acceptance that they can in fact collect this information and that it is correct. like israel's 'secret nuclear program', there's no point having a psychological threat if you don't threaten people with it. so you get a book like the 'samson option' or an edward snowden. like some robbed monk who bursts through the city gates to warn of an insurmountable omnipotent god who knows all our sins and that we need to practise purification rituals on our data with vpns etc. i mean if they really do have this insurmountable power, why not undermine it and force them to demonstrate it. what they fear is people believing all the data they collect is useless.

[DrEvil]

They collect a lot of fluff, because they collect everything, but even seemingly useless stuff like "aw, it's so adorable" under a cat picture helps them build out your profile. When they have enough "useless" data on you they can model you and predict your behavior, and if they can predict your behavior they can change your behavior. Maybe you're not important enough to give two shits about today, but you might be in ten years, and the important people of tomorrow are growing up always online.

Politics are going to be fun when "they" have access to a pool of millions of potential candidates for various official positions, and can guide and groom them their entire life to get there. No more looking for that one perfect candidate in a haystack when the entire haystack is mapped down to the last straw and you have a ranked list of the top 10K potentials. The first "products" are probably already getting their feet wet.

As for Thiel, I'm surprised it took this long for him to be revealed as an official snitch. His business model the last few decades has been population-wide snitching for profit.

[Belligerent Savant]

These platforms -- including website analytics tools such as Meta Tracking Pixels [see here for an overview: https://www.freshpaint.io/blog/direct-r ... sing-hipaa] -- collect quite a bit of PII [Personal Identifiable Information] and PHI [Protected Health Information], in addition to mere 'fluff'; these identifiers, in aggregate, are also incorporated into any profiling generated of user activities (and often sold to marketers and other entities, let alone whatever certain letter agencies may or may not be doing with the data). There's alot of details/info that can be gleaned from this data: your relatively precise geolocation, mobile/computer devices utilized and browsers, spending habits/purchases, searched interests, comments on social media or blog postings (especially if a user is browsing while logged in via a profile account like Facebook or 'Disqus', a platform that can track any/all comments posted across participating websites while logged in, etc).

[Full Disclosure: as part of my line of work I have intimate/in-depth knowledge of these types of tools, both as an advisor to corporate clients to help them remain compliant with various HIPAA or Privacy/Regulatory Compliance laws, and also to facilitate large-scale electronic record reviews on behalf of law firms for litigation; historically, some of these litigations have involved massive class action lawsuits leveled against a number of companies [pharma, energy, investment firms, etc], as well as whistleblower-related litigation and fraud cases (involving financial institutions, silicon valley startups, energy companies, life sciences/pharma companies, etc. I have signed myriad NDAs because of this; the information I've observed/analyzed over the years certainly informs a number of positions I hold on many of the topics raised here).

[DrEvil]

Yeah, the default position for anyone doing anything online should be that everything you do is tracked, collated, sold and used to influence you in every possible way (legal or not. The fines are just the cost of doing business. The really bad stuff is outsourced to disposable sweatshops, and you're very shocked when it's revealed what they were up to and you definitely didn't use any of that data in your business and you definitely won't be using those people again. Why, it's an outrage!).

Even just walking around gives them a ton of data on you through the GPS (where, how long and how often, add payment info for what and how many), accelerometer, gyroscope (gait, physical disability, "rhythmic movements" at certain locations), microphone (breathing patterns, everything you and those around you say out loud), wireless networking ( Scientists Can Now Use WiFi to See Through People's Walls : https://www.popularmechanics.com/techno ... ugh-walls/ ) and depth-sensing camera/IR (3D mapping of your surroundings) in your phone. Not to mention smart watches with temperature sensors (fever? No entry!) and heart monitors. You can even get phones with full Predator-style heat vision (pair it with the wifi wallhack for some real life science fiction).

Take all that and you have the training data for the Person of Interest AIs.

[stickdog99]

They collect a lot of fluff, because they collect everything, but even seemingly useless stuff like "aw, it's so adorable" under a cat picture helps them build out your profile. When they have enough "useless" data on you they can model you and predict your behavior, and if they can predict your behavior they can change your behavior. Maybe you're not important enough to give two shits about today, but you might be in ten years, and the important people of tomorrow are growing up always online.

Politics are going to be fun when "they" have access to a pool of millions of potential candidates for various official positions, and can guide and groom them their entire life to get there. No more looking for that one perfect candidate in a haystack when the entire haystack is mapped down to the last straw and you have a ranked list of the top 10K potentials. The first "products" are probably already getting their feet wet.

As for Thiel, I'm surprised it took this long for him to be revealed as an official snitch. His business model the last few decades has been population-wide snitching for profit.

Good post. Instead of recruiting on college campuses (both officially and unofficially) or waiting for grant or scholarship applications (that cite Ayn Rand as a favorite novelist) to come to them, these agencies now have their own (hyper)Linked In profiles for almost every PMC-trained individual on Earth.

[stickdog99]

These platforms -- including website analytics tools such as Meta Tracking Pixels [see here for an overview: https://www.freshpaint.io/blog/direct-r ... sing-hipaa] -- collect quite a bit of PII [Personal Identifiable Information] and PHI [Protected Health Information], in addition to mere 'fluff'; these identifiers, in aggregate, are also incorporated into any profiling generated of user activities (and often sold to marketers and other entities, let alone whatever certain letter agencies may or may not be doing with the data). There's alot of details/info that can be gleaned from this data: your relatively precise geolocation, mobile/computer devices utilized and browsers, spending habits/purchases, searched interests, comments on social media or blog postings (especially if a user is browsing while logged in via a profile account like Facebook or 'Disqus', a platform that can track any/all comments posted across participating websites while logged in, etc).

[Full Disclosure: as part of my line of work I have intimate/in-depth knowledge of these types of tools, both as an advisor to corporate clients to help them remain compliant with various HIPAA or Privacy/Regulatory Compliance laws, and also to facilitate large-scale electronic record reviews on behalf of law firms for litigation; historically, some of these litigations have involved massive class action lawsuits leveled against a number of companies [pharma, energy, investment firms, etc], as well as whistleblower-related litigation and fraud cases (involving financial institutions, silicon valley startups, energy companies, life sciences/pharma companies, etc. I have signed myriad NDAs because of this; the information I've observed/analyzed over the years certainly informs a number of positions I hold on many of the topics raised here).

And if you are a gig worker of almost any sort (even an online teacher or tutor), you now typically have give permission for a data aggregator company to create a full profile of your identity. For the companies who set up these "disruptive" services, the data of their vast network of employees is one of their most valuable products that they in turn shill to whoever is willing to pay, including multiple federal "security" agencies.

[JackRiddler]

Shocked, shocked, that a primary architect of the new global machinery of mass surveillance and control in depth and scale, a fervent high priest of transhumanism and techno-eugenics for the rich and hellscape for you, is also a loyal provider of retail personal snooping for the Feds.

[drstrangelove]

They collect a lot of fluff, because they collect everything, but even seemingly useless stuff like "aw, it's so adorable" under a cat picture helps them build out your profile. When they have enough "useless" data on you they can model you and predict your behavior, and if they can predict your behavior they can change your behavior.

but the data doesn't tell them why people behave how they do. for instance, a culture war may be ignited by the mass media and someone may use the internet as an outlet to ventilate the hatred being forced into them with. "fucking kill republicans" or "fucking kill liberals" etc. thus the data they produce would show hate. however, they may be behaving hatefully as a means to keep this out of their day to day thoughts and feelings in the real world. meaning based on the data you may expect them to behave radically in the real world, but they won't, precisely because they behave radically online as a sort of waste disposal of what's been feed to them causing them discomfort. when we eat food we shit out all the excess waste.

if data is rational and humans are not, then using purely rational means to predict human behavior is irrational. the whole thing is a contradiction.

google may know all the books i order online, but it doesn't know the change that occurs in my mind when i read them, or if i stop reading them 20 pages in(i don't use ebooks ). if an ai was fed all the comments i've made on RI which express the views on things informed from books i've read, i would imagine it would still come across numerous contradictions, as over the few years i've been on here, my views have changed or shifted based on things that are beyond even my own comprehension.

i read plato's republic as an input, but express opposition to it as an output. how does this make sense to the input/output feedback loop? because of other inputs it is unaware of + my lived experience through the senses which it has no control of.

law of averages only predict average behavior. people that cause change through creativity and ingenuity don't behave averagely. thus this method cannot be used to predict change. if they can't predict change they can't stop it.

data surveillance is entirely reactionary, not predictive. thus its threat is as a reactionary tool used as a means of force, either through blackmail or manipulation.

if it can be used to predict behavior then you must accept it can be used to 'prevent' 'bad' behavior. there in lies the danger. whether it is the 'future' 'bad' behavior of a person, or even a virus.

[DrEvil]

They collect a lot of fluff, because they collect everything, but even seemingly useless stuff like "aw, it's so adorable" under a cat picture helps them build out your profile. When they have enough "useless" data on you they can model you and predict your behavior, and if they can predict your behavior they can change your behavior.

but the data doesn't tell them why people behave how they do. for instance, a culture war may be ignited by the mass media and someone may use the internet as an outlet to ventilate the hatred being forced into them with. "fucking kill republicans" or "fucking kill liberals" etc. thus the data they produce would show hate. however, they may be behaving hatefully as a means to keep this out of their day to day thoughts and feelings in the real world. meaning based on the data you may expect them to behave radically in the real world, but they won't, precisely because they behave radically online as a sort of waste disposal of what's been feed to them causing them discomfort. when we eat food we shit out all the excess waste.

But that behavior can be modeled. Based on our models, synthesized from the observed behaviors of millions of people, thousands with similar profiles to drstangelove, drstrangelove has a 98% probability that online venting will not lead to real-world action (factors considered: posting history, online interactions and the sentiments of the interactees, wider popularity of sentiment expressed, the degree to which high-probability individuals express the same sentiment, the risk of high-probability individuals taking real-world actions, the probability that high-probability individuals actions will influence and change the probability score of low-probability individuals like drstrangelove, etc., etc.). Low priority, but sentiments expressed online warrant further attention. Next.

And it's not necessarily the behaviors of individuals that are of greatest concern to "them". If your concern is a specific individual old fashioned surveillance is still good. Population scale models are probably more useful, helping to predict trends, public sentiment and potential pushback to policies coming down the pipe, and where to focus your efforts to neutralize the pushback. Better models means more fine-grained control, so instead of carpet-bombing state capitols with lobbyists you can identify the weak points and pivotal people and apply more finely targeted pressure there for a fraction of the cost. It's all about improving the control methods they already have.

if data is rational and humans are not, then using purely rational means to predict human behavior is irrational. the whole thing is a contradiction.

But humans aren't irrational 99% of the time, and even the irrational behavior often has a rational explanation if you dig down a bit, and all of that can be modeled. Online posturing and venting isn't irrational for instance, it's exactly what it sounds like, venting, releasing some pressure with little or no consequences. Perfectly rational, no different than complaining about politicians at the pub in ye olde days.

google may know all the books i order online, but it doesn't know the change that occurs in my mind when i read them, or if i stop reading them 20 pages in(i don't use ebooks ). if an ai was fed all the comments i've made on RI which express the views on things informed from books i've read, i would imagine it would still come across numerous contradictions, as over the few years i've been on here, my views have changed or shifted based on things that are beyond even my own comprehension.

If they compile all the data they have on all the people who read a specific book, and how their behavior changed after reading it, they have a pretty good idea what happens in your mind after reading it. If they really want to know they'll target you specifically for more invasive methods. At a stretch, the mic in your phone could detect the sound of you turning pages and determine what page you're on, or if you check your messages the camera gets a glimpse of the book.

i read plato's republic as an input, but express opposition to it as an output. how does this make sense to the input/output feedback loop? because of other inputs it is unaware of + my lived experience through the senses which it has no control of.

That's just it, it doesn't make sense, and it doesn't have to. The other day I typed "chihuahua lion" into Stable Diffusion, and out popped a photo of a tiny chihuahua lion regally surveying its domain, the back yard. It doesn't make any sense, but it still gave me a great result because it's been fed insane amounts of data that allows it to take "irrational" inputs and output rational results, because very few things actually make no sense if you have enough data.

law of averages only predict average behavior. people that cause change through creativity and ingenuity don't behave averagely. thus this method cannot be used to predict change. if they can't predict change they can't stop it.

The outliers require regular people to act or react in relation to their behavior for it to have any effect, and those regular people can be predicted and controlled. Just be on the lookout for outliers, and make sure you can get in on the ground floor to guide things, and you're good.

data surveillance is entirely reactionary, not predictive. thus its threat is as a reactionary tool used as a means of force, either through blackmail or manipulation.

For now. As models improve and the data available to train the models multiplies, prediction will become easier and easier. Right now there's still a significant proportion of the population who don't live their entire lives online, but that number is shrinking fast.

if it can be used to predict behavior then you must accept it can be used to 'prevent' 'bad' behavior. there in lies the danger. whether it is the 'future' 'bad' behavior of a person, or even a virus.

There's probably a ton of people who watched Person of Interest and thought to themselves "I want that!", and set about building it. New tools for old assholes. The potential for massive automated abuse is staggering, and should be fucked with at every turn.
The grim fate that could be ‘worse than extinction’
https://www.bbc.com/future/article/2020 ... telligence

Control has always had an element of prediction involved. You can't constantly be reacting and expect to win, you need to be a few steps ahead of the other guy.

[drstrangelove]

But that behavior can be modeled. Based on our models, synthesized from the observed behaviors of millions of people, thousands with similar profiles to drstangelove, drstrangelove has a 98% probability that online venting will not lead to real-world action (factors considered: posting history, online interactions and the sentiments of the interactees, wider popularity of sentiment expressed, the degree to which high-probability individuals express the same sentiment, the risk of high-probability individuals taking real-world actions, the probability that high-probability individuals actions will influence and change the probability score of low-probability individuals like drstrangelove, etc., etc.). Low priority, but sentiments expressed online warrant further attention. Next.

they can't model intent. they have no way to predict whether a comment is made in jest, to vent, or to threaten prior to its potential actualisation. they can only claim they can.

if they claim they can, and it is a claim since they can't actually prove what hasn't happened until it does -- data modelling isn't real -- then people could keep pointing out variables they need to control for - such as racial groups, geography, income, education, local laws, criminal record, sexual orientation, lead content in the water supply of every town they've lived in since birth, and on and on and on and on.

and so they begrudgingly keep adding all these controls to their model, and to their horror suddenly realise that the more controls they add to it the more they shrink their sample size.

which means they must either start rejecting certain controls to protect their sample size, which means their models become arbitrary and can be dismissed as non predictive, or they keep controlling for variables and shrink their sample size until it's random.

the power of their models will be based, as they were during the pandemic, on people's acceptance of these models in their minds as accurate predictions of the future.

don't we all remember Nate Bronze and his great predictive polling models of the 2016 general election? in 2019 australian exit polling, the polling literally after people have voted, failed to predict the election victory of scott morrison. it turned out data models had no way to control for people --- lying to them.

i wonder how often people lie online?

[DrEvil]

I don't see why they couldn't be able to model intent. Look at what a person has said over the last then years, paying particular attention to posts where they express some kind of intent (I'm gonna go protest about X at city hall!), then look at later data to see if said intent was carried out or not. Cross reference with all other available data to create a profile of this person and the likelihood they're only blowing hot air. They care a lot about topic X, so they're more likely to carry out intent in relation to that. They claim to care about Y, but other data indicates it's just keeping up appearances and fishing for social acceptance, etc. Repeat for millions or billions of people, and you have a model that can assign a value to the probability that any one person will carry out their stated intentions in relation to a particular topic (this is obviously wildly simplified and probably entirely wrong. It's more of a thought experiment in working out how something could be done, not how it will be done).

They won't be able to make entirely accurate predictions about the actions of specific individuals, but with enough data they can probably get a generalized model that tells them a specific person has an 85% chance of doing X when they said they will.

I'm not saying that these things are possible now, but that they could be, and way faster than you think, and that you're wildly over-estimating how unpredictable the average person is. We are biological systems, and systems can be mapped and understood, and you don't need to model every single drop of water in a waterfall to predict where the average drop will end up.

[Belligerent Savant]

.

Predictive Analytics. It's been in use for some years now in the Legal "eDiscovery" realm to help dramatically reduce the total amount of documents that would require human eye review (among other methods employed, one scenario involves a 'seed set' of highly responsive documents that are fed into the machine learning system, which in turn pulls out a number of key identifiers that it then seeks to locate in a much larger population of data. Similar methods are also utilized by big corps to spot flag-worthy/anomalous behavior patterns on systems that utilize live monitoring of IT infrastructure in highly regulated industries, such as broker-dealer firms, etc).

What is described below is the publicly-available tech. We can only speculate the black box tech in use by certain subsets within letter agencies.

https://cloud.google.com/learn/what-is- ... 20behavior.

Predictive analytics defined

Predictive analytics is the process of using data to forecast future outcomes. The process uses data analysis, machine learning, artificial intelligence, and statistical models to find patterns that might predict future behavior. Organizations can use historic and current data to forecast trends and behaviors seconds, days, or years into the future with a great deal of precision.

How does predictive analytics work?
Data scientists use predictive models to identify correlations between different elements in selected datasets. Once data collection is complete, a statistical model is formulated, trained, and modified to generate predictions.
The workflow for building predictive analytics frameworks follows five basic steps:

1. Define the problem: A prediction starts with a good thesis and set of requirements. For instance, can a predictive analytics model detect fraud? Determine optimal inventory levels for the holiday shopping season? Identify potential flood levels from severe weather? A distinct problem to solve will help determine what method of predictive analytics should be used.
2. Acquire and organize data: An organization may have decades of data to draw upon, or a continual flood of data from customer interactions. Before predictive analytics models can be developed, data flows must be identified, and then datasets can be organized in a repository such as a data warehouse like BigQuery.
3. Pre-process data: Raw data is only nominally useful by itself. To prepare the data for the predictive analytics models, it should be cleaned to remove anomalies, missing data points, or extreme outliers, any of which might be the result of input or measurement errors.
4. Develop predictive models: Data scientists have a variety of tools and techniques to develop predictive models depending on the problem to be solved and nature of the dataset. Machine learning, regression models, and decision trees are some of the most common types of predictive models.
5. Validate and deploy results: Check on the accuracy of the model and adjust accordingly. Once acceptable results have been achieved, make them available to stakeholders via an app, website, or data dashboard.
...

https://www.ibm.com/topics/predictive-a ... lsrc=aw.ds

What is predictive analytics?

Predictive analytics is a branch of advanced analytics that makes predictions about future outcomes using historical data combined with statistical modeling, data mining techniques and machine learning. Companies employ predictive analytics to find patterns in this data to identify risks and opportunities. Predictive analytics is often associated with big data and data science.

Today, companies today are inundated with data from log files to images and video, and all of this data resides in disparate data repositories across an organization. To gain insights from this data, data scientists use deep learning and machine learning algorithms to find patterns and make predictions about future events. Some of these statistical techniques include logistic and linear regression models, neural networks and decision trees. Some of these modeling techniques use initial predictive learnings to make additional predictive insights.

Of course, I've already expressed here many times my scrutiny of 'models' when applied in other far less controlled and highly variable environments, such as when attempting to predict illness/mortality trajectories (Re: covid; the highly flawed Imperial College models) and when utilized to attempt to predict weather/climate trends years into the future. These examples generally utilize different types of modeling techniques, however.

[DrEvil]

But regardless of what you think of the accuracy of various models now, I think we can agree that they're not going to get less accurate in the future. Obviously people are a bit more complex than legal discovery, but we're now at the point where an entire generation has grown up in the digital panopticon. Hundreds of millions of people (at least), all tracked from childhood to adulthood, with profiles that would make the Stasi blush, plus all the other data sets that compliment it: financial records, traffic patterns, power usage, buying patterns, fashion trends, music, movies, pictures, art, all the news, sports, weather, crop data, politics, housing, legal documents, memes, camera footage, radar data, encrypted communications just sitting on a disk somewhere waiting for the day they can crack it, etc., our collective digital subconscious, all of it digitized by default and ready to be fed into the machine so it can model our environment and how we behave within it.

It must be twenty years since I first read about the US government's desire to build a world simulation, and they're finally at the point where it might be possible. Something like Google Earth, but instead of zooming in and getting a year old picture from street view you get a real time representation of everyone and everything happening there (this part I'm pretty sure they've had for years. I have a hazy memory of reading about a classified Google Earth-alike). Then task an AI with keeping tabs on suspicious behavior and to notify an operator for review as necessary.

It could for instance detect a demonstration before it started by looking at the movement patterns of individuals known to be affiliated with protest movement X, and, through training on historical data, know that this kind of movement in relation to each other is simple day to day life, while this kind of movement is what happens when they're planning something, then dispatch police and pick them all up before they even get to the protest site.

Edit: typo