Fresh Cambridge Analytica Leak

[Raoul]

Tweet

— Twitter API (@user) date

This is going to be quite a story....

 

[edcunited1878]

FFS....the lengths those with influence and money will go is fecking frightening. They've lost touch with any ounce of reality for so long. feckin hell.

 

[Brwned]

The unpublished documents contain material that suggests the firm was working for a political party in Ukraine in 2017 even while under investigation as part of Mueller’s inquiry and emails that Kaiser says describe how the firm helped develop a “sophisticated infrastructure of shell companies that were designed to funnel dark money into politics”.

Hopefully there's more focus in the media this time on these longstanding, fundamental issues in politics instead of exclusively focusing on the more trendy (but less influential) subject of social media.

 

[Wednesday at Stoke]

Hopefully there's more focus in the media this time on these longstanding, fundamental issues in politics instead of exclusively focusing on the more trendy (but less influential) subject of social media.

The question is if anyone is going to prosecute them.

 

[VorZakone]

FFS....the lengths those with influence and money will go is fecking frightening. They've lost touch with any ounce of reality for so long. feckin hell.

They're fecking relentless aren't they. Incredible dedication to their 'cause'.

 

[Pexbo]

The question is if anyone is going to prosecute them.

When it’s the ones in power who have benefitted? Absolutely no chance.

 

[Brwned]

The question is if anyone is going to prosecute them.

Unfortunately the lawmakers don't seem particularly bothered by dark money in politics. Wonder why...

 

[Jippy]

Unfortunately the lawmakers don't seem particularly bothered by dark money in politics. Wonder why...

Nor do those most likely manipulated into voting how they did, eg Brexiteers, Trump supporters etc...

 

[Javi]

Don't think targeted advertising is generally illegal and haven't heard anything on its use being illegal when pursuing ideological gain, that might vary by country though. Not sure there is anything new in this article. Cambridge Analytica (or the susequent firm, thought they rebranded) gathers personal data and places personalized ads. Manipulation is wrong but the problematic part is that voters believe these things.

 

[VeevaVee]

No one will care. Especially those most influenced by it.

 

[Ødegaard]

Why did we stop with public hanging again?

 

[Pexbo]

Don't think targeted advertising is generally illegal and haven't heard anything on its use being illegal when pursuing ideological gain, that might vary by country though. Not sure there is anything new in this article. Cambridge Analytica (or the susequent firm, thought they rebranded) gathers personal data and places personalized ads. Manipulation is wrong but the problematic part is that voters believe these things.

The problem is that technology has moved way faster than the laws. This is ethically very very dodgy and much of what they do is actually illegal due to how they collect the data and how they use it, they just count on profits out running the fines.

The legal system needs to catch up and to understand what they are doing in order to figure out what should be illegal.

Subliminal messaging is illegal, when it was found out how advertisers were using it laws were passed to ensure it cannot be used because it’s unethical. The same needs to happen here. It’s not just that they are micro targeting adverts that is the problem, it’s that they are data mining personal data in order to classify people psychologically and then target them with emotive misinformation that subconsciously changes their opinions.

Its absolutely and entirely unethical and should be illegal without any doubt.

 

[Javi]

The problem is that technology has moved way faster than the laws. This is ethically very very dodgy and much of what they do is actually illegal due to how they collect the data and how they use it, they just count on profits out running the fines.

The legal system needs to catch up and to understand what they are doing in order to figure out what should be illegal.

Subliminal messaging is illegal, when it was found out how advertisers were using it laws were passed to ensure it cannot be used because it’s unethical. The same needs to happen here. It’s not just that they are micro targeting adverts that is the problem, it’s that they are data mining personal data in order to classify people psychologically and then target them with emotive misinformation that subconsciously changes their opinions.

Its absolutely and entirely unethical and should be illegal without any doubt.

Don't really know anything about subliminal messaging but if that works and they'd use it that would be another question. But there is nothing about that in the article so not sure how you get there.

As far as the data mining goes I don't see anything in that "new leak" that suggests the data was obtained illegaly. Lots of people freely share their personal views etc. online for everbody to see. Mining this data isn't illegal because these persons chose to share it which brings us again back to it being a demand side problem. Voters get manipulated all the time because they don't know shit about what they are voting for. Misleading ads and targeted ads are placed all the time as well.

So what is it that you want to be illegal? I get that it's unethical but I don't think the law can help.

 

[Classical Mechanic]

The problem for the left, in the UK at least, is that they're always miles behind on utilising stuff like this. They just aren't smart or savvy enough.

 

[Brwned]

Don't really know anything about subliminal messaging but if that works and they'd use it that would be another question. But there is nothing about that in the article so not sure how you get there.

As far as the data mining goes I don't see anything in that "new leak" that suggests the data was obtained illegaly. Lots of people freely share their personal views etc. online for everbody to see. Mining this data isn't illegal because these persons chose to share it which brings us again back to it being a demand side problem. Voters get manipulated all the time because they don't know shit about what they are voting for. Misleading ads and targeted ads are placed all the time as well.

So what is it that you want to be illegal? I get that it's unethical but I don't think the law can help.

I don't know if what Aleksandr Kogan did was illegal but it certainly wasn't in compliance with Facebook's regulations, which are presumably designed with the law in mind. I've seen very little on what they were actually able to do with those 87 million "stolen" profiles, but presumably you can build some pretty useful predictive models based on people's friend groups / social networks.

I don't think we know enough about the mechanics or effects of these hyper targeted social media campaigns with disturbing names like "attitudinal inoculation" to be so blasé about their deployment or how they should be dealt with. There has been no evidence of subliminal messaging, but there's obviously lots of supraliminal messaging being deployed at a frequency and with a degree of versatility and targeting that we've not seen before on such a scale. It's plausible that could be more dangerous for both individuals and society more broadly.

 

[Sky1981]

I don't know if what Aleksandr Kogan did was illegal but it certainly wasn't in compliance with Facebook's regulations, which are presumably designed with the law in mind. I've seen very little on what they were actually able to do with those 87 million "stolen" profiles, but presumably you can build some pretty useful predictive models based on people's friend groups / social networks.

I don't think we know enough about the mechanics or effects of these hyper targeted social media campaigns with disturbing names like "attitudinal inoculation" to be so blasé about their deployment or how they should be dealt with. There has been no evidence of subliminal messaging, but there's obviously lots of supraliminal messaging being deployed with a level of frequency, versatility and targeting we've not seen before on such a scale, and it's plausible that could be more dangerous for both individuals and society more broadly.

In our hands? Meh...

But in some other hands it could be huge. From designing the next bond to catering the presidential slogan.

You dont need to sublimal. If you know for sure 90percent of the caf is losing hair, your chances of selling a hair product here is better than ever. If you know 90percent of the caf are bald, chances are they'd welcome a bald james bond.

Only some crude examples, but it is effective in the right hands. Doesnt have to be highly intellectual, some numerical facts about little things can be manipulated.

 

[Brwned]

The problem for the left, in the UK at least, is that they're always miles behind on utilising stuff like this. They just aren't smart or savvy enough.

It's worth noting that the main campaign strategist on the right, Isaac Levido, said they used very little of this hyper targeted social media that's all the rage and instead relied on broad messages in mass media...in one of their most successful campaigns in recent history.

https://www.thetimes.co.uk/article/...triumph-with-skill-and-sharp-slogan-ddg2h0t5g

 

[Brwned]

In our hands? Meh...

But in some other hands it could be huge. From designing the next bond to catering the presidential slogan.

You dont need to sublimal. If you know for sure 90percent of the caf is losing hair, your chances of selling a hair product here is better than ever. If you know 90percent of the caf are bald, chances are they'd welcome a bald james bond.

Only some crude examples, but it is effective in the right hands. Doesnt have to be highly intellectual, some numerical facts about little things can be manipulated.

I currently work in the industry that functions almost entirely based on those principles, and work alongside a company that loosely helped Cambridge Analytica perpetrate this scandal, so I get the idea...but I think it's mostly a load of nonsense. The people that do these things don't really know how they work. Cambridge Analytica were full of people who make small fortunes convincing people they can do things they can't. Maybe some folks in MI5 or whatever have a complete understanding of the human brain and how to pull those levers with incredible accuracy, but unless we se some evidence of that, it's just the same broad fear about government mind control. These political consultants just capitalise on that fear with bullshit.

 

[Kentonio]

I currently work in the industry that functions almost entirely based on those principles, and work alongside a company that loosely helped Cambridge Analytica perpetrate this scandal, so I get the idea...but I think it's mostly a load of nonsense. The people that do these things don't really know how they work. Cambridge Analytica were full of people who make small fortunes convincing people they can do things they can't. Maybe some folks in MI5 or whatever have a complete understanding of the human brain and how to pull those levers with incredible accuracy, but unless we se some evidence of that, it's just the same broad fear about government mind control. These political consultants just capitalise on that fear with bullshit.

It doesn't require you to have a perfect understanding of the human brain to manipulate people en masse. As the Russians showed perfectly clearly when they targeted the left wing of the Democratic party with apparently positive messages which were designed to drive a wedge between the left and centre of the Democrats and result in the left not turning out to vote for Clinton in as big a numbers as expected. Or their campaign posting apparently pro-African American messages that served to drive AA voters into not voting. Once you have people's habits and basic ideologies its not exactly complicated to design ways to push them in the directions you want.

 

[Javi]

I don't know if what Aleksandr Kogan did was illegal but it certainly wasn't in compliance with Facebook's regulations, which are presumably designed with the law in mind. I've seen very little on what they were actually able to do with those 87 million "stolen" profiles, but presumably you can build some pretty useful predictive models based on people's friend groups / social networks.

I don't think we know enough about the mechanics or effects of these hyper targeted social media campaigns with disturbing names like "attitudinal inoculation" to be so blasé about their deployment or how they should be dealt with. There has been no evidence of subliminal messaging, but there's obviously lots of supraliminal messaging being deployed at a frequency and with a degree of versatility and targeting that we've not seen before on such a scale. It's plausible that could be more dangerous for both individuals and society more broadly.

They had stolen some data earlier, iirc for the Trump election. That's got to be punished – no question there. Not sure who you mean with «blasé» but I am all for researching the impacts of what you call "supraliminal messaging" but before we have results that can point to problems I don't see what should be done from a legal point of view. What is the danger you think is plausible?

 

[Brwned]

It doesn't require you to have a perfect understanding of the human brain to manipulate people en masse. As the Russians showed perfectly clearly when they targeted the left wing of the Democratic party with apparently positive messages which were designed to drive a wedge between the left and centre of the Democrats and result in the left not turning out to vote for Clinton in as big a numbers as expected. Or their campaign posting apparently pro-African American messages that served to drive AA voters into not voting. Once you have people's habits and basic ideologies its not exactly complicated to design ways to push them in the directions you want.

The new problem you've outlined there is that it was done by bad actors in an entirely hidden fashion, and this is something social media is particularly vulnerable to due to their commercial models. But the actual messaging themselves, causing voters to drop out or to turn against particular candidates...that takes place in mass media too. The methods are different but the effects are no less significant. People with money use the media to change people's opinions with various kinds of messages, often very strategically and cynically, mostly without us realising. If that's what we're defining as manipulation, the conversative party just did that very effectively through mass media and the influence on popular opinion is quite easy to see.

This notion that we have transformative technologies in the form of social media and predictive analytics to build a deep profile of the entire population's deepest desires and most powerful political triggers, and activate them on a scale never imagined before...it's mostly hyperbole. The people saying this are those who benefit commercially and politically. The assumptions informing such practices aren't even well developed yet, never mind having the technology to deliver on it. Political advertising in mass media the US has been doing the same thing, blatantly and out in the open, for decades. And it's more transparent, but only marginally so - dark money still gets in there.

Social media opens up new methods with its more targeted approach, but it effects less people. It presents new problems but the dystopian portrayal of it is just fictional. It's scary because it's new. If American political attack ads were shown to people who had no concept of this kind of manipulation before, they'd be terrified too. We're just normalised to it and focus on the new stuff, rather than the larger problems that enable all of these forms of manipulation.

 

[Olly Gunnar Solskjær]

A new film reveals how Cambridge Analytica, collaborating with a software company, has created a platform for US churches that targets the poor, the addicted and the disabled — to radicalize them for far-right politics.

This evidence, I assume, pointed to some form of meddling in the Brexit referendum.​

​

Yes. We documented it for a year, interviewed all the MPs, and then, of course, the political will to follow what the MPs recommended in their report was not that great, because it did prove that Brexit was interfered with.​

​

And we thought, OK, where does this piece go now? Because Charles had placed himself at the nexus of people giving evidence to the committee, people kept bringing him evidence on the side. So somebody brought a whole bunch of evidence about who Cambridge Analytica collaborated with in the United States other than the Trump campaign.​

​

And what did you discover?​

​

It turned out to be far-right-wing churches, conservative churches in the US. And they've built a platform that targets mentally ill or vulnerable people in order to draw them into church, to monetize them through donations. That's the short-term goal. To help them is the facade for it, but ultimately the aim is to convert them to the politics of the far right.​

​

And we went to as many churches as we could. We spoke to as many people as we could. Charles looped in a senior academic from Melbourne and a professor of journalism at Columbia, and a whistleblower who used to work for SCL (Strategic Communication Laboratories Group), the parent company of Cambridge Analytica.​

​

And it ended up with these three tiers. We looked into the data side and then ultimately ended up finding that the people who built that platform had ties to the White House essentially through an enormous secret non-profit organization, one of the most powerful organizations in the United States.​

​

Can you tell me a bit more about that?​

​

Charles Kriel: What initially happened is that a Koch brothers-funded charity commissioned Cambridge Analytica, along with a software company called Glue, to build a software platform that could be used by churches in order to target vulnerable people.​

​

And these are people who are suffering from addiction, financial distress, who might be struggling with opioid dependence or they might be dealing with bipolar issues. And all of these options are available in the software that has been deployed to the churches. And once those people are identified, they can target them with social media. And once brought into the church, they can also be recruited into the politics of the far right.​

https://www.dw.com/en/us-religious-...people/a-55062013?maca=en-rss-en-all-1573-rdf

 

[calodo2003]

A new film reveals how Cambridge Analytica, collaborating with a software company, has created a platform for US churches that targets the poor, the addicted and the disabled — to radicalize them for far-right politics.

This evidence, I assume, pointed to some form of meddling in the Brexit referendum.​

​

Yes. We documented it for a year, interviewed all the MPs, and then, of course, the political will to follow what the MPs recommended in their report was not that great, because it did prove that Brexit was interfered with.​

​

And we thought, OK, where does this piece go now? Because Charles had placed himself at the nexus of people giving evidence to the committee, people kept bringing him evidence on the side. So somebody brought a whole bunch of evidence about who Cambridge Analytica collaborated with in the United States other than the Trump campaign.​

​

And what did you discover?​

​

It turned out to be far-right-wing churches, conservative churches in the US. And they've built a platform that targets mentally ill or vulnerable people in order to draw them into church, to monetize them through donations. That's the short-term goal. To help them is the facade for it, but ultimately the aim is to convert them to the politics of the far right.​

​

And we went to as many churches as we could. We spoke to as many people as we could. Charles looped in a senior academic from Melbourne and a professor of journalism at Columbia, and a whistleblower who used to work for SCL (Strategic Communication Laboratories Group), the parent company of Cambridge Analytica.​

​

And it ended up with these three tiers. We looked into the data side and then ultimately ended up finding that the people who built that platform had ties to the White House essentially through an enormous secret non-profit organization, one of the most powerful organizations in the United States.​

​

Can you tell me a bit more about that?​

​

Charles Kriel: What initially happened is that a Koch brothers-funded charity commissioned Cambridge Analytica, along with a software company called Glue, to build a software platform that could be used by churches in order to target vulnerable people.​

​

And these are people who are suffering from addiction, financial distress, who might be struggling with opioid dependence or they might be dealing with bipolar issues. And all of these options are available in the software that has been deployed to the churches. And once those people are identified, they can target them with social media. And once brought into the church, they can also be recruited into the politics of the far right.​

https://www.dw.com/en/us-religious-...people/a-55062013?maca=en-rss-en-all-1573-rdf

Well, if the targeted can believe such religion, they are totally susceptible to even more egregious levels of batshittery.

 

[Sweet Square]

Tweet

— Twitter API (@user) date

 

[VorZakone]

Tweet

— Twitter API (@user) date

I thought we were being manipulated and all that?

 

[Pexbo]

Tweet

— Twitter API (@user) date

No mention of AggregateIQ, the data firm owned by Bannon who were outsourced by CA for all the actual data mining?

 

[Sweet Square]

No mention of AggregateIQ, the data firm owned by Bannon who were outsourced by CA for all the actual data mining?

There's a mention of it here but that's about it

The ICO hence noted “it is suspected” that some parts of the original Kogan data may have been used in connection with political campaigning for the US 2016 presidential election, albeit in modelled form:

For example, it is understood SCL (through contracts with firms including AIQ) deployed advertising on the Facebook Platform which was targeted to specific voter demographics informed by the profiling that had been undertaken by SCL/CA and GSR

Sources at Cambridge Analytica, however, have always disputed this, claiming that the data was only being quarantined for modelling comparison reasons. As it stands, the final report offers no compelling evidence to dispute that.

 

[Brwned]

Tweet

— Twitter API (@user) date

Shocker. @Pexbo ready to admit you got caught up in the hype yet?

 

[Pexbo]

Shocker. @Pexbo ready to admit you got caught up in the hype yet?

https://bylinetimes.com/2020/10/09/brexit-spin-vote-leave-cambridge-analytica-ico-dcms/

 

[Brwned]

https://bylinetimes.com/2020/10/09/brexit-spin-vote-leave-cambridge-analytica-ico-dcms/

As far as I can see, that article doesn't touch on the main thrust of your argument. It's quibbling about incidental details rather than the point you made, which was entirely in line with the hype: that these analytics techniques combined with targeted advertising were a game-changer going forward, and that they had a significant impact on either campaign. Do you still believe that a) these techniques were able to do anything remotely close to what they claimed, and b) that social media was able to harness that power to be a primary cause of the shock results?

 

[Pexbo]

As far as I can see, that article doesn't touch on the main thrust of your argument. It's quibbling about incidental details rather than the point you made, which was entirely in line with the hype: that these analytics techniques combined with targeted advertising were a game-changer going forward, and that they had a significant impact on either campaign. Do you still believe that a) these techniques were able to do anything remotely close to what they claimed, and b) that social media was able to harness that power to be a primary cause of the shock results?

a) Yes, I’m not sure what you think I think their techniques were? I understand them very well, it was less data mining to produce accurate predictive models and more reinforcement learning through relentless trial and error, posting advert after advert, making subtle changes to text, font and images and establishing what worked and what didn’t. Their micro targeting was never based on sophisticated predictive models, they had access to voter registration records so they knew their audience and they knew where they were located so the microtargetting was on districts in swing states they knew would decide the election. Which leads on to...
b) It was a matter of 60,000 votes across 5 states which decided the election. All 5 of those states were subject to relentless disinformation campaigns targeted with laser precision at the districts they knew would be decisive.


As for the Brexit campaigns, the key point is that 3 leave campaigns were having their digital campaigns (which were full of disinformation) managed by AggregateIQ, a company separate but intrinsically linked to CA and it’s AIQ that handled the data so an very soft and incredibly politically sensitive investigation into a company which spent 3 days bleachbitting their offices before the investigators were allowed in was always going to be piss poor.


Anyway, here’s some more leaks:

Tweet

— Twitter API (@user) date

 

[Brwned]

a) Yes, I’m not sure what you think I think their techniques were? I understand them very well, it was less data mining to produce accurate predictive models and more reinforcement learning through relentless trial and error, posting advert after advert, making subtle changes to text, font and images and establishing what worked and what didn’t. Their micro targeting was never based on sophisticated predictive models, they had access to voter registration records so they knew their audience and they knew where they were located so the microtargetting was on districts in swing states they knew would decide the election. Which leads on to...
b) It was a matter of 60,000 votes across 5 states which decided the election. All 5 of those states were subject to relentless disinformation campaigns targeted with laser precision at the districts they knew would be decisive.


As for the Brexit campaigns, the key point is that 3 leave campaigns were having their digital campaigns (which were full of disinformation) managed by AggregateIQ, a company separate but intrinsically linked to CA and it’s AIQ that handled the data so an very soft and incredibly politically sensitive investigation into a company which spent 3 days bleachbitting their offices before the investigators were allowed in was always going to be piss poor.


Anyway, here’s some more leaks:

Tweet

— Twitter API (@user) date

Yeah, I understand them too. Been working with them for years at this point. The way you talk about them gives off the impression you think because you understand how they work, you have this deep insight that you feel compelled to share. I think you'd be surprised how many people on here understand them too.

You're now saying it wasn't about data mining and models but about targeting. Here's what you said in this very thread:

It’s not just that they are micro targeting adverts that is the problem, it’s that they are data mining personal data in order to classify people psychologically and then target them with emotive misinformation that subconsciously changes their opinions.

If you look, you'll find other posts of yours echoing that point. It wasn't in question whether they were using targeted ads, nor is it particularly interesting, because everyone is. They were just using more of it than their political opponent. What was in question is this more pervasive idea which was the source of the hype. Targeted advertising was not something that made people think "this has changed elections as we know it".

So let's try it again. Do you believe that they can classify people pyschologically, in a substantive and useful way, do you believe that they can use that to subsconsciously change people's opinions, and do you believe that can win elections?

 

[Pexbo]

Yeah, I understand them too. Been working with them for years at this point. The way you talk about them gives off the impression you think because you understand how they work, you have this deep insight that you feel compelled to share. I think you'd be surprised how many people on here understand them too.

The post you have quoted above was a very casual and superficial comparison of a technique that was found to be unethical over half a century ago (and subsequently made illegal) to a technique that has been discovered in the last half decade, found to be unethical and has resulted in new laws to curtail and my suggestion was that we still haven't fully understood the implications of it or gone far enough to stop.

If you care to read the entire post again rather than out of context quoting the most controversial claim:

The problem is that technology has moved way faster than the laws. This is ethically very very dodgy and much of what they do is actually illegal due to how they collect the data and how they use it, they just count on profits out running the fines.

The legal system needs to catch up and to understand what they are doing in order to figure out what should be illegal.

Subliminal messaging is illegal, when it was found out how advertisers were using it laws were passed to ensure it cannot be used because it’s unethical. The same needs to happen here. It’s not just that they are micro targeting adverts that is the problem, it’s that they are data mining personal data in order to classify people psychologically and then target them with emotive misinformation that subconsciously changes their opinions.

Its absolutely and entirely unethical and should be illegal without any doubt.

And then read your post on the subject just two posts below it...

I don't know if what Aleksandr Kogan did was illegal but it certainly wasn't in compliance with Facebook's regulations, which are presumably designed with the law in mind. I've seen very little on what they were actually able to do with those 87 million "stolen" profiles, but presumably you can build some pretty useful predictive models based on people's friend groups / social networks.

I don't think we know enough about the mechanics or effects of these hyper targeted social media campaigns with disturbing names like "attitudinal inoculation" to be so blasé about their deployment or how they should be dealt with. There has been no evidence of subliminal messaging, but there's obviously lots of supraliminal messaging being deployed at a frequency and with a degree of versatility and targeting that we've not seen before on such a scale. It's plausible that could be more dangerous for both individuals and society more broadly.

... you might notice that we are broadly saying the same thing. You'll also realise that I never claimed they employed subliminal techniques, only brought it up as an example of unethical advertising behaviour that was banned once reviewed. It's unfortunate if you think I've been arguing that they have some how managed to psychologically model their potential audience and then ply them with some sort of subliminal messaging voodoo tricks to indoctrinate part of the voting public to vote for their causes because I've never claimed anything of the sort.

You're now saying it wasn't about data mining and models but about targeting. Here's what you said in this very thread:

It’s not just that they are micro targeting adverts that is the problem, it’s that they are data mining personal data in order to classify people psychologically and then target them with emotive misinformation that subconsciously changes their opinions.

If you look, you'll find other posts of yours echoing that point. It wasn't in question whether they were using targeted ads, nor is it particularly interesting, because everyone is. They were just using more of it than their political opponent. What was in question is this more pervasive idea which was the source of the hype. Targeted advertising was not something that made people think "this has changed elections as we know it".

Just to be clear this is their technique as I understand it and why I believe it's unethical and illegal:

1. Collect huge amounts of personal data.
2. Use traditional machine learning techniques to data mine and create initial predictive models likely based on the fields you can use on social media to target adverts (Age, gender, job, interests, behaviour and friends).
3. Use voter registration records and polling data to target important geographical areas.
4. Apply predictive models to target adverts.
5. Use reinforcement learning to optimise the adverts they use based on engagement statistics.
6. Use advert engagement statistics and changing behaviour of audience to further optimise the predictive models.
7. GOTO 3.

So do they use data mining?
Yes.

Did they data mine people's personal data which was illegally obtained?
Yes.

Did they "psychologically profile people"?
Not exactly, they created very broad psychological profiles which were then used to target ads which individuals fit into and each individual no doubt fit into hundreds if not thousands of profiles. There's a misconception that they were running this algorithm against individual people and creating a specific profile for them. This would be an entirely pointless task anyway as you can't target individuals for adverts on social media platforms.

Did they use reinforcement techniques?
Yes, by posting the same advert sometimes hundreds of times with subtle changes each time and as I said: this is the most effective or at least catalytic part of the strategy because it meant they could improve the accuracy of their models and the adverts they were targeted with.

So let's try it again. Do you believe that they can classify people pyschologically, in a substantive and useful way

The narrative you are trying to push here is that I believe they can plough through terabytes of personal data across hundreds or thousands of data points to produce some sort of perfect psychological classification models which can then be used to target adverts at specific individuals. That's not something I have ever claimed. Adverting by it's very nature is about influencing a person's sub-conscience and the techniques CA use are designed to subconsciously change peoples opinions or, more commonly, deeper entrench opinions they already have using factually incorrect or misleading information. I'm not sure what there is to argue with here, are you suggesting that advertising does not have any effect of the sub-conscience?

do you believe that they can use that to subsconsciously change people's opinions

Do I believe that targeted advertising works?

and do you believe that can win elections?

Do I believe it can win elections? Elections aren't won by any single strategy so this is reductio ad absurdum. Do I believe that targeted disinformation campaigns have had a significant impact on a number of elections and referendums over the last decade? Absolutely.

The argument all along has always been that the quality or productivity of the end is absolutely inconsequential. It's the ethics or legality of the means which people should be focusing on.

 

[esmufc07]

I agree with @Brwned

 

[Brwned]

Ok, so we agree they can't psychologically profile people and change their opinions by using those psychological models in targeted advertising. It was a load of bullshit made up by political consultants, perpetuated by the data science community in an act of self-aggrandizement, and hyped up by the media who couldn't understand what they were talking about. That dystopian reality is not here now, and may never be, because it's based on a theory of the mind we have absolutely no way of proving yet. The reality that is here is the same reality we've had for decades, that politicians can use popular media to produce disinformation campaigns. And they find it easier to hide their money. The seismic shift never really happened.

 

[Pexbo]

Ok, so we agree they can't psychologically profile people and change their opinions by using those psychological models in targeted advertising. It was a load of bullshit made up by political consultants, perpetuated by the data science community in an act of self-aggrandizement, and hyped up by the media who couldn't understand what they were talking about. That dystopian reality is not here now, and may never be, because it's based on a theory of the mind we have absolutely no way of proving yet. The reality that is here is the same reality we've had for decades, that politicians can use popular media to produce disinformation campaigns. And they find it easier to hide their money. The seismic shift never really happened.

I agree the media got caught up in the buzz of psychoanalysing and profiling though big data but I think you’re being naive if you think they didn’t manage to employ it to social media at a relatively crude level to degrees of success. Replace “psychoanalysing“ and “predictive modelling” with “identifying important demographics” which is essentially what they were doing and they were successful with it.

Advertisers have targeted ads for years, kids toys get advertised during cartoons, bookies advertise during sports events. That’s essentially what CA did. It was different because rather than having scheduled events they can advertise between, they were able to identify specific traits and target advertisements accordingly.

So an example of the process would be something like:

1. They identify a certain district which is important to the election.
2. They drill the data down to people who can vote in this district.
3. They then build a rough model of these voters and look for the most common traits or interests. Let’s say they post a lot about, interact with posts about or like pages about guns, immigrants and Jesus.
4. They put out the first wave of adverts, a few permutations of a white family, obvious Christian iconography being threatened by evil immigrants, unable to protect themselves because the Democrats took their guns away.
5. They see which permutation was most successful and who with, they retrain their predictive model with better quality audience data and run a new wave of adverts based on the best permutations.
6. Rinse and repeat.


So it’s really not as sophisticated as has been touted, maybe not the laser focus that was claimed, more like a rough stick that was gradually whittled into a fancy spoon. The problem is that the rough stick was identified as a really good candidate for a fancy spoon by unethical means using illegally obtained wood.

 

[VeevaVee]

Gotta say, I agree with @Brwned too here

 

[Brwned]

I agree the media got caught up in the buzz of psychoanalysing and profiling though big data but I think you’re being naive if you think they didn’t manage to employ it to social media at a relatively crude level to degrees of success. Replace “psychoanalysing“ and “predictive modelling” with “identifying important demographics” which is essentially what they were doing and they were successful with it.

Advertisers have targeted ads for years, kids toys get advertised during cartoons, bookies advertise during sports events. That’s essentially what CA did. It was different because rather than having scheduled events they can advertise between, they were able to identify specific traits and target advertisements accordingly.

So an example of the process would be something like:

1. They identify a certain district which is important to the election.
2. They drill the data down to people who can vote in this district.
3. They then build a rough model of these voters and look for the most common traits or interests. Let’s say they post a lot about, interact with posts about or like pages about guns, immigrants and Jesus.
4. They put out the first wave of adverts, a few permutations of a white family, obvious Christian iconography being threatened by evil immigrants, unable to protect themselves because the Democrats took their guns away.
5. They see which permutation was most successful and who with, they retrain their predictive model with better quality audience data and run a new wave of adverts based on the best permutations.
6. Rinse and repeat.


So it’s really not as sophisticated as has been touted, maybe not the laser focus that was claimed, more like a rough stick that was gradually whittled into a fancy spoon. The problem is that the rough stick was identified as a really good candidate for a fancy spoon by unethical means using illegally obtained wood.

To me, replacing "psychoanalysing" with "identifying important demographics" is like replacing targeted advertising with mass advertising. They exist in the same field but their meaning is worlds apart, relatively speaking. Politicians have been identifying important demographics for decades, and the advertising they run on mass media was always aimed at reaching demographic groups in specific contexts. The advertising the Biden campaign are running now in Phoenix during a show of Exlatón delivers a specific message to a very clear demographic group and the message is contextually relevant to both that group and the "mood state" they expect viewers to be in.

Targeted advertising can reach smaller, more specific groups, and A/B testing can help them refine the messaging, but my personal view is that's not much of a step change. The people who run this kind of advertising and analyse this kind of data are very keen to claim that it is, but the evidence is pretty murky. It suggests targeted advertising is particularly good at some stuff without ever assessing what it's bad at and providing a full picture of its overall effectiveness, and evidence that "proves" its effectiveness requires you to make some assumptions that might have been plausible a decade ago but now are very debatable.

Ultimately, if targeted advertising was so powerful, you'd think that's where the money would go. But in the most expensive election campaign in US history, it's not even close.

We forecast that in 2020, political advertising in the US will capture $6.23 billion. Roughly 15.3% of that spend will go toward digital and 69.0% will go to TV.

But we can agree to disagree on targeted advertising. My contention was with the idea they built psychological profiles of the US population and had discovered the tools to manipulate people with really effective subliminal messaging. Proper dystopian shit, and definitely the thrust of the media hype. I'd read your posts in this thread as supporting that notion. Now I know you don't!

 

[Kentonio]

Ultimately, if targeted advertising was so powerful, you'd think that's where the money would go. But in the most expensive election campaign in US history, it's not even close.

Why are you linking to an article showing spending from the 2018 midterms and using it as a comparison to 2020 general election spending?

But we can agree to disagree on targeted advertising. My contention was with the idea they built psychological profiles of the US population and had discovered the tools to manipulate people with really effective subliminal messaging. Proper dystopian shit, and definitely the thrust of the media hype. I'd read your posts in this thread as supporting that notion. Now I know you don't!

Has anyone except you actually claimed they were doing any subliminal messaging?[/QUOTE]

 

[Brwned]

Why are you linking to an article showing spending from the 2018 midterms and using it as a comparison to 2020 general election spending?



Has anyone except you actually claimed they were doing any subliminal messaging?

I'm not 100% sure what you're referring to in the 1st one. Maybe I've mixed links. As I understand it, the 2020 elections are the most expensive in history, and a small proportion has been spent on digital / targeted advertising. Which suggests it's not quite as important as suggested. Do you have different information or are you just here to quibble?

Feel free to switch subliminal messaging to "emotive misinformation that subconsciously changes their opinions". I'm not sure how great the distinction is between the two, but if you replace those words, the core point remains the same.