Facebook can’t count, says Cambridge Analytica Cambridge Analytica has disputed Facebook’s claim that it had access to 87 million records from The Social Network™. “Today Facebook reported that information for up to 87 million people may have been improperly obtained by research company GSR,” the company said in a statement. “Cambridge Analytica licensed data for no more …
In this case the truth is likely to be more than either case as both of them are doing a damage limitation exercise. An educated guess would be up to 90% of the electorate using Facebook which translates to > 80% of electorate in total. If they had less, that means that the slurp bot developers would need to be sacked.
So for USA you are looking at >200M and for UK Leave referendum you are looking at >30M.
ZuckerBorg confession that 2Bn out of its 2.2Bn users have had their profiles swiped by bot at least once is a good starting point here. If it was swiped by bot, that means it entered the pool of data to be sold and reused. If we apply that ratio to the user numbers and correct for users outside voting age we will reach more or less the same result.
"The statement also says: “We did not use any GSR data in the work we did in the 2016 US presidential election.” Which rather pours cold water on the “Cambridge Analytica is wot won it for Trump” theory behind much of the interest in the company."
Looks like that cold water you poured on it just turned into a cloud of steam,,,,,,,
C'mon, we expect better reporting on this story here (and have been getting plenty of it too, just to be clear). Considering who set this company up, and what was exposed in the programme, we should not be believing statements like this, let alone writing "analysis" that starts from assuming their thuthfulness.
The statement also says: “We did not use any GSR data in the work we did in the 2016 US presidential election.”
When a company like this says something like that, you have to look at exactly what they said. They did not use any GSR data in the US election, fine. However:
1) They did work in the 2016 election
2) Their work involves mining data to provide targeted micro ads
3) They used another set of data
Perhaps the pertinent questions now will be: what data DID you use in the 2016 US election, where did it come from, are you still using it. People are acting like the GSR quiz was the only profile harvesting tool in operation.
"Perhaps the pertinent questions now will be: what data DID you use in the 2016 US election, where did it come from, are you still using it. "
Also, which elections did you use it? According to their CEO, they were involved in 44 US election races in 2014. That would be the year the data in question was collected. Even if we believe everything they say, the US presidential election is far from the only one; they claim to have been involved in a great many others, as well as having openly boasted about manipulating elections in other countries. That one specific data set can't be tied to one specific election doesn't really seem particularly important.
We keep hearing these numbers in the news as if they have a direct relation to some particular result. If Cambridge Analytica had 30 million records it does not automatically mean that so many millions of people had their minds changed about who to vote for. Yes, they like to brag, but they have no proof. It seems that the significance of the numbers varies according to who is quoting them and at what time.
Sounds just like what I used to do as a student canvasser back in the 80s knocking on halls of residence doors .... at the time various people were confused as to why after asking who they intended to vote for I didn't try to persuade them to change their allegiance - I explained the main task was identifying your party's supporters and then making sure they got out to vote on the day. (Also, I had a selection of leaflets to distribute over the campaign - not enough to put one of each by each door so I selected which leaflet was more likely to be relevant to "idealistic" first year and which to "job hunting" final students and distributed tehm accordingly ...so I suppose I was guilty of evil targetting messaging as well!)
"If Cambridge Analytica had 30 million records it does not automatically mean that so many millions of people had their minds changed about who to vote for. Yes, they like to brag, but they have no proof."
Intent can be criminal in itself. You can be guilty of attempted murder even if your effort had essentially no chance of success. If a company brags about their ability to manipulate elections by questionable means, they don't actually have to be successful in order for people to get upset about it.
No, they probably didn't persuade that many Clinton supporters to vote Trump, however:
The election was pretty close, and they didn't need to persuade that many people
They might have persuaded Trump supporters to get out and vote rather than stay at home
They might have persuaded Clinton supporters to stay at home, on the basis that she was going to win by a landslide anyway; because who on earth would be mad enough to vote for that orange clown.
Due to the very nature of Social Media, even 30 million can cause a lot of damage.
Here how it works. Hit the 30 million with targeted ads. Say 10 million take the bait and like or share it with all their friends. Some of those do the same. After a while it becomes a meme on Social Media and untraceable back to the original source.
To be honest 1 is to many
OK. I'll bite.
"Due to the very nature of Social Media, even 30 million can cause a lot of damage.
Here how it works. Hit the 30 million with targeted ads. Say 10 million take the bait and like or share it with all their friends. Some of those do the same. After a while it becomes a meme on Social Media and untraceable back to the original source.
To be honest 1 is to many"
1. How much do you think it would cost to send ads to 30m? Go find out. It is not cheap!
2. 10m out of 30m to like or share?!?! LOL!
(For comparison) Direct Marketing has a standard return rate of around 0.01. To even imaging 10m out of 30mil would respond is bonkers.
I think so many people are overstating the effects of this.
You have been served ads.
Are you all as fickle as to be directed by these ads? No. You are not.
So why do you all think everyone else is?
Finally, social media is a business with 2 products. The platform and us.
The platform is the product we buy, and we pay for it by being served ads (targeted via our data).
We are the product the advertisers buy. They buy it with money that funds the service/company.
I thought most people understood this? I do.
I expect my data to be used to target ads because they tell me they will do this.
A lot of this complaining is like my nan moaning about ads on ITV. . .
>Hit the 30 million with targeted ads. Say 10 million take the bait and like or share it with all their friends.
It doesn't take nearly that many.
Hit half a dozen people with the same factoid, and it can become 'true' for them and all their like-minded acquaintances.
Hit a few thousand of those 'half dozens' and you've got a polling swing.
"CA's role is spreading disinformation."
And how is this different from Tony Blair 1997-2007, Gordon Brown 1997-2010, Peter Mandleson 1997+.
Or indeed Vince Cable at his recent party political conference telling the assembled, and the country via the main news channels, essentially that every single Leave voter was/is racist.
Jeez, whatever CA did with Facebook data, if they had had as much success in misleading the electorate and gaining their vote as Blair/Brown did then Trump would have won at least 55 states.
What they are trying to do is to distract the focus from ''they abused personal information'' to ''how many ?''
This is not too far from how political messages work these days -- sod the facts, produce vague, emotional messages that most listeners will interpret differently as being good for them.
OK: politicians have always done that, it used to be called 'spin'; but these days it seems more deliberate - we are in a 'post truth' era where people believe things despite clear evidence to the contrary.
In reality every Facebook APP using the API v1 had access to all of the FB users data, and access to their friends data as well. But the scandal is even bigger. FB apps had permissions on iOS and Android to read the users phone book, SMS, photogallery, and uploaded all data to FB cloud (not visible to the user, but analysed). And Facebook creates shadow profiles of those contacts that have no account yet. Websites with embedded Facebook "share buttons" and Facebook comment sections track web surfers and create & update those shadow profiles as well. So it's not about 85mio accounts, it's about 2 billion accounts multiplied by the number of FB apps using API v1. suckerberg..
' “Cambridge Analytica licensed data for no more than 30 million people from GSR, as is clearly stated in our contract with the research company. We did not receive more data than this.”
The statement also says: “We did not use any GSR data in the work we did in the 2016 US presidential election." '
Does anyone believe a word CA says after they explained nicely to the Channel 4 reporter on camera the depths to which they would sink to help a client win an election?
As little credibility as Facebook has (and it's very little), CA has even less. Their disclaimers are meaningless.
But, I think when you're arguing about how many tens of millions of people were affected, you've already lost sight of the real issue here. It's like arguing "hey, I only hit you 10 times, not 20!".
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
