Hacker News Reader

> isnt that close to how Cambridge Analytica happened?

Close to. But different.

Cambridge Analytica drew from Facebook’s servers. This plug-in pulls from users’ computers. Cambridge Analytica was not academically affiliated. This is legitimate research. Cambridge was closed source. This is open.

Most importantly, users are explicitly sharing their data with NYU. There is no carrot of a personality quiz to obscure the quid pro quo.

Cambridge Analytica was academically affiliated. The data they got was laundered through a University of Cambridge researcher Aleksandr Kogan, and their app claimed to be using the data for legitimate academic research. I’m not saying it’s exactly the same, but the fundamental challenge is still present: how can Facebook be sure that these researchers won’t share the data with some controversial organization?

No they were not. Please see this post from Dr David Stillwell from the Psychometric Centers at Cambridge University who responded to a post here a few weeks ago about this very point:

https://news.ycombinator.com/item?id=24912986

Thanks for the link, but I'm sure you can understand my skepticism of a naked assertion that the reporting on this issue was wrong - especially since this was Mr. Stillwell's only comment ever.

Is it really that implausible that someone have have forwarded him a link to a site he has no interest in and he saw a comment he felt compelled to comment on? Since when is the number of of past comments a barometer for legitimacy? I mean Hacker News even gives you the option to post from a throwaway account.

Further the comment is simply refuting one Wired article with an earlier article Wired published on the same subject in which he participated in. It's not like some crazy idea was being posited in the comment.

Lastly Dr Stillwell has made the assertion many times that the app and data he and Dr Michal Kosinskis developed for their "MyPersonality Quiz" app at Cambridge University is neither the same app or data as Aleksandar Kogan's My Digital Life" app that Kogan developed for Global Science Research and their client SCL Elections. The "MyPersonality Quiz" app was simply based on Stillwell's earlier work.

I really don't follow your point here. It's plausible, but lots of things are plausible - it's plausible that the original reporting is correct, plausible that Stillwell is embarrassed about his involvement, plausible that Cambridge was involved in some way that didn't involve Stillwell. You seem to have reached a stronger conclusion, that Stillwell's statement is definitely correct and the reporting was definitely wrong, and I don't see how you got there by just looking at plausibility.

I actually provided a link that to a well-researched article by Wired Magazine. I have also pointed out that the differentiation between the two different apps is well-documented. In that I have given you the two key points that you could easily use in your own Google search.

What have you provided? You have provided exactly nothing to substantiate your "strong" conclusion. It's hard to see how you yourself "got there" based on nothing other than it's just your opinion.

The story is unfortunately complicated, so many news outlets get it wrong. myPersonality was my app that ran from 2007-12. Michal and I published papers based on its resulting data, which was collected with user consent and only from users who used the app (not their friends). We worked in the Psychometrics Centre - a research group in Cambridge University. Prof. Kogan joined Cambridge University later on in 2013 as an independent faculty member not in the Psychometrics Centre, created his own app 'mydigitallife,' and went on to work with Cambridge Analytica using data from users of his app and their friends.

So key points from my perspective are (1) My app isn't the same as Kogan's app, (2) I didn't work with Cambridge Analytica, (3) Kogan was independent faculty, not part of the Psychometrics Centre.

As you've noticed from my post history, I don't usually use this website. I do, however, have an alert set up that pings me links when someone uses my name online. I have it set up so that I can attempt to correct journalists who write about me and my research group and get the facts wrong.

Hope this clarifies things

Edit: Since I'm here. I think it's critical that NYU's Ad Observatory continues and that academics have freedom to collect data and report results that might embarrass rich and powerful companies. Tweet #9 in this thread is the key one ( https://twitter.com/doctorow/status/1329873620353515520 ) - how do we know that FB's own ad library isn't good enough? Because Ad Observatory found that it wasn't reporting all instances of political ads.

cambridge analytica was absolutely academically affiliated when they requested research data from FB, they just lied about their research which is why the doors are closed now.

The methods were vastly different.

i think the point is that no academic is owed any of this data, and while FB was collaborating with academic institutions, its clearly not worth the risk

NYU isn't collaborating with FB though. It's collaborating with FB users, who explicitly agree to do this.

Does FB get to dictate how its users get to share their own data?

I would argue yes, particularly social media data. For one thing most interactions on these platforms involve at least two parties interacting. If one of these parties decides to send the transactions to a chrome extension, it seems they are making a decision unilaterally for all participants. Certainly if anyone is going to do that it will be the platform itself.

I'd also add that Cambridge Analytica used obfuscation to mask the true reason they wanted the data: NYU seems to be totally upfront about the collection.

The plug-in is open sourced: https://github.com/OnlinePoliticalTransparency/social-media-...

You can download a release straight from there and import it into your browser. Browsers also do some reviews of the plugins that are in their stores, so every update should be tested by both Google and Mozilla (even if only superficially).

I'd say the odds of them managing to pull off scraping anything else without anyone noticing are rather slim.

with respect, thats not the issue. With facebook they only collect "limited" information. Its what they do with it afterwards that matter.

How is it stored, is it secure, how do they anonymise it, who has access to the raw data, what happens when there is a break, is there monitoring, etc etc etc.

They are making a massive honey pot, and we just have to trust them that they are doing good job. (much like facebook....)

Please don't throw the baby out with the bathwater. There's no obligation to opt-in, and there's risk in everything we do and don't do.

The tool would need to guarantee that the data of a user opting in doesn't contain private information about his contacts which may not have done so.

Facebook doesn't have the greatest security track record. While it wouldn't be productive to argue that one group is inherently more trustworthy than the other it's worth pointing out (as others already have) that the code is available to the public.

Code being public doesn't mean its secure. See php, django, gnupgp, openssl etc etc etc.

Again, the issue isn't the collection, its the storage, processing and disposal that really matters here.

> its the storage, processing and disposal that really matters

That you can say about almost all web apps? Then almost maybe you shouldn't use the Internet, I start thinking

The last part is the crux of this dispute to me. If the users are opting into this, isn't that the same argument for letting them opt into the Facebook platform itself? Why must we protect them from NYU and not from Facebook?

Thanks for the link. I agree odds would be slim, but this can potentially set a precedent for others to do similar things. Imagine, in the future, a new company comes up with similar add-in claiming to scan ads, and take more data than needed. Facebook needs to be careful of that too. (though i think that is not the reason they are shutting this down, just that privacy is favorable to them in this case :)). There has to be a better way to do this, and that needs serious discussion.

Also, as an aside, the add-in records the following things: country, age, gender, language etc. and the text of the ad. Not so clear what happens when ads have personal info (or friends' names) attached like xyz likes page. Payload seems to suggest they take every bit of info so it could include that as well maybe.

What is the alternative? No visibility for the general public into hyper targeted ad campaigns?

If the users are opting into this, isn't that the same argument for letting them opt into the Facebook platform itself? Why must we protect them from NYU and not from Facebook?

> If Facebook allows third parties to scrape data (who can incentivize users any which way) where does it stop?

What say should Facebook have here though? I'd argue that when I am viewing data through my browser it's on my computer at that point. I've downloaded it so why should Facebook be allowed to decide what I can and cannot do with it any more than they can decide what I can do with my emails that I also view on my computer?

...but Facebook doesn’t have to let you use their service (for free) if they don’t want to.

If they don’t like something you are doing, they can do whatever they like.

Are you sure there isn’t more nuance to the situation?

Like what?

It's not running on their servers at this point: It's on my computer.

Now, I don't know what the US laws exactly say about that but in terms of common sense, FB should not be allowed to make any claims to that data when it's on my computer.

Sure, they can say "well, no more data for you, you're banned" but that's a different argument.

In addition, having control over users of FB is one thing as they signed up to it and agreed to the terms but NYU didn't so I have no idea how they can even claim to enforce them especially if NYU are scraping data from the users computer (I assume they aren't making a connection to the FB servers!)

Also, not that it matters for this argument, I do not have a FB (or any social media) account: I tried it, maybe 10 years ago, and found it silly :)

Edit: Guys, this isn't Reddit! I'm all for downvotes but not drive-by downvoting... a simple "I downvoted you because..." would suffice.

Edit2: Mea Culpa! As has been pointed out, I replied to someone that wasn't actually commenting on my post... not sure how I missed that!

I agree with you. I was referring to the person above me.

Of course they can but my point still stands: what I do with data that is, by every definition, running on MY computer is my business, not Facebook's. Them banning me for doing that (possibly violating their Terms) is a separate argument.

In addition, NYU didn't sign up to Facebook so they have no need to abide by their Terms. At least, they shouldn't have to abide by them... not sure what the laws in the US say about it.

I'm not for government intervention for things like this but the courts (I believe) have already ruled that scraping data is legit.

Edit: added that I think they shouldn't be able to have a say

> I've downloaded it so why should Facebook be allowed to decide what I can and cannot do with it any more than they can decide what I can do with my emails that I also view on my computer?

While I agree with your opinion, that data would still be Facebook/the advertisers' intellectual property and your actions could be considered infringing the terms you agreed with.

Though NYU may not have agreed to those terms, encouraging others to violate the terms and provide them the data could be considered the intellectual property damages needed to at least start a civil court case

Fair point but I would argue that NYU are only looking at the page and saying "this user viewed ad XYZ" and recording that fact.

They're not (I assume) copying the advert and storing it on their servers. I could see Facebook arguing that taking a copy of it is a violation of IP somehow.

But simply looking at the page and recording some of the meta data is the same as someone asking me on the phone "what advert did you just view?".

Would that be violating IP?

Looks like they store the ads in a public DB

https://adobserver.org/ad-database/

Samsung TVs take screenshots of displayed content and send them to their servers. They use it to track what you watch, whether it's the latest blockbuster DVD release or your private homemade porn video :). It's all sent to them.

That doesn’t count as copyright infringement because it doesn’t benefit the end user.

/sarcasm

While I agree with your opinion, that data would still be Facebook/the advertisers' intellectual property

I read something, now it is in my brain. Does Facebook still own it? I suspect they believe they do.

I think the ad's creator would own it but they've licensed it to Facebook.

Isn't(technically speaking) the addon taking data only from the users PC and not from facebook servers? At the end of the day it shouldn't be facebook's call what goes on the user's device.

One thing about these addons is that you can actually take them apart and see what is going on if you don't trust that the addon is built from their github code.

I do think that users opting in is an insufficient response. A lot of users opt in for things they do not understand. (Terms and conditions of any site or app for example). For tech savvy users with the plug in, they will be smart enough to know what they are accepting, though that is not true for everyone.

I like the idea of open source code and that is helpful but there is got to be a better way. And I also believe that the research like this should be supported. All things aside, we can arrive at solutions without the need of an article like this which neither poses such questions, nor provides context.

> I do think that users opting in is an insufficient response. A lot of users opt in for things they do not understand. (Terms and conditions of any site or app for example). For tech savvy users with the plug in, they will be smart enough to know what they are accepting, though that is not true for everyone.

Everything you just said here applies to Facebook themselves as much as the researchers. If we allow people to opt-in to sending data to Facebook (by creating an account), why would be against giving those same users the option to send the data somewhere else?

> isnt that close to how Cambridge Analytica happened?

Yes, its exactly this.

"Facebook changes based on past scandal for the better, and this is bad. here is my 9th grade essay with no research just feeling"

Facebook _should_ be held to account.

I personally don't think that the advertising is actually that much of a problem. It all has paper trails because they have to pay. Advertising is not the worst part of facebook is its users.

for that to change facebook needs to understand that "freedom of expression" disappears when you sign the terms and agreements, which specifically limit your freedom of expression. If they actually enforced their community standards properly, the place would be much better.

Instead they carve out exceptions for celebs, based on precedence that not even employees can find.

Isn't how human nature works? Exceptions, greed, gossip, deception? You can't leave that to a company to enforce with T&C.

> "I personally don't think that the advertising is actually that much of a problem"

Facebook was used to illegally infuence elections in Uk and break spending rules and because of their closed nature, nobody realised untill a whistleblower stepped forwards. Its a massive problem, and maybe we would not have Brexit without it.

www.bbc.co.uk/news/amp/uk-politics-44856992

A TV ad, by comparison, is aired publically and you can see it and ask who paid for it.

The question is the degree to which advertising actually influences behavior. It must have some effect, I’d agree, but was Cambridge Analytica’s ad targeting so powerful that a 1.3 million vote margin was entirely their doing? I’m pretty skeptical.

It was highly targeted advertising on targets which were highly susceptible to be turned. Whatever they were turned by, doesn't matter whether it was factual or not (cause that is the low standard for advertising).

We have Edward Bernays standard of advertising coupled with the power of Facebook available for everyone, including people who lie through their teeth to get what they wanted.

Furthermore, there is evidence Russian trolls influenced the Brexit campaign, and if that isn't enough: it was a moment in polls where leavers indeed won, but before and after the referendum it was more votes for stay and for leave.

I'm actually glad The Netherlands quit with referendum. Elections are very tough nowadays because of how easy people are influenced.

> Elections are very tough nowadays because of how easy people (WHO VOTE AGAINST MY POLITICAL ORIENTATION) are influenced.

Let's be honest, none of this were a scandal if Cambridge Analytica would have worked for Hilary or the remain campaign. I dont deny all this stuff influence people, it clearly does, but that is the price you have to pay for having a society with freedom of speech and universal vote. If you want to get rid of, either you have to take more draconian censorship laws or you eliminate universal voting for adults, both cornerstones of a modern democracy.

Brexit is just an example. Manipulation of citizens is of all time. The danger lies in how precise the targeting nowadays is, together with internet being world-wide and advanced.

Its the price we all paid for some of us having a dishonest ML data hoarder.

I'm involved in the integrity of elections in The Netherlands. I have a political preference, but I am able to leave that out when I manually count votes.

I also am aware of ways which undermine our democracy: electronic voting, strategic voting, and abused ML. I believe a high integrity on elections benefits society on the long term.

However given the international forecast of our societies (ie. what used to be called globalization 20 years ago), and difference in population and cheap labour, a country like Russia or China can heavily influence our society. In this case with the help of an American behemoth, but there is no reason other sources cannot be abused.

I don't have the solution to this problem. It warrants further investigation though. Hopefully before it is too late.

> Brexit is just an example.

Brexit could only take place because the political class spent 25 years using the EU as an excuse for everything that went wrong.

The current PM spent his formative years making up stories about the EU to send back to the telegraph.

FAcebook's advertising only works (assuming that it does actually work) if people are receptive to the message. Those messages have to be thought up by someone.

Thats not facebook's fault. thats the fault of political discourse.

The Pro EU side _lost_ because they failed to follow the standard rules of election winning: Make the other side seem like they are going to make you poorer, or less patriotic (or both)

They spent their time shouting at the sceptics calling them racists. Waffling on about economics, never emotion, never empathy, Never ambition.

Much as its comforting to think that the likes of russia, google facebook and china are at fault, its really not the fault of advertising.

> It was highly targeted advertising on targets which were highly susceptible to be turned

This doesn't gel with me... how many people saw the ads, how many times, and of those, how many were actually influenced by it?

Going by my sample size of two (my sister and my wife) they both claim that they never even look at ads that appear on screen.

I honestly am calling bullshit on the whole ad industry here... show me evidence that a significant number of people are influenced by ads that makes it worth spending money on (if unbiased data actually exists).

Are there any advertisers out there that are seeing a quantifiable return on their ad spend?

It is bullshit I agree. We all know what the intent is with an advertisement. You don’t have to be a “tech-elite” to understand ads.

There should be more transparency around how ads are displayed on the internet.. but giving researchers full access to a user’s data and their friend’s data is a pretty poor one. Can’t a better system be designed for this?

Content, however, is very dangerous and it’s not clear what a good solution is. It’s not clear what the purpose of content is on the internet.

> It’s not clear what the purpose of content is on the internet

Something I have pointed out to others in the past, and on here too, is that the Internet existed just fine before all these ad-related sites appeared. Of course, nothing on the scale of FB, Google etc, but I'd argue that we're worse off now than before.

When you rely on advertising you then have to make sure that every piece of content works towards making the advertising worthwhile. You then maximise SEO, and clicks and focus on metrics and bullshit like engagement (whatever that means).

You have to calculate whether a blog post or article is worth doing because your sponsors now care about that: you stop making content for the love/joy/sake of the content and everything becomes bait for your ads.

20 years ago, the only ads were about punching a monkey and finding out that I was the 1 millionth visitor to a website and could click the ad to claim my prize :)

I agree with your skepticism of referendums, because they do seem to be influenced more by cultural battles than policy outcomes. But I don't follow the connection you're drawing. If survey polls both before and after the election showed a different result, that just indicates the survey polls were flawed - any opinion changes Cambridge Analytica was able to create should be present both in survey polls and the election result.

Survey polls can and do manipulate people. True, unlikely in a referendum, but not impossible. Say the polls say its 55/45 for your team, and its raining heavily (weather is known to influences outcome). Would you be motivated to go? Maybe you would, but some would not.

Another one (not referendum specific; my comment regarding strategic voting wasn't specific to referendum either). Say your preferential candidate is Charlie. You have no idea what other people vote. You'd vote for Charlie. However, say you know its going to be between Alice and Bob. You really don't want Bob to win. So you vote for Alice.

If you consider mass media and hype surrounding survey polls, these cause strategic voting between whoever ends up popular. Consider, for example, the way a candidate gets chosen for Republican or Democratic party. Prime example of strategic voting.

Some see these examples as opportunities to get their preference [for a candidate or political goal] in higher regard. I don't; I see these factors as factors which potentially harm the democratic outcome on the long term. Now, like Earth, democracy is strong and adaptive. It can take a bump. But it isn't endlessly going to accept stomps and beatings. Hence, I believe we need to try to maintain the integrity and authenticity.

My understanding was that its power came from discouraging voting, not encouraging voting for a particular candidate.

Attempted murder is still manslaughter.

Breaking election laws in an attempt to influence an elections outcome is still breaking election laws whatever the outcome. It doesn’t have to be entirely their doing.

Sure. I'm not defending Cambridge Analytica here - just saying that, on a practical level, it seems unlikely that they had a decisive impact on any elections.

Said whistleblower (assuming you're talking about Christopher Wylie) got banned from fb. That's my issue with all this. Did the Russians or Cambridge Analytica do bad stuff? Probably yes. Is Facebook also incredibly hypocritical and self serving? Yes

>Facebook was used to illegally influence elections

yes, the electoral commission found that. sadly its only able to issue a tiny fine. Not only that its being eroded by populists from both sides of the political spectrum. Its not really a facebook problem but an EC one.

This transgression was found by going over the accounts, not by leaks.

> A TV ad, by comparison, is aired publically and you can see it and ask who paid for it.

political TV ads are illegal in the UK. Long may it continue.

We are also supposing that facebook adverts are actually effective. They have the power to somehow corrupt a morally "good" person into someone that votes for the "other side". This is plainly nonsense.

The real issue is that facebook is a warped mirror of society. without rules and order, chaos and bad actors reign. This means "censorship" or if it were treated as the press, editorial standards.

That is not true, according to an investigation ran by a third-party https://www.bbc.com/news/amp/uk-politics-54457407 Do you trust the third-party investigation, or a couple of whistleblowers one of which designed the entire system and another of whom sold the system. No doubt CA is a wake-up call for data security and privacy.. but a sales person telling you how freakin’ powerful the system is and how they can control the world with their ML models. Well we’ve all heard that.

By the way it’s not just TV ads to consider. Mail is a huge part especially for political campaigns and NO one has visibility to that. You can’t just cherry-pick one distribution model of advertisements.

The thing is, I don't see how Facebook have any business allowing or forbidding users to install software on their own devices. I mean, if this was about FB banning users for TOS violations, that's one thing. But claiming a third party is not legally allowed to provide software that does something FB do not like? That's a whole different can of worms.

IANAL, but if NYU makes software designed to violate a contract between user and FB, isn’t that torturous interference? IOW, I think what you just described is not legal. I can’t make something that is designed to violate a contract, then step back and pretend they aren’t encouraging users to violate a contract w fb.

I'm curious, who is on the other side of the debate here? Because I've seen more people argue for a flat earth than a non-evil facebook.

There are several arguments to be made for a non-evil Facebook. But people are unlikely to engage with a starting comparison like that. Perhaps that's why you haven't see any opposing arguments.

Please do.