Facebook’s Ad Scandal Isn’t a Fail, It’s a Feature
nytimes.com
nytimes.com
But worst of all, there's no incentive to fix it.
Most cases of bad actors on platforms are already a violation of the terms of service. But until it starts to affect the platform provider financially, everyone will keep funneling money in to the problem. The bad actors pay for the ads, they pay for the phone numbers, they pay for Internet access, and thus it's too easy to look the other way.
Then the providers come to us, the victims of the crimes whose time is wasted, whose lives are scammed. They say it's too big a problem, and gosh darn it they just can't do anything. They have their pockets turned inside out and shrug at us like we're supposed to believe them. But it doesn't affect them, it affects us, and they've never cared about us, just our money.
You can even use this advertisements to deanonymize data. And that's extremely-dangerous. If you want a list of people from an ideology, or minority, or whatever. You can send an ad targeted to that people, in the city that you are interested in. But the ad is for getting some free, unrelated to your motives, sample product. People will send you their name and address without even knowing that they saw your ad only because they are part of that minority/ideology/...
You can also, and it has already been done, target undecided people in an election. This is nothing new, but the precision and scope has increased drastically. And usually you only need a few percentage points in one state to swing an election. The money that could buy you half a percentage point, now is going to buy you a lot more.
Facebook offers a great feature for ad companies to be able to specifically target people, but it is dangerous for society as a whole. I think that the GDPR (http://www.eugdpr.org/key-changes.html) is a good step in the right direction. But probably it is not enough.
The article misses completely the point. Moving the subject from "privacy" to "antisemitism".
The only real problem I can think of is when these targeted ads are used as a way to lie to one set of targeted people while saying something completely different to another set.
All of this to serve better ads is at the cost of your privacy, everyone's privacy. You can actually find things you might be interested in just fine by searching for them instead of seeing some ad twice that you might be interested in. Serving better ads requires more data on consumers and once the treadmill starts, there is no getting off unless as a consumer you decide to take back control. This is why ad-blockers are as big as they are.
This isn't some externality either, what happened in the election is the externality. To the extent that Facebook interactions fed and influenced the election, it was because better ad targeting; and more consumer time on the site, which helps better ad targeting is the business model. So, I agree with the article very much.
However I don't believe the "precision ads don't work" hype. I've noticed a dramatic increase in the quality and relevance of ads in the last few years, even as I mostly stopped using Facebook, Snapchat, Instagram, and Google. Amid all these bot scandals, I would still be shocked to dind that hyper-precision targeting is ineffective without adblockers and such.
Crying out "ads don't even work!" is just a desperate attempt by anti-ad people (like me) to find some indication tha the ad emperor has no clothes.
And both are true in the cases people most often complain about. I buy a camera and then get to see ads for cameras everywhere. That's both precise and completely ineffective. Basically, the smarts are is in the targeting side, the dumbness is in choosing what ad to show...
Slightly OT, where does the 2b monthly active user statistics come from and how was it measured? I could only find Facebook self reported stats and it is not clear what do they mean by users, and most articles seem to use user/account/person interchangeably.
OTOH Twitter's definition of an "Active User" is laughable: an account must simply follow 30 other Twitter accounts and must also be followed by one-third of the accounts it follows to qualify as an active user.
[1] https://www.fool.com/investing/2016/11/08/monthly-active-use...
I looked at active public-posting G+ activity in January of 2015, and estimated about 4-6 million people matching that classification, based off of a sampling from the G+ sitemap files:
https://ello.co/dredmorbius/post/naya9wqdemiovuvwvoyquq
Stone Temple Consulting expanded that sample 10x (to 500k profiles) and didn't cheap out as I had on the sampling (I used a single sitemap file after various tests suggested that the sitemap files themselves were sufficiently randomly constructed), and pretty much fully confirmed my findings, as well as extended the analysis to look at a few other possible explanations. They found 3.5 million users with 50+ posts ever and also posted within the past 30 days:
https://www.stonetemple.com/real-numbers-for-the-activity-on...
I will note that this is G+ and not Facebook.
I did another analysis, looking at content posted online as determined through selected keyword searches, limited to domains, and using Google's reported result count as an indicator of total results. There are numerous possible issues with that methodology, though the errors should be reasonably consistent.
Basically, I looked for a common English word ("this"), a set of probable indicators for substantive conversation (the Foreign Policy "Top 100 Global Thinkers" list from 2015), and an arbitrarily selected string suggesting less-than-substantive conversation "kim kardashian". I searched for these across roughly 100 domains, TLDs, CCTLDs, etc.
https://www.reddit.com/r/dredmorbius/comments/3hp41w/trackin...
Facebook reported 2.6 billion results for "this", which should be a rough proxy for total public user posts.
Twitter was about 1.2 billion. Reddit and G+ about 1/10th as much: 120k and 160k, roughly.
That would suggest Facebook has at a public posting level at least 10x the volume of G+ -- something like 30-50 million accounts.
I also calculated an FP/1000 score (FP top 100 results per 1,000 pages), and ... somewhere in my records ... an FP:KK ratio -- the relative ratio of FP top 100 results per mention of Ms. Kardashian.
Metafilter, Wordpress, and Reddit all ranked quite highly on that last.
I also looked at similar scores for media sites (Financial Times and The Economist do well), alternative media (generally better than mainstream), educational, and other classifications (Harvard has an impressive online presence, along with several other private schools -- major public universities such as the University of California ranked surprisingly low compared to expectations).