In reply to Streetwiseguy :
The algorithm doesn't need to be super smart to pull people in. Typically ads on Facebook and Google are just designed to display what's relevant to the user at any given moment in hopes you'll click on it to investigate. They're really only good for items which you can afford to throw away 99% of the monetized clicks. Sidebar: A 1% conversion rate is considered pretty good.
Amazon's ad strategy of just display random crap you looked up is different. They get a cut or handlers fee for whatever you buy so they'll just keep throwing E36 M3 at the wall to see what sticks. Amazon can do this because they control the ad delivery vehicle, rather than Google or Facebook, so it costs next to nothing for them to deliver the ad from AWS. Obviously it isn't smart as items are often something you purchased or haven't looked at in weeks but it costs virtually nothing to take the chance.
Steaks on the other hand, Omaha and others know that you're just going to drive over to Ralphs and buy whatever is on sale there when you want it so there's no point in trying to follow you around to sell you a subscription based on a known interest in steak (eg. Searched for recipies.) They might, however, display an ad if you're searching "how to buy half a cow" on Google because there's a good probability you want recurring steak in your freezer.
You guys aren't thinking insidiously enough. The issue at hand wasn't that they were trying to sell shoes or steaks it's that they were trying reduce the unknown quantity of middle of the road voters by appealing to a single issue which would flip the switch for their candidate without being forthcoming about who was supporting the effort. CA was building highly scientific psychological profiles on individual users under the guise of a personality test to accomplish this. Since 68 percent of Americans get their news in some form from Facebook, especially those with less education, and Facebook is designed as an echo chamber, the personality test identified the blocks of voters as well as their first and second level connections which the resulting data suggested could be swayed with targeted messages.
That's why so many of the ads and fake groups were mostly tailored as: "She's coming for your guns" and "Hillary belongs in jail for Benghazi." Serious personal story here about how this works. I was was sitting in a remote FOB in Afghanistan when the Benghazi situation took place. There were definitely some thoughts of "Holy E36 M3 would the President and Secretary of State seriously also let us out to dry if it was politically expedient?" If I didn't go out of my way to read WaPo, CNN, Quartz, and NYT in addition to whatever appeared in the Google newsfeed (mainly because I was aware of Red Feed Blue Feed through the tech community and users here (Thanks ArsTechnica and AlfaDriver in particular)) I'd probably have fallen into that trap too reading only right leaning publications.
The issue at hand is that various interested parties, both CA and Russia, were trying to gain more sway within the government by affecting the elections's outcome with a candidate partial to their interests. That's a lot different than "Look at these flashy shoes." The data mined from Facebook potentially allowed this to happen so, yes, I truly believe there needs to be an investigation and some fines for losing control of the data leveraged.
As to if CA had a real effect on the election, I'm not the data scientist analyzing the results, nor do I know if the data can be tickled enough to get some meaningful results on why people voted the way they did, but I'd bet there was at least some effect. I saw first hand how the rise of Breitbart articles through Facebook polarized my coworkers against Hillary Clinton but not necessarily for Trump, as well as saw what proper use of data analytics did for President Obama in 2012, so I'm certainly not discounting the probability.
edit: I'll look at this again in the morning I have a feeling I'm half coherently rambling because work and my kids wore me out.