The New Yorker article was odd to me for a New Yorker article, it had a lot of historical context without a lot of details of what Simulmatics actually did for the Kennedy campaign. Perhaps there isn't a lot of historical record/people who want to talk about it. I was left feeing like I had learned that the Simulmatics founders had an idea, and successfully sold it to the Kennedy campaign and made some money, but not sure to what extent it actually provided value in that actual campaign.
With some people suggesting the same might be true of Cambridge Analytica.... the more things change? (I have no opinion on what value either CA or Simulmatics actually provided, I don't know enough).
FYI, these are all excerpts or reviews of the same book by the historian Jill Lepore. (I've quite enjoyed her writing over the years and highly recommend it.)
I've changed the title to be that of the book under review. Actually we've started doing that for most book review posts (e.g. https://news.ycombinator.com/item?id=24474073). It seems to make for better discussion.
What’s fascinating is that prior to CA helping targeting in the previous election cycle, this capability was kind of embraced by politicians and industry as the next evolution in political campaigning... but few recognized an issue with that kind of data usage at the time...
I remember media singing praises for Obama campaign about how effective they were with social media. Using that to contrast him to mccain who apprently didn't know how to use email.
This realization really pushed me away from mainstream media entirely. I explicitly remember and can still find articles about Obama's campaign being cutting edge by using "behavioral data" (https://theweek.com/articles/451328/how-obama-won-internet). Yet when the same actions were taken by conservative candidates it was ridiculed. As an independent I just couldn't trust these sources any longer.
Yes the methods of data collection were slightly different between Obama's campaign and what Cambridge Analytica did and sold to conservative candidates. Obama's campaign claimed they didn't do as much with the data and it was within the FB ToS, but these claims don't change the ethics and since 2004 both parties have been building their data infrastructure out to sway voters (https://slate.com/news-and-politics/2012/10/obamas-secret-we...).
Both parties are nefarious in their use of data and attempt to persuade voters through micro-targeted outrage, and it makes me wary of ever voting for either of the two parties' candidate.
The solution is to not allow targeting for political ads. Costs will go up as you'll be competing for same eyes, and you'll have to spend more to reach everyone you want - plus has added benefit of allowing to see what messaging and narratives opposition is perpetuating.
How is this even remotely practical? You're only allowed to have political ads on nationally shown TV programs? Choosing to show an ad in Wisconsin and not California is "targeting".
I think only allowing the advert to be targeted at the level that the campaign is being run in is reasonable.
i.e. if you are campaigning for Governor of NY, you can target New York state, but you can't target Manhattan, or Long Island, or men over 50, or people who liked the NRA.
So about the only allowed form of advertising is mail addressed to "Occupant" sent to every postal address that is in the jurisdiction the office represents?
Nearly everything else will generally only reach a subset of the voters, and that subset will have known demographic differences from the overall voter demographics for the jurisdiction.
Run an ad at in the 8-8:30 PM Sunday slot on every FOX TV station in a state, for example, and you'll get quite a different demographic than if you ran an ad on those same stations on Sunday afternoon. The first ad would be during "The Simpsons" and the second would probably be during an NFL football game.
that would kill list building and fundraising which are large digital expenditures. non-presidentials barely spend on digital persuasion but they will spend on fundraising if it raises
Doing so would destroy the ability of non-establishment candidates representing salient issues, e.g. police violence in particular communities, from effectively running grassroots campaigns.
It might be a solution although I think regulating this specific form of political ads would have unintended negative consequences.
I think things like hard cap campaign contribution limits (as in overall campaign fundraising cap rather than by donor type) could do more good to even the playing field. And it should reduce the web of "IOUs" that elected officials come into office with.
The choice of what issues to run on is in itself targetting regardless of where it is run. I am not sure what good that would even do given it is already well known that political stances at rallies vary by venue.
Interesting question arises - does advertising to Republicans or Democrats cost more or less than the other? Either way the mechanism would be good to level the playing field further.
That's a good question, the only thing I could find (and take it with a grain of salt) was that Trump spent $5 per vote compared to Clinton's $10 per vote if you assume all the money they raised was spent. https://www.reuters.com/article/us-usa-election-spending/at-...
It is interesting to think of campaign efficiency (or propaganda efficiency lol) as CPV (cost per vote) to make it parallel to advertising metrics like CPC and CPI.
Is there any evidence the Obama campaign did what Cambridge Analytica/Trump was being criticized for?
Which was harvest private data (including messages and news feed) from Facebook users and their friends via an app that was ostensibly non-political, and then in turn sell it to a super PAC that for microtargeting. I'll note that the super PAC involved was likely coordinating with a campaign in violation of FEC campaign finance laws.
The CA scandal resulted in a $5 billion fine to Facebook by the FTC.
I think it's disingenuous to say what the Obama campaign and Trump/Cambridge did were close to the same.
Cambridge Analytica were accused of taking data mainly on what pages users 'Liked' - not private messages or news feed. (Unless you have a source?)
Obama campaign boasted about having access to the entire friends network. In other words, they know who each person was and what the connections between them were. Facebook openly said that they helped the Obama campaign get and utilise 100% of the friends network data on Facebook.
Fairly sure that's not anywhere near the whole picture of what CA was accused of? I thought CA ran surveys and from that data could make implications based on the friends and friends if friends of the person who took the survey?
My understanding is the quizzes were just viral Facebook apps to get API authorizations which back then let you download a lot of friends' basic data.
Simple scraping would get you a lot of the same data back then but it would be more involved and less solid legally. CA just figured out an easy way to do a huge scrape of Facebook.
I'd be surprised if many other organizations out there aren't sitting on similar datasets.
Even without asking any psychometric questions you could easily seed this through the dataset based on just a few examples or ground truths. This part seems rather unspectacular. With all the likes you can already tell who's alike. You just need to label the clusters.
CA's sell was this data would allow them to target super precisely and send different messages to different groups. But there's a big bottleneck at message creation. If you have 10 ads you can only target 10 groups uniquely. I'd be much more impressed if CA actually had software to manage ad creation, placement and performance tracking.
If that disclosure statement means what it is apparently meant to imply, it would of course be substantially more interesting to hear what you know, rather than talking about accusations, no?
But if you actually want a refresher, and for anyone else,
Well I'm not willing to provide any proof of my identity, so you have no reason to take what I say as legitimate. Indeed, I've seen lengthy posts of utter nonsense from others who claimed to work at CA.
Usually, even the generally available facts are a lot better than most people's understandings of the situation. For example, most people seem to think that this scandal proved that Facebook sell their user data, whereas it actually is alleged that CA took advantage of the well-documented public API that Facebook used to have, where any app vendor could download data on friends of friends who had given permissions to the app.
If you have specific questions, I would be happy to point you in the right direction.
Ya, the double standard is obvious. I laughed when I read all these articles about Trump which were the same about Obama on this issue, but with a completely different tone.
The Obama administration was praised for gathering the same information about users without their consent. Obama's campaign manager said that Facebook told them that if it was anyone else gathering that much information, they would have blocked it, but that Facebook was "on their side". There's absolutely a double standard.
Do you have a source for that? Your comment is the top Google result when I search for that "on their side" quote with various nouns about Obama and Facebook data collection.
EDIT: This was a few spots down in the search results and explains how Obama's data collection was different than CA.
The issue was never that CA did this with "false pretences" (they didn't sign up users directly, that was done by an academic who then sold the data) or that by signing users up on your own website, you were doing some above board. The issue was that Facebook allowed access to not only your account but your friend network. That was what Obama used, that is what CA used, that is why both tried to get the Facebook login, and, btw, the end goal was exactly the same: microtargeting of ads.
Btw, there are probably hundreds of other companies who are doing the same thing. A lot of the "data science" startups are using this data in their products, there was an Israeli startup (mentioned here recently) that somehow had click data from Google, individual search data from Google...this isn't just politics, this personalised, identifying data that is somehow everywhere.
Does someone not being in jail somehow prove what they did is legal? Is someone always held individually liable for corporate misdeeds?
There have been multiple fines related to this scandal from multiple governments. There have been Congressional hearings on it in the US. There have been numerous lawsuits. I did say "potentially" in that first comment because I can't say definitively that they acted illegally, but there is a lot of smoke here. If people are going to make the accusation that it is a double standard, where is the smoke surrounding Obama's use of data?
the phrase "potentially illegal" makes no sense to me.
I can claim that your comment is "potentially illegal" ..means nothing.
> There have been Congressional hearings on it in the US. There have been numerous lawsuits.
doesn't mean its illegal. Its just politics to discredit opposition's election victory.
This is equivalent to "hillary's emails" which also had lawsuits and congressional hearings. Now no one cares about it. Same with CA, no one will hear about it once democrats win.
> where is the smoke surrounding Obama's use of data?
There isn't because republicans didn't think it would be politically useful like "hillary's emails".
I like how you pulled out that one piece that is politically motivated without addressing the more apolitical fines or lawsuits. None of that has existed in relation to Obama as far as I'm aware. And it isn't like Republicans would let Obama slide on something if there was any chance to call him out for it.
The main problem with CA was that they tricked users into giving up FB data, pretended to be a college research study, violated FB's terms, and then when FB told them to get rid of the data they didn't.
Even if we accept the idea that CA was decisive to Trump's victory (I don't think it was, not even close), I think that you are absolutely on point here. I do think democrats were caught by surprise in 2016 and, instead of blaming themselves for their debacle, they sourced it elsewhere.
This doesn't show it worked. It shows that CA is connected to other similar companies either through a direct relationship or a tortured, barely tenuous connection (the slide on how some employees went to work for other companies in the same space is silly). Regardless, at no point does it attempt to even show that anything that CA did made one lick of a difference or that anything they did was even unprecedented in the space. For example, I don't believe for a second that Democratic campaigns haven't been using similar social media aggregation techniques in present and past elections. I even remember glowing articles written about how the Obama campaign was able to harness social media for purposes of political targeting and messaging.
So what are we talking about here?
And the entire thing is even more ridiculous in the larger context of how incompetent and disorganized the Trump 2016 campaign was.
Meanwhile, one aspect of the 2016 presidential coverage that drives me nuts is the omission that HRC's voter file was stolen and likely used by GOP, Trump, Russia.
Meaning they knew exactly who to target.
A campaign's voter file is it's Crown Jewels, it's secret sauce. It is the campaign.
HRC knew they had a problem in MN, WI, PA. They just couldn't figure out what was going on in time. Plus a zillion other things going on.
Again, will read your link soonest. It might help me fill in a few more of the puzzle pieces.
> HRC knew they had a problem in MN, WI, PA. They just couldn't figure out what was going on in time.
Yeah, the problem is that they took those states (and it’s MI not MN) for granted as safe blue states (which they had been for decades) and bought into the “Hillary is so far ahead in the polls she doesn’t even think about Trump anymore”[0] hype that they didn’t see the need to campaign there.
Okay. I quickly scanned the linked graphcommons presentation (?). Instead of responding to that directly, I'll share the thesis I've been chewing for a while.
I worked on election integrity stuff for about a decade. Opposing the touchscreens, tabulators, and other blackboxes. Advocating for better administration. That kind of stuff.
Like with all learning, my understanding of the problem dramatically changed over time.
The biggest threat to election integrity is change. With great reluctance I came to the view that any future changes had to be slow-rolled, even the phase out of the touchscreens. Think of it as how first responders stabilize the patient before transporting that patient.
The second biggest threat is noise. It's the asymmetry of bullshit principle. It's just so fucking hard to figure out what's going on. We spend all our time chasing wild gooses. So much slips past our attentions.
Only then can I think of the actually machinery of democracy. Stuff like form of elections, procedures, transparency, etc.
This frame and my own campaign work (stumping, GOTV, etc) informs how I regard CA and others.
The biggest threat CA poses is Steve Bannon's "flood the zone" with bullshit strategy. With so much activity below the radar, it's impossible to talk rationally about this stuff. Because there's always more to the story. Which is the trap, the exact purpose of Bannon's strategy.
The skeptics say CA, social media, etc. didn't effect voter behavior, specifically costing HRC the election.
That's entirely besides the point.
Other factors absolutely moved voters. We could stack rank them. Comey's betrayal is probably the biggest. My two personal faves are HRC's reliance on negative partisanship (too many attack ads), and HRC's desperate need to tangle with corporate media (instead of just creating HillaryTV and speaking directly to voters like Trump did with Twitter).
But there are many fights in the margins that most commentators completely ignore or outright dismiss. What we election integrity people called "death by a thousand cuts." It's never just one thing that makes the difference. That's too obvious and will eventually be corrected. Rather it's the cumulative damage of 1,000s of small attacks. Asymmetric war fare.
By keeping the contest "close", it becomes possible for the jitter (noise) to make the difference, depending on timing of the snapshot (when you measure).
And now we get to what CA and others did. Buried in all the noise, they used HRC's stolen voter file to precisely target and overwhelm "undecided" and low motivation voters. Maybe 500,000 voters in total. Tiny percentage of the electorate.
And Trump got lucky. It was just barely enough. Smaller in impact than say Jill Stein's spoiler. But on top of every thing else, it made the difference.
("Undecided" in quotes, because that's another topic onto itself.)
I have only seen this concept on HN and not elsewhere. Can you explain what you mean? What about CA “didn’t appear to work”? I went researching just now and didn’t find anything online to back up that vague opinion, so what exactly do you mean?
There is certainly no way to know to what extent CA content changed voter patterns and actions. I don’t understand how you could be so confident you are right.
However, if CA is capable of delivering one should view them as more than “just“ a business... Perhaps CA’s owners found more value in not helping Cruz win?
The key feature of CA is that they were affiliated with the Trump campaign and therefore were made to be the scapegoats for what EVERYONE was doing. And yes, what they were doing specifically was a nothing burger.
Cambridge Analytica was a pissant compared to the data political parties easily buy from advertising data companies. The whole controversy surrounding them is like getting upset about grocery store discount cards but ignoring that Facebook exists.
Also, campaigns don't have to buy or amass that data anymore. Facebook is the largest purchaser of consumer data. So now if you use Facebook for campaign ads you get all that "for free".
Lastly, wait until you find out what the NSA knows about you. Surely that data would never be used for nefarious purposes...
And a couple subsequent letters from readers on the article:
https://www.newyorker.com/magazine/2020/09/07/letters-from-t...
The New Yorker article was odd to me for a New Yorker article, it had a lot of historical context without a lot of details of what Simulmatics actually did for the Kennedy campaign. Perhaps there isn't a lot of historical record/people who want to talk about it. I was left feeing like I had learned that the Simulmatics founders had an idea, and successfully sold it to the Kennedy campaign and made some money, but not sure to what extent it actually provided value in that actual campaign.
With some people suggesting the same might be true of Cambridge Analytica.... the more things change? (I have no opinion on what value either CA or Simulmatics actually provided, I don't know enough).
I haven't listened to/read the NPR piece yet.