Hacker News Reader

> Sales would say anything to convince buyers (campaigns) and had no idea how the technology works. You can take any sentences uttered by them and quote it to the data scientists, their reaction would be… bow, shake their head, “Yeah, not, that’s… not it.”

This is frankly par for the course in data science. I work as a data scientist/statistician for a relatively newly established (~5 years) analytics team at a large company, and the first order of business was weaning our teams off of the snake oil that grifter consulting companies were making millions selling them. The things I've seen referred to as "machine learning" and "AI" are absolutely bonkers. The reality is that most of the analytics services being sold B2B is total garbage, and yet it thrives because most customers aren't aware of why its bad.

To be fair to these consulting companies, my impression is at least some of the time they aren't knowingly selling BS, and that often times they just have very poor analysts/data scientists who have a very surface level knowledge of what they are doing, but not enough depth to realize why they can't do/aren't doing what they think they are. And the actual sales people know literally nothing but buzzwords.

These days we've been able to largely get the company away from vendors where possible and to rely on internal analytics teams, but it is still an issue when our team will come back to business partners with an answer someone doesn't want to hear, and they have external vendors essentially promising they can solve literally any problem with AI, and magically always produce the safe answer the customer wants to hear (but they never are willing to share code, the low level details of their process, or really anything that can make their analyses reproducible).

Relevant quote from TFA: Dave Karpf, who has written two books on the use of data in modern campaigns, calls Cambridge Analytica “the Theranos of political data.” Eight GOP political consultants told Ad Age’s Kate Kaye that the company was “all hat and no cattle.”

Maybe it's because I'm from Texas but "all hat and no cattle" is about as damning an insult as they come. It's the equivalent of "full of shit".

Or as they would say in Cambridge, "all mouth and no trousers"

Or as they say in Yorkshire: all fur coat and no knickers

Apparently they would be more likely to say "all mouth and trousers", which is believed to be a corruption of the more logical but less commonly heard "all mouth and no trousers" [1].

[1] https://www.phrases.org.uk/bulletin_board/19/messages/511.ht...

Never heard that saying once in Cambridge, lived here 20 yrs

Lol, well you must never have run into me either between 1990-93 or again 1999-2002

Michael Lewis' Podcast `Against the rules` did a good job in bringing a different perspective of the entire thing Kogan did with data was not significant yet he was made a scapegoat by journalists https://atrpodcast.com/episodes/the-alex-kogan-experience-s1...

That is indeed a very interesting perspective

Ignoring the unrelated 30-min digression about grammar (sic), I’m not tempted to excuse him: he shared personal-level data, including names, with a company that was openly Orwellian. They didn’t hide what they were doing. People started paying attention when they went from sowing division and crafting racist campaign in South-East Asia and Africa to Western countries, but they did not hide their methods before or after. I guess the podcast is making a good case that he’s awkward and naive but that shouldn’t prevent him from seeing the consequence of his sharing his research data.

The biased story seems to ignore that Facebook did ask CA to delete the data as soon as they knew about the sale. I can’t imagine that Kogan didn’t hear about that. On that, I wouldn’t be able to say anything more relevant than the last sentence of the podcast on “creating your own reality”.

Kogan’s avoiding public opinion and his self-excuses are blocking the entire research on social networks. Now, Facebook PR expects that any similar academic collaboration will end badly — and I’m not surprised that those are stalled.

> There was a moat between the operations/sales team... and the technical team.

It's one thing to claim two teams _don't_ work together regularly; it's another to say there's a moat between them, indicating it was difficult or impossible for the teams to interact due to an external factor designed into the organization's structure. What does this moat mean practically, and why is this claimed division pertinent to the points the documentary makes?

Intellectual moat: the sales team refused to acknowledge that over-selling was going to hurt them in the long run; the technical team regularly asked them to re-phrase copy on sales material that was not technically true, the sales would say yes and not do it, or change the written statement but make the false claim, verbatim, in front of cameras. Numbers in their presentations didn’t add up, they confused information and inferences, interpolation, models and individual data, etc.

there’s an open-source model for parsing Google Street View, classifying the car parked and making precinct-level predictions

Mind blown. The lengths campaigns will go to for the win. Can you share any more stories like that?

> The lengths campaigns will go to for the win.

In general, if there's enough money to use for a specific goal, humans will go to great lengths to win. Look at finance companies building microwave towers to lower trade latency, hiring the top minds for salaries in the millions, building in-house supercomputers, etc...

Obama famously used technology in his campaign, why people are surprised that the other side is using it is beyond me.

Also, while individuals can make unique decisions, in general humans are very predictable, which is why fields like economics and sociology are even able to exist.

> Obama famously used technology in his campaign, why people are surprised that the other side is using it is beyond me.

I suspect a lot of the difference in judgement is less about the tools and more about the messaging that it allowed: What CA did, quite effectively under the brand SCL Election, was dog-whistling and demotivation campaign, like “Do So” in Trinidad presented in the Big Hack.

If you can tell which one of your voters are racist, you can be a lot more blunt about what you mean by “revisit immigration policies”. People suspect that targetting allowed Trump to talk as if no one else was in the room — although, it’s not like he was very respectful in public.

One the other hand, people are likely to assume that Obama’s analytics, even if they were more detailed, where about pointing potential voters are the relevant part of his program: you are a teacher? Let’s put the promises about increasing the budget for education in bold. Veteran? Talk about red-tape at the VA. Everybody can see those, but people might not care as much. This is still targeting, probably more effective than cloak-and-dagger messaging, but it doesn’t come off as problematic.

Of course, a lot of the tools are similar: scraping Facebook Graph should be frowned upon in both cases, it wasn’t in 2008, 2012; it was in 2016, although it’s hard to see if that’s more awareness, and from whom, or partisanship.

I do think that the distinction is key, and why Facebook’s main visible effort was to make all political ads publicly visible: it avoided to let the argument about secret offensive ads fester. That might have some unintended consequences, though.

I read your supposed differences and see none. Both Democrats and Republicans harvest information voters thought was private to microtarget their message to people. The only difference seems to be you dislike the message that was then presented. But that isn't a problem with the methodology, thats a problem with the message. If your point is that Trump ran on a campaign of racism and division, fine, I'm with you on that. But the idea that the Trump campaign utilization of Cambridge Analytica was scandalous and the Obama campaign's analytic efforts that do the exact same thing is somehow above water is ridiculous.

Either harvesting people's data and using it to message to them is a problem or it isn't. The message being sent is a distinct issue.

> If your point is that Trump ran on a campaign of racism and division, fine, I'm with you on that.

What I find rather funny, were you to ask a Trump voter you'd likely hear them say the Democrats are the ones running divisive, racist campaigns, and that they've previously pioneered digital advertising in the political space only shows their pot-kettle policies in regards to the successful Trump campaign.

Your counterpoint is finely drawn. The problem is that many people say "Obama did it too!" when they're in favor of the content of Trump's messaging. The act of using finely targeted data to drive racism, hate, and falsehood is the issue - you are opposed to that, but others aren't.

You are correct the message being sent is the issue.

- Do you tell population a and population b contradictory things?

- Do you spread what you know is misinformation?

- Do you attempt to manipulate someone into working against their own interests. For example tricking a population whose interests and preference can best be expressed as A over B over C into voting for A so that C can win.

- Is your messaging a net harm to society? Deliberately playing up racial tensions to both sides determent for votes seems to fit the bill.

- Are you actual intents hidden because your interests are actually at odds with your target and they would act materially different if they understood your actual motivation?

The democratic usage of data appears to be primarily based on figuring out who could effectively message and the republican usage seems to be figuring out who they could manipulate and harm.

It's not that hard to see the difference.

For the car to politics: https://news.stanford.edu/2017/11/28/neighborhoods-cars-indi...

For unexpected findings… without giving details, it’s fairly easy to stumble on religion and practice from super-market data. Interestingly, having religious data is banned in continental Europe. Cue to an embarrassing moment when consultants back out of a big project with retailers.

I’ve seen more fun stuff with betting and pay-day — some betting website have a score per football team, how much of their fans are salaried monthly, weekly, receive State-stipend, etc.

Overall, findings like those (say, parking occupancy and revenue) are generally done at Edge-funds so they are not known until long after the fact.

> there’s an open-source model for parsing Google Street View, classifying the car parked and making precinct-level predictions.

I tried searching for this but wasn't able to come up with anything. Does anybody know what this is referring to?

Edit: Found this in another comment https://news.stanford.edu/2017/11/28/neighborhoods-cars-indi...

As an outsider to me the worst was targeting people that were paranoid (also not in some technical psychological sense, again as an outsider) while violating agreements and exploiting that for elections.

That would be the worst if it happened.

I don’t think that Cambridge Analytica had the ability to do that.

I appreciate your POV, thanks. I do think psychographic profiling was pretty roundly proven by reporting and testimony at this point though. You might be interested in this piece from today though, it takes a bigger view outside CA as well:

"It was somewhat surprising that Facebook succeeded in hogging the limelight in a field so rich with villains. The bigger picture…was the way in which the Cambridge Analytica story opened a window onto a new constellation of international billionaires, corrupt politicians, and war profiteers who were apparently amassing enormous power. That story isn’t only about technology, data, and psychographic profiling; it’s also, at root, a story about the consequences of entrenched economic inequality, the privatization of essential public assets and government functions, including even national security, and the challenge to conventional foreign policy posed by the bargains being struck between international kleptocrats…"

https://www.nybooks.com/daily/2019/08/27/the-oligarch-threat...