Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

nytimes.com

Post reply on HN
Print view View original

111–120 of 138 posts

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#111

Post by mswen »

resters wrote: post #46

This is how marketing works. They could have used music taste or religious affiliation or nearly any of the data add-ons Experian sells. This isn’t sinister in any way, merely clever as it seems marketers had been behind the times. There are entire magazines and television shows that have equally strong affiliation. If spend on fashion related stuff was smart, it was because it was a bargain compared to the many othe…

>"Marketers have been behind the times"

I have to laugh. I learned psychographics from Bill Wells back in the mid 1990s. He had come back into academia in after spending years at DDB Needham creating one of the first psychographic initiatives/programs that had teams of statisticians, social scientists and used mainframes for statistical analysis in support of product design, message refinement and targeted marketing and advertising based on "personas".

One of the fascinating things to me was that Bill told of doing a summer project during his own graduate school experience at Stanford in which it took him a whole summer to manually (using a mechanical calculator) calculate a Principle Components Analysis on some consumer data.

It boggles my mind in some ways that we have a bunch of CS trained start-up guys who are rediscovering both techniques and consumer behavior research that is sometimes 30 or 40 years old and think they have invented something new! OK, I am sounding like an old fart.

Now, I will admit that many marketing executives I have met over the years were not really comfortable with quantitative approaches but there is a whole subset that have been applying data science types of approaches for a very long time.

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#112

Post by SlowRobotAhead »

elliekelly wrote: post #101
axaxs wrote: post #41

Earlier quoted context omitted.

Not sure why downvoted, I agree. A 'western' and 'outdoorsy' company tends to have conservative customers, I think any 12 year old could tell you that. I'd rather see unexpected/unexplained trends, things the average person wouldn't already assume.

I think it actually goes a bit deeper than using certain brands to identify political leanings. That was step one, but they also rated users' openness, conscientiousness, extraversion, agreeableness, neuroticism, depression, anger and vulnerability.[1] Then they showed targeted messages to people they decided were right leaning and vulnerable to their messaging. [1] https://www.cnn.com/style/article/christopher-wylie…

Yea... they did all that... then fell all the way back to “who buys wrangler jeans?”.

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#113

Post by bsder »

AndrewKemendo wrote: post #76
chrisco255 wrote: post #65

Are you telling me that people who buy camo clothes are more likely to be conservative? That's crazy talk. Next thing you'll tell me is people who dye their hair pink or blue are more likely to be liberal.

Next thing you'll tell me is people who dye their hair pink or blue are more likely to be liberal. I have been curious about this particular recent feature. Does brightly colored/neon hair indicate the person is part of some particular group? My exposure to this trend has been primarily with people on the internet, specifically Twitter and it seems to be predominantly women, and predominantly associated with being ou…

> Does brightly colored/neon hair indicate the person is part of some particular group?

Yes. The "I'm non-conformist and quirky!" girl.

Of course, they're "quirky" exactly the same way as all the other "quirky" girls.

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#114

Post by IgorPartola »

tobr wrote: post #94
IgorPartola wrote: post #80

Earlier quoted context omitted.

Look, if you take a group of 100 people who are into Warhammer 40k, chances are that at least 80% of them will have played StarCraft too. Collapsing that group into a single person and calling her Wendy helps some people reason about them. If you prefer to think of them a Venn diagrams and that helps you sleep at night, no crime in that.

But you’re not learning that from a persona, you’re learning it from talking to 100 real people. You can put that into a persona if you like, but it won’t give you any new knowledge.

A persona is a heuristic model, that’s all. It seems you feel like it’s dehumanizing. Would it make you feel better if we called it “human behavior model 748263”? There is nothing inherently bad about saying “if you like video games, you likely own a gaming PC or a console”? What if you call that association “748263”? What if instead of that number you called it “Wendy”? Does any of this make any difference to you?

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#115

Post by sizzle »

tobr wrote: post #95
tlow wrote: post #82

Earlier quoted context omitted.

This is not necessarily the case and makes light of an entire field. Often times personas are developed by ethnographers, human centered designers, design researchers and are based upon actual users or prospective users. Furthermore, data analytics can be used to generate personas. Personas can be representative of key forks in behavior patterns or beliefs and are a useful tool when making design and product decision…

I’m aware that personas are often based upon insights from real people, but I question the value of repackaging the insights as a new, made up person. It introduces distractions and invites stereotypes. Just use real quotes and observations, statistics if it makes sense, and focus on describing typical or recurring situations or behaviors that are relevant to what you’re using the research for. A hair color or favori…

"A hair color or favorite snack isn’t going to be helpful."

It's about humanizing the product design process, don't pay any mind to extraneous information like hair color etc. At the end of the day, personas should be representative of the people you are developing products for and that's inherently valuable to the product development process and all the people involved who don't have a chance to be in front of users and get a feel for their needs, which determine how successful your product will be when they are aligned with what you are coding.

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#116

Post by autokad »

tomrod wrote: post #87
autokad wrote: post #11

Earlier quoted context omitted.

one interesting one was that I took 5000 twitter profile photos and averaged the pixel values by male / female. unsurprisingly, the male average had short hair while the woman long. but what did surprise me was that the shirt area of the males had a blue hue and the women had a red one. its not like anyone in particular was wearing red/blue by sex, but on average they were.

Cool! Could you share the results?

https://twitter.com/autokad/status/689896734542069760

https://twitter.com/autokad/status/689897031414886400

keep in mind these profile pictures are not centered, some are group photos, etc. its not 'super strong' but it seems evident to me, let me know what you think

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#117

Post by spiralganglion »

bsder wrote: post #113
AndrewKemendo wrote: post #76

Earlier quoted context omitted.

Next thing you'll tell me is people who dye their hair pink or blue are more likely to be liberal. I have been curious about this particular recent feature. Does brightly colored/neon hair indicate the person is part of some particular group? My exposure to this trend has been primarily with people on the internet, specifically Twitter and it seems to be predominantly women, and predominantly associated with being ou…

> Does brightly colored/neon hair indicate the person is part of some particular group? Yes. The "I'm non-conformist and quirky!" girl. Of course, they're "quirky" exactly the same way as all the other "quirky" girls.

This is my social group. Neither I, nor my friends, dye our hair because we want to be non-conformist or quirky — we dye our hair because it looks cute. It’s not at all difficult to think of fashion choices that’d be more distinct, but it’s often the concern that we’ll be seen as “weird” that keeps us from choosing them. I look how I do because I want to be liked and respected, same as anyone making the choice to be fashionable.

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#118

Post by tobr »

andrewflnr wrote: post #106
tobr wrote: post #94

Earlier quoted context omitted.

But you’re not learning that from a persona, you’re learning it from talking to 100 real people. You can put that into a persona if you like, but it won’t give you any new knowledge.

It's very common for a new perspective on knowledge you already have to enable breakthroughs in deriving new knowledge from it. In particular, imagining a person rather than a statistical abstraction will tend to activate useful neural wetware that simulates other humans.

Agree 100%, but I don’t understand how that leads to personas. Just talk about real people instead and you can get all of these benefits.

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#119

Post by downandout »

TheOtherHobbes wrote: post #67
downandout wrote: post #48

Earlier quoted context omitted.

> It was definitely effective and stretched ad dollars further. The evidence seems to indicate that not only is the technique not effective, but that it wasn't actually used in the campaign. More than one article, in traditionally liberal publications that would normally be on his side, refers to Mr. Wylie's techniques as "snake oil". Also remember that we only know about Cambridge Analytica because Mr. Wylie announc…

Everything you wrote is wrong - at least as far as CA's contribution to the Brexit vote. In reality CA used very specific targeted advertising on FB. https://mobile.twitter.com/Nealb2010/status/1059068463933743... The point about these campaign structures isn't that they're generically effective. In fact they don't need to be generically effective. They just need to be able to swing a vote by a few percent to define…

I didn't mention Brexit, I was talking about the election that made this Wylie character famous. Nevertheless, again, despite the claims in random Twitter posts here and there, mountains of evidence point to psychographic profiling being largely ineffective regardless of what type of vote it is being used on. “Swinging the vote by a few percent” is a major undertaking. It is not “trivially easy to do,” as you put it, certainly not through scientifically dubious methods that (in the case of the 2016 election) are not even actually used.

Mr. Wylie is a programmer that violated confidentiality agreements with his employer to try to gain notoriety and presumably raise his income. That will make him difficult to employ ever again in any traditional setting, even among those that are sympathetic to his political beliefs. He seems to believe, or at least wants others to believe, that he is today’s Edward Snowden - even though the two couldn’t be more different. I don’t know how well his monetization efforts are working out, but based on the fact that he is now making clickbait-type public statements that can be easily debunked to get media coverage, it seems that it may not have worked out as he planned.

Re: Cambridge Analytica Used Fashion Tastes to Identify Right-Wing Voters

#120

Post by ashildr »

walshemj wrote: post #108
Jedi72 wrote: post #9

Earlier quoted context omitted.

It was hilarious as an engineer who needed to learn marketing 101 for a startup, its pretty much just stereotyping your customers. "Wendy is a 24-30 yr old professional, slightly insecure, straight, likes wine and dogs."

It's all ok in marketing until you have "the time" :-) Btw this is what the "marketing gels" at Burberry still refer to the time when chavs got into Burberry check baseball caps. And this is a direct quote from a mate who works for a big 4 consultancy "gels" being the upper class young women who will be working there. I suspect that Mr Bannon wearing Barbour isn't exactly taken well by their marketing team.

Gels?
Post reply on HN