a thoughtful web.
Good ideas and conversation. No ads, no tracking.   Login or Take a Tour!
comment by b_b
b_b  ·  4866 days ago  ·  link  ·    ·  parent  ·  post: "Choose your friends wisely, for they may someday determine your APR."
Let's envision this scenario: Against odds, a poor black kid from the ghetto goes to college, graduates, gets a job and wants to buy a house. The mortgage banker looks at him and says, "Son, your always pay your bills and have a stable income, but your friends all wear do-rags and have multiple priors. I'm sorry; I wish there was something more I could do." WTF is the function of social media anyway?




kleinbl00  ·  4866 days ago  ·  link  ·  
Q: Why do civil libertarians hate having their DNA and fingerprints in a public database?

A: Because once the information is available, a vendor will figure out a way to discriminate against you because of it.

It's not even evil on their fault. They're actuaries. They're just using the best data available to them to maximize their profits and minimize the costs to their customers.

Statistically speaking, the ghetto kid gone good is at a greater risk of running into financial difficulty than a WASP from Grosse Pointe who went to Exeter. The actuaries are just putting numbers to it. Genetic predisposition to breast cancer? Well of course you should pay higher insurance rates. That's just logic.

The problem is this:

Anything with a "rate" is a probability calculation in which the vendor is making a profit based on the risk he's taking on the recipient. The recipient feels he has a right to equal treatment while the vendor feels he has a right to discriminate. Until the government steps in and says "you're not allowed to know this" any vendor who operates on a statistical playing field is going to say "you'll get a better rate if you pee in the cup."

I could probably get a better rate if I let Progressive lojack my car. But I'm not about to. Know why?

Because fuck them, that's why.

b_b  ·  4866 days ago  ·  link  ·  
Statistically speaking, the ghetto kid gone good is at a greater risk of running into financial difficulty than a WASP from Grosse Pointe who went to Exeter. The actuaries are just putting numbers to it.

Therein is the problem with these types of actuarial data. They are purely correlative. How can these data be applied to this person? Usually, they can't. Being from a certain zip code, or having a certain race don't cause one to behave a certain way. We make correlative judgments all the time, but it doesn't make them correct.

The breast cancer example is qualitatively different, because there is a causal link between BRCA1 mutations and cancer risk. There is no causal link between the friends one keeps and whether mortgage payments are delivered on time. But even in the case of breast cancer, I think there is still a moral argument that can be made that a woman with that mutation shouldn't pay more for health insurance, since its a shared risk pool, and it was no fault of hers that she has it. The same applies to people from the ghetto. They didn't choose to be born there; only morons like me choose to live in the ghetto.

Playing with correlative data works well when shooting dice, but I don't think its a great idea to use it too liberally when talking about humans. But really what we're talking about here is people willingly giving copious amounts of personal data to companies. That, I suppose I, have little to say about, because its their own goddamn fault if it bites them in the ass. Can't save stupid.

kleinbl00  ·  4866 days ago  ·  link  ·  
All statistics are purely correlative. Your numbers are only as good as your model, and your model is imperfect by design. Any statistical data is an attempt to apply gestalt policy to individuals - if that worked, would we still be individuals?

As to the "causal link" and payments, you're simply mistaken. Play a game with me. Write down your five closest friends. Write down what you think they make in a year. Now add those five numbers together and divide by five. Now tell me that number.

I'm going to repeat that number back to you, and tell you your annual earnings.

You think because it isn't genetic, it isn't true?

It isn't their fault, either. Take customer loyalty cards. Did you know that simply having a Kroger card means that someone can subpoena all your shopping habits since you first got the card?

b_b  ·  4866 days ago  ·  link  ·  
I agree that statistics are correlative, but they are most useful when they can applied to learn about a causal relationship. Correlative data that don't tease out a causal relationship aren't informative or useful. Consider for example, there is a strong correlation between sales of hot chocolate and reported cases of frostbite. According to these superficial treatments of statistics, if we ban hot chocolate sales, we should curb the instances of frostbite. Obviously, this an extreme example that it doesn't take very much brain power to see through, but that's the purpose of evaluating in the extreme.

An easier and more precise way to determine my earnings/holdings would be to look at my W2s and bank statements. I think this is one of the reasons I'm sensitive to this issue. When I bought my home it was a huge pain in the ass and nearly didn't get funded, but only because of the zip code in which I bought it, not because of my income, lack of down payment, history of paying bills, or the like. Its very frustrating to be treated like a data point, when you're actually a human.

Also, I refuse to get a Kroger card for exactly that reason.

kleinbl00  ·  4866 days ago  ·  link  ·  
I'm not arguing with any of this, merely pointing out that if you live near a grocery store and want to save money, you're likely to get some sort of loyalty card (for the longest time, I used one I found in the parking lot, happily racking up rewards for a total stranger).

I'm also arguing that any data that can be collected will be collected and the responsibility for proper treatment of data that can be mined comes down to legislation and vendor practice, not the consumer. It's far too easy to hide implications and effects from the consumer and they're far too easy to bribe.