Comment by najirama | Hacker News Reader

Comment by najirama | Hacker News Reader

Parent Full thread

[Edit: spelling mistakes]

Based on your response, you do not even understand the terms I've been using.

'Set of classes,' means everything per this discussion - as the discussion has been premised on the statement you made (that skin color is an indicator of class).

So when you say things like...:

"If you're trying to predict someone's behavior, distinguishing someone who makes $59,999 from someone who makes $60,000 makes no sense."

..You suggest that you either do not understand my argument, or, you do not understand what has been meant by the word "class" in this discussion.

So I'll say it again; Classes are defined by an interval on the set of natural numbers. This isn't a discussion about guessing incomes with any greater precision than those provided to us by said classes.

The data provided to us by the census bureau document is the average. Note that the "households" in the census include BOTH family and non-family households.

So, when you say things like...:

"The poverty threshold for a family of four is $22,025. Black households have a "surplus" income of $12,193. White households have a surplus income of $33,505."

...You hint that you didn't quite understand that the average provided was calculated including single-households as well as non-single households with kids. So your whole reply is a bit of a straw-man...

...But it isn't even really a good one. We are talking about race as a predictor so let us control for income and work backwards. Imagine two families, one black, the other white, both living with an income of $34,218, both with four children. Is it then your contention that the Black family is still more likely to produce a thief?

If so, I do consider your mind boggled.

I have said several times that I'm not talking about discrete classes, yet you insist on arguing that point. I do not think that race can predict whether someone is lower, middle or upper class. I do think that race can predict income. I don't understand why you continue to tell me what I meant by class when I have told you what I actually meant three times.

I understand the data completely. The family of four example was an illustration of the marginal utility of income. The point is that not only are $55,530 and $34,218 quite different numerically, but they're also vastly different in terms of the luxuries a household can afford.

You've spent so much time arguing semantics that you haven't even tried to disprove my assertion that households earning $55,530 and $34,218 will likely have different behaviors. If my assertion is true, then race is a valuable predictor of behavior.

Controlling for income makes no sense, as my claim was about outward predictors of behavior. No, I don't think black families and white families with the same incomes will have significant differences in criminality. I think that data does actually show that blacks commit more crime than whites even when you control for income, but I don't think the difference is big enough for it to be a valuable predictor of behavior.

Good night.

[Edit: 'its' changed to 'opposite' for clarity, 'there' to 'their']

1. "I have said several times that I'm not talking about discrete classes, yet you insist on arguing that point."

Regardless of whether or not you meant to be "talking about discrete classes," you were. Socio-economic classes ARE discrete classes. So if you are talking about some other type of class, it is a mistake of comprehension on your behalf.

2. "I do not think that race can predict whether someone is lower, middle or upper class."

An explicit contradiction. I'll quote you directly:

"In America, skin color is a good predictor of class."

So which is it? Is race a good "predictor" of class or is it not?

3. "I do think that race can predict income."

This is the same thing as predicting class. Income is both necessary and sufficient for determining class, so if you could predict income by race, you could predict class by race. But you cannot do this (remember that guessing 'middle class' over and over is not allowed as it isn't information providing.).

4. "...you haven't even tried to disprove my assertion that households earning $55,530 and $34,218 will likely have different behaviors."

Ummm...that's because the opposite is taken for granted in the model of class. The middle class interval is traditionally considered to be from 30K to 75K - the class intervals are chosen precisely because the buying power in each strata is considered to be effectively equivalent. Again, if you don't understand this, then you do not understand the nature of classes - it just is what it is.

5. " If my assertion is true, then race is a valuable predictor of behavior."

No. Even if your assertion is true, all it shows is that income is a predictor of class. Well, duh - classes are defined by incomes. Behavior has no relation to any of these variables - that is, bad behaviors are distributed in equal proportions across the class strata. Income may predict risk-levels associated with bad-behaviors, but not necessarily their frequency.

6. "No, I don't think black families and white families with the same incomes will have significant differences in criminality."

Exactly. So then, there is no predictive ability that race has for any arbitrary class strata. It immediately follows then, that race has no predictive ability in terms of class. QED.

Though it is tangential at best to this discussion, let me say that I find it absolutely hilarious that in this thread and the one above it, I have written provably true statements and have been down-modded, whereas the people who I have been responding to have written provably false statements (outright contradictions!) and have been up-modded.

If the point-system rewards shallow group-think, and punishes truth - 'What is the point of it?', I must ask.

Heck, this has now become my new standard for judging content quality - let me see how many provably true, on-topic things one says in a thread, so that their score drops below unity.

I think you were downmodded because you were consistently arguing against points I explicitly told you I wasn't making. You are not some sort of martyr for truth.

You also tried to use the model of class to prove that people with vastly different incomes behave the same. You cannot prove with a model without proving that the model is accurate. The model you're using is wrong.

As I said before, I was never trying to appeal to the model of class anywhere in this discussion. I asked you to pretend I said socioeconomic status instead so we could stop arguing points I never made, but you wouldn't do me that favor.

Reply on news.ycombinator.com