“LessWrong has lost 52% of it’s giftedness since March of 2009” is an incredibly sensationalist way of describing a mere 7-point average IQ drop. Especially if the average is dropping due to new users, because then the “giftedness” isn’t actually being lost.
Honestly, so long as the drop is due to lower-IQ people arriving rather than higher-IQ people leaving, I can’t see why it’s such a big deal—especially if the “new” people mostly just lurk. Now, if the average IQ of only the people with > 100 karma in the last 30 days was also dropping with time...
A. Well, 52% is the real figure. That’s just math.
B. We don’t know why the average IQ dropped. It could be because some of the gifted users have left.
C. Saying “if the data is good, LessWrong has lost 52% of it’s giftedness” in the context of an average should be interpreted as “if the data is good, LessWrong’s average has lost 52% of it’s giftedness”. This is correct. I feel that this is picking at details, but I’ll go add the word “average” so that nobody else does it.
If you feel there is some deeper problem with my statement, please continue. If not, does this address your criticism?
Although 52% is a figure you calculate with math, describing it as “losing giftedness” is not math. Math is not about dividing numbers by other numbers; it’s about figuring out which numbers are the correct ones to divide.
What you have calculated, as far as I know, is that the IQ mean has moved 52% of the way towards an arbitrary cutoff point. I have two objections to this:
The arbitrary cutoff point. Why should we specifically care about whether someone’s IQ exceeds 132?
More importantly, even if we did care about this, the correct measure would be the number (or proportion) of people on LW with an IQ of 132+. We can get some idea of what this is from gwern’s reply, but the decrease in this is nowhere near 52%.
Edit: to be clear, I think that the “7-point drop in IQ” figure is the initial figure to report. Further analysis might reveal that this drop is due to new users arriving with a new IQ distribution, or to older users leaving at a rate dependent on IQ, or a mix of the two; that would also be useful to know, and report details on.
No matter what cutoff point I choose for the giftedness calculations, it will be argued that it is the wrong cutoff point. There are a lot of definitions of giftedness, and there’s a lot of controversy over how giftedness should be defined. There’s nothing I can do about that.
The reason I chose IQ 132 (or rather the top 2% which can vary from one test to another) is explained in the comment I linked to about this. Briefly: if you have that IQ, you qualify as gifted by most IQ based definitions of giftedness.
The most relevant thing about this IQ is that the research on giftedness tends to be done on people who have IQs over 132. I could have picked 110, but there would be very little research to read about “gifted” people with an IQ of 110. Conversely, had I picked 180, you’d be hard pressed to find any research at all. I looked once and found exactly one book on that IQ range. It’s full of case studies. Those people are so rare, that this is about all they could do.
I picked the top 2% because although there’s no standard, it is as close to a standard as I’ve got.
I decided to do the math anyway, and I’m content with having chosen a number that’s connected to something meaningful: most of the research on giftedness. (Yes there are meaningful connections between IQ and all kinds of things. It’s a common myth to assume IQ is only a number. It’s not.)
Since there really isn’t an IQ threshold that could have been chosen that would not be controversial, what would you have done?
Since there really isn’t an IQ threshold that could have been chosen that would not be controversial, what would you have done?
I would have reported the IQ difference (and possibly some of the figures in this comment, which if you prefer you can also calculate for 132) rather than a dubious measure of distance to the 132 threshold.
In hindsight I regret objecting to the cutoff since my main source of dismay is the second point in my previous comment.
The reason I chose IQ 132 (or rather the top 2% which can vary from one test to another) is explained in the comment I linked to about this. Briefly: if you have that IQ, you qualify as gifted by most IQ based definitions of giftedness.
One user with IQ 120 leaving and one with IQ 100 entering decrease the average IQ, but describing that as “losing giftedness” sounds kind of fucked up to me.
“LessWrong has lost 52% of it’s giftedness since March of 2009” is an incredibly sensationalist way of describing a mere 7-point average IQ drop. Especially if the average is dropping due to new users, because then the “giftedness” isn’t actually being lost.
Some absolute figures:
Well, I agree, but “mere” probably isn’t a sensationalist enough way to describe a 7 point drop in IQ.
Okay, I agree, maybe that was pushing it a little.
Honestly, so long as the drop is due to lower-IQ people arriving rather than higher-IQ people leaving, I can’t see why it’s such a big deal—especially if the “new” people mostly just lurk. Now, if the average IQ of only the people with > 100 karma in the last 30 days was also dropping with time...
A. Well, 52% is the real figure. That’s just math.
B. We don’t know why the average IQ dropped. It could be because some of the gifted users have left.
C. Saying “if the data is good, LessWrong has lost 52% of it’s giftedness” in the context of an average should be interpreted as “if the data is good, LessWrong’s average has lost 52% of it’s giftedness”. This is correct. I feel that this is picking at details, but I’ll go add the word “average” so that nobody else does it.
If you feel there is some deeper problem with my statement, please continue. If not, does this address your criticism?
Although 52% is a figure you calculate with math, describing it as “losing giftedness” is not math. Math is not about dividing numbers by other numbers; it’s about figuring out which numbers are the correct ones to divide.
What you have calculated, as far as I know, is that the IQ mean has moved 52% of the way towards an arbitrary cutoff point. I have two objections to this:
The arbitrary cutoff point. Why should we specifically care about whether someone’s IQ exceeds 132?
More importantly, even if we did care about this, the correct measure would be the number (or proportion) of people on LW with an IQ of 132+. We can get some idea of what this is from gwern’s reply, but the decrease in this is nowhere near 52%.
Edit: to be clear, I think that the “7-point drop in IQ” figure is the initial figure to report. Further analysis might reveal that this drop is due to new users arriving with a new IQ distribution, or to older users leaving at a rate dependent on IQ, or a mix of the two; that would also be useful to know, and report details on.
No matter what cutoff point I choose for the giftedness calculations, it will be argued that it is the wrong cutoff point. There are a lot of definitions of giftedness, and there’s a lot of controversy over how giftedness should be defined. There’s nothing I can do about that.
The reason I chose IQ 132 (or rather the top 2% which can vary from one test to another) is explained in the comment I linked to about this. Briefly: if you have that IQ, you qualify as gifted by most IQ based definitions of giftedness.
The most relevant thing about this IQ is that the research on giftedness tends to be done on people who have IQs over 132. I could have picked 110, but there would be very little research to read about “gifted” people with an IQ of 110. Conversely, had I picked 180, you’d be hard pressed to find any research at all. I looked once and found exactly one book on that IQ range. It’s full of case studies. Those people are so rare, that this is about all they could do.
I picked the top 2% because although there’s no standard, it is as close to a standard as I’ve got.
I decided to do the math anyway, and I’m content with having chosen a number that’s connected to something meaningful: most of the research on giftedness. (Yes there are meaningful connections between IQ and all kinds of things. It’s a common myth to assume IQ is only a number. It’s not.)
Since there really isn’t an IQ threshold that could have been chosen that would not be controversial, what would you have done?
I would have reported the IQ difference (and possibly some of the figures in this comment, which if you prefer you can also calculate for 132) rather than a dubious measure of distance to the 132 threshold.
In hindsight I regret objecting to the cutoff since my main source of dismay is the second point in my previous comment.
One user with IQ 120 leaving and one with IQ 100 entering decrease the average IQ, but describing that as “losing giftedness” sounds kind of fucked up to me.
For that matter, it is currently the case that users with IQ 132 entering would also decrease the average IQ (and cause LW to “lose giftedness”).