3 ms·

That's kind of the point though: intelligence has many disputed definitions, but the general consensus is that extreme intelligence (or genius) involves the abi

That's kind of the point though: intelligence has many disputed definitions, but the general consensus is that extreme intelligence (or genius) involves the ability to grasp concepts others can't and apply them in novel ways, and need not savant-like working memory or ability to compute nearly as fast as a dumb old calculator. On the other hand the metrics and methodology of IQ tests distinguish between higher performers based on speed of puzzles that are comparatively simple to understand out of practicality. Of course processing speed for rotational symmetry puzzles also correlates pretty well with [in]ability to learn advanced topics at the lower end of the scale and people at the upper end of test performance are seldom considered stupid, but this thread is about the outliers. If we're still doing running analogies, Bolt and Kichoge are both huge outliers in running respective distances certainly proves they're significantly healthier than the average human, but maybe there's nothing wrong if they get sick or die at similar ages to normal people because peak human health isn't best defined by the outcome of a race.

dahart 11mo ago
> the general consensus is that extreme intelligence (or genius) involves the ability to grasp concepts others can’t and apply them in novel ways Whose consensus? Can you back up that claim with some links? The only actual metrics we have for intelligence are based on a speed race, and are not based on concepts that you and I can’t understand. Everyone can answer the questions on an IQ test correctly given unlimited time. I would claim that because IQ tests do not involve the ability to grasp concepts other can’t, there is no such consensus. Higher IQ is literally defined as someone being able to answer questions you can understand faster than you. > Both and Kichoge [sic] are both huge outliers in running A small nitpick, Bolt and Kipchoge are the fastest at what they do, but they are not huge outliers at all, they’re a very tiny bit faster than their competition, and there’s plenty of competition. This actually brings up a pretty important statistical point - that measuring outliers in terms of standard deviations washes away ALL of the absolute differences between people. It’s a relative ranking score, not a direct measure of intelligence. Someone with an IQ of 210 might be 0.1% smarter than someone with an IQ of 130. Or even less. The popular idea that a high IQ score somehow means someone has extreme intelligence far higher than others is a complete and total myth, there’s no basis for believing it’s true aside from conflating or misunderstanding what a standard deviation is. Standard deviations tell you about the distribution. It is a metric that gets exponentially smaller the higher the number is, and is always measuring in relation to the group, and does not correlate directly with the metric it’s capturing. There’s really no such thing as extreme intelligence with people who can grasp concepts others can’t. And there’s no such thing as a 210 IQ either, as others have pointed out. Everything over ~145 is almost completely meaningless and even if it had meaning simply does not demonstrate that very high IQ is appreciably smarter than your average scientist. Isn’t it actually already obvious there isn’t a cabal of super geniuses who have invented the things the rest of us don’t understand?
- notahacker 11mo ago
  > Whose consensus? Can you back up that claim with some links? I mean even the people who conceived speed-based tests defined intelligence broadly as “the global capacity of a person to act purposefully, to think rationally, and to deal effectively with his environment” (Weschler), which aligns with other dictionary and colloquial definitions and scientists with broader interests than psychometrics, and of course daily arguments on here about whether neural networks have achieved general intelligence which focus on their performance on benchmarks humans struggle to pass, not their processing speed or token memory. Psychometricians' focus on speed for actual metrics is driven by expediency (and the belief that batteries of simple speed tests correlate well with the ability to learn to perform more complex tasks to exceptionally high levels) rather than because they think peak intellectual ability is characterised exclusively by rapid puzzle solving. Needless to say genius is defined even more in terms of accomplishment and insightfulness, which is why lists of geniuses are dominated by scientists and creatives who conceived things their peers couldn't (some of which give even grad students with an interest in the subject headaches trying to understand when being walked through it), not people who are particularly fast at solving simple puzzles. Sure, 150 IQ people mostly just answer more questions within a time limit than 120 IQ people on some types of test, which is a function of processing speed and prioritization, but there are absolutely people who score low because they can't grasp what half the questions are looking for no matter how much time they're given. Not all of them are on the left side of the distribution, and some tests like Raven's APM give very similar rankings when taken untimed. And there are definitely plenty of smart people who couldn't say, solve Fermat's last theorem, despite being exceptional in their knowledge of mathematics and ability to solve related problems and being interested enough to try. I think we both agree there's a tendency of people to fetishise extreme intelligence without realising that not only can you possess it and not do anything useful with it (and possess it and have massive flaws in some areas of judgement), but you can also solve pretty significant problems without be notably special in metrics related to it. Also agree that obviously IQ is a constructed scale that doesn't capture absolute differences and for this and many other reasons distinctions over 145 are pretty meaningless. But I think that aligns pretty well with "maybe psychometricians really don't know what makes exceptionally smart people exceptionally smart"... (As for Bolt/Kipchoge, 'massive outlier' is a matter of perspective, but the amount of competition their best time is >1% faster than, selected from a pool of highly motivated full time professionals which in turn comes from a pool of millions who attempted sprint races at some point in their youth, is a point in their favour.)

dahart 11mo ago

> the general consensus is that extreme intelligence (or genius) involves the ability to grasp concepts others can’t and apply them in novel ways Whose consensus? Can you back up that claim with some links? The only actual metrics we have for intelligence are based on a speed race, and are not based on concepts that you and I can’t understand. Everyone can answer the questions on an IQ test correctly given unlimited time. I would claim that because IQ tests do not involve the ability to grasp concepts other can’t, there is no such consensus. Higher IQ is literally defined as someone being able to answer questions you can understand faster than you. > Both and Kichoge [sic] are both huge outliers in running A small nitpick, Bolt and Kipchoge are the fastest at what they do, but they are not huge outliers at all, they’re a very tiny bit faster than their competition, and there’s plenty of competition. This actually brings up a pretty important statistical point - that measuring outliers in terms of standard deviations washes away ALL of the absolute differences between people. It’s a relative ranking score, not a direct measure of intelligence. Someone with an IQ of 210 might be 0.1% smarter than someone with an IQ of 130. Or even less. The popular idea that a high IQ score somehow means someone has extreme intelligence far higher than others is a complete and total myth, there’s no basis for believing it’s true aside from conflating or misunderstanding what a standard deviation is. Standard deviations tell you about the distribution. It is a metric that gets exponentially smaller the higher the number is, and is always measuring in relation to the group, and does not correlate directly with the metric it’s capturing. There’s really no such thing as extreme intelligence with people who can grasp concepts others can’t. And there’s no such thing as a 210 IQ either, as others have pointed out. Everything over ~145 is almost completely meaningless and even if it had meaning simply does not demonstrate that very high IQ is appreciably smarter than your average scientist. Isn’t it actually already obvious there isn’t a cabal of super geniuses who have invented the things the rest of us don’t understand?

notahacker 11mo ago
> Whose consensus? Can you back up that claim with some links? I mean even the people who conceived speed-based tests defined intelligence broadly as “the global capacity of a person to act purposefully, to think rationally, and to deal effectively with his environment” (Weschler), which aligns with other dictionary and colloquial definitions and scientists with broader interests than psychometrics, and of course daily arguments on here about whether neural networks have achieved general intelligence which focus on their performance on benchmarks humans struggle to pass, not their processing speed or token memory. Psychometricians' focus on speed for actual metrics is driven by expediency (and the belief that batteries of simple speed tests correlate well with the ability to learn to perform more complex tasks to exceptionally high levels) rather than because they think peak intellectual ability is characterised exclusively by rapid puzzle solving. Needless to say genius is defined even more in terms of accomplishment and insightfulness, which is why lists of geniuses are dominated by scientists and creatives who conceived things their peers couldn't (some of which give even grad students with an interest in the subject headaches trying to understand when being walked through it), not people who are particularly fast at solving simple puzzles. Sure, 150 IQ people mostly just answer more questions within a time limit than 120 IQ people on some types of test, which is a function of processing speed and prioritization, but there are absolutely people who score low because they can't grasp what half the questions are looking for no matter how much time they're given. Not all of them are on the left side of the distribution, and some tests like Raven's APM give very similar rankings when taken untimed. And there are definitely plenty of smart people who couldn't say, solve Fermat's last theorem, despite being exceptional in their knowledge of mathematics and ability to solve related problems and being interested enough to try. I think we both agree there's a tendency of people to fetishise extreme intelligence without realising that not only can you possess it and not do anything useful with it (and possess it and have massive flaws in some areas of judgement), but you can also solve pretty significant problems without be notably special in metrics related to it. Also agree that obviously IQ is a constructed scale that doesn't capture absolute differences and for this and many other reasons distinctions over 145 are pretty meaningless. But I think that aligns pretty well with "maybe psychometricians really don't know what makes exceptionally smart people exceptionally smart"... (As for Bolt/Kipchoge, 'massive outlier' is a matter of perspective, but the amount of competition their best time is >1% faster than, selected from a pool of highly motivated full time professionals which in turn comes from a pool of millions who attempted sprint races at some point in their youth, is a point in their favour.)