wu :: forums
        (http://www.ocf.berkeley.edu/~wwu/cgi-bin/yabb/YaBB.cgi)
      

        riddles >> general problem-solving / chatting / whatever >> Ranking search results based on IQ
        
(Message started by: amichail on Jun 22nd, 2007, 6:55pm)
Title: Ranking search results based on IQ
Post by amichail on Jun 22nd, 2007, 6:55pm
Instead of using PageRank, how about computing an "IQRank" that takes into account the IQs of people?

For example, you could consider the IQ of the creator of the site as well as the IQs of people who link to the site.

The creator of a site would have a greater incentive to undergo IQ testing so as to have some chance of ranking highly.

Not sure what sort of incentive you could have for encouraging people linking to a site to undergo IQ testing.
Title: Re: Ranking search results based on IQ
Post by SMQ on Jun 22nd, 2007, 7:21pm
And what do you do with corporate content?  More importantly, do you honestly believe that content created by (or linked by) people who score well on a standardized IQ test is inherently likely to be more valuable than content created by people who score only average?  Because it seems to me that the tendency of the Internet toward egalitarianism is one of its great strengths.

--SMQ
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 22nd, 2007, 8:30pm

on 06/22/07 at 19:21:47, SMQ wrote:
Because it seems to me that the tendency of the Internet toward egalitarianism is one of its great strengths.

--SMQ

Is it really a strength?  Imagine instead an internet that gives greater rewards to smarter people. What would it look like?

What would arguments in online discussion groups look like if we can filter out posts by less intelligent people?

Would we get more or less spam?
Title: Re: Ranking search results based on IQ
Post by Sameer on Jun 22nd, 2007, 9:15pm
How do you define a "less intelligent post". IQ of a person doesn't necessarily determine how intelligent his posting is...
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 22nd, 2007, 9:19pm

on 06/22/07 at 21:15:14, Sameer wrote:
How do you define a "less intelligent post". IQ of a person doesn't necessarily determine how intelligent his posting is...

See the literature on the g factor.  It really does appear that people form a total order with respect to intelligence.
Title: Re: Ranking search results based on IQ
Post by Obob on Jun 23rd, 2007, 12:05am
I think that what Sameer is saying is that the intelligence of a person need not be correlated with the amount of thought and time that goes into making a post.  Just because somebody is intelligent doesn't mean that he will post things that are relevant to the discussion, promote further discussion, and aren't hastily constructed.

Besides, the practicalities of inputting an IQ test score into a search engine or ranking scheme are somewhat ridiculous.  Almost as ridiculous as flying a remote controlled model airplane over the internet.
Title: Re: Ranking search results based on IQ
Post by Barukh on Jun 23rd, 2007, 5:55am

on 06/22/07 at 18:55:47, amichail wrote:
Instead of using PageRank, how about computing an "IQRank" that takes into account the IQs of people?

For example, you could consider the IQ of the creator of the site as well as the IQs of people who link to the site.

How would you define this rank (as formally as possible, please)?
Title: Re: Ranking search results based on IQ
Post by rmsgrey on Jun 23rd, 2007, 7:06am
I is teh sm@rt3r!!!one!
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 23rd, 2007, 7:21am

on 06/23/07 at 00:05:38, Obob wrote:
Just because somebody is intelligent doesn't mean that he will post things that are relevant to the discussion, promote further discussion, and aren't hastily constructed.

You can find exceptions to everything.  This is just a heuristic that may work well in practice.


Quote:
Besides, the practicalities of inputting an IQ test score into a search engine or ranking scheme are somewhat ridiculous.

You could try to approximate it.  For example, you could start by assigning the highest IQs to mathematicians, physicists, and theoretical computer scientists from famous universities and research labs.    Then you might assume that they would associate mostly with smart people  and give those people high IQs and so on.
Title: Re: Ranking search results based on IQ
Post by Aurora on Jun 23rd, 2007, 9:19am
I'm sure that all of these people associate with those who's IQs differ greatly from their own, plus there are people with a very high IQ which isn't reflected in their profession.

Also, what about the people who aren't associated with anyone else there? ???
Title: Re: Ranking search results based on IQ
Post by Obob on Jun 23rd, 2007, 10:12am
If I remember correctly, google sort of already "guesses" intelligence in the way you are now describing.  Getting linked to from a .edu website dramatically increases your chance of being listed highly.
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 23rd, 2007, 10:40am

on 06/23/07 at 10:12:43, Obob wrote:
If I remember correctly, google sort of already "guesses" intelligence in the way you are now describing.  Getting linked to from a .edu website dramatically increases your chance of being listed highly.

Do you have any references?

As for "IQRank", I would guess that the result would be sort of like what you see on reddit.com, a social news service with more intelligent users than similar sites.
Title: Re: Ranking search results based on IQ
Post by Obob on Jun 23rd, 2007, 1:03pm
No, I don't have a reference for that.  I just remember being told this by a CS professor.
Title: Re: Ranking search results based on IQ
Post by Sameer on Jun 23rd, 2007, 2:50pm
IQ tests do not actually reflect a person's intelligence or even IQ for that fact. People are knowledgeable in different areas and respond differently to different situations. As for e.g. a person not good at math may be an exceptional painter. Would he score high on our current IQ tests?

I read the Pagerank paper long time ago so don't remember the specifics but it is more like a indicator of linking depths. How much a page is linked from other sites, etc. Which is why there was a recent news from google indicating how they are going to stop people from illegally linking to their sites using to artificially increase this linking score and hence their pagerank.

I think rmsgrey's post sums it up. Would you classify as a high IQ or low IQ post? That small sentence describes so much!!

I think the system of classification needs to be an adaptive neural network. When you join the discussion your "IQ" is lowest. As you contribute more to the discussion it rises based on the input you provide with relevance. If you can think up of devising a method of detecting this and updating the adaptive "IQ" rank system you can assign the IQ to that poster. Isn't this kind of gets similar to our forum where slowly you start at Newbie and rise to the level of "Uberpuzzler". Of course here it is entirely on the number of posts, but maybe you can adapt that to the relevancy of the posts.

Can you please summarize the "g factor" literature?
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 23rd, 2007, 2:54pm

on 06/23/07 at 14:50:48, Sameer wrote:
Can you please summarize the "g factor" literature?

See:

http://en.wikipedia.org/wiki/General_intelligence_factor
Title: Re: Ranking search results based on IQ
Post by towr on Jun 24th, 2007, 7:36am
*shakes head sorrowfully at the suggestion*

Really..

IQ is not particularly correlated with people knowing what they are talking about. They have to have actually researched it a bit, rather than be generally intelligent and have pulled an opinion from their ass.
Also, IQ is partly determined by age; it's supposed to represent one's mental age divided by one's physical age. Should I prefer the opinion of a ten year old with an IQ of 140 over that of a 40 year expert with an IQ of 110?


on 06/23/07 at 07:21:54, amichail wrote:
You could try to approximate it.  For example, you could start by assigning the highest IQs to mathematicians, physicists, and theoretical computer scientists from famous universities and research labs.
So, can that be taken to imply you consider professors in the humanities of inferior intellect? The mathy brain is the only real brain?
Title: Re: Ranking search results based on IQ
Post by rmsgrey on Jun 24th, 2007, 9:30am

on 06/23/07 at 14:50:48, Sameer wrote:
IQ tests do not actually reflect a person's intelligence or even IQ for that fact.

Doesn't that depend on how you define IQ?

The practical definition seems to be that IQ is whatever it is that IQ tests measure. How that links to actual intellect is an open question.
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 24th, 2007, 9:36am

on 06/24/07 at 07:36:26, towr wrote:
So, can that be taken to imply you consider professors in the humanities of inferior intellect? The mathy brain is the only real brain?

This is a complicated question.  See for example:

http://www.hno.harvard.edu/gazette/2003/10.23/01-creativity.html

"We didn't find this," Carson notes. "We saw creativity increase as IQs climb to 130 (the average score of Harvard students), and even up to 150."

But even so, the humanities are likely not as competitive as mathematical fields.

So yes, I think the best way to identify high IQ individuals is to do a search as I have described. Sure, some people in the humanities may be super smart also, but being a humanities professor at a prestiguous institution would not imply this.
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 24th, 2007, 9:38am

on 06/24/07 at 09:30:27, rmsgrey wrote:
Doesn't that depend on how you define IQ?

The practical definition seems to be that IQ is whatever it is that IQ tests measure. How that links to actual intellect is an open question.

Have you read this?

http://en.wikipedia.org/wiki/General_intelligence_factor
Title: Re: Ranking search results based on IQ
Post by Barukh on Jun 24th, 2007, 11:15am

on 06/24/07 at 09:36:20, amichail wrote:
Sure, some people in the humanities may be super smart also, but being a humanities professor at a prestiguous institution would not imply this.

...and being a math professor at  a prestigious institution would?
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 24th, 2007, 11:17am

on 06/24/07 at 11:15:07, Barukh wrote:
...and being a math professor at  a prestigious institution would?

Yes, it very likely would.
Title: Re: Ranking search results based on IQ
Post by towr on Jun 24th, 2007, 12:11pm
Seems like a bit of a prejudice..

Of course the really intelligent people are in artificial intelligence; cause that's where I am. ;)
Title: Re: Ranking search results based on IQ
Post by Obob on Jun 24th, 2007, 12:34pm
It may not be the case that a humanities professor is the best person to ask about some technical or mathematical question.  Even being a math person, though, I would very much disagree with the statement that math/sciences people are a priori smarter than humanities people.  Especially when it comes to talking about established professors at well-known universities.  Just because somebody doesn't know how to solve a differential equation or something doesn't mean that their grasp of other concepts like philosophical theories, languages, and so on and so forth does not compensate for this.
Title: Re: Ranking search results based on IQ
Post by Barukh on Jun 24th, 2007, 10:45pm

on 06/24/07 at 12:11:50, towr wrote:
Seems like a bit of a prejudice..

Agree. But afterall, it depends on how do define "smart people".
Title: Re: Ranking search results based on IQ
Post by towr on Jun 25th, 2007, 1:15am

on 06/24/07 at 22:45:10, Barukh wrote:
Agree. But afterall, it depends on how do define "smart people".
Yes, I suspect there is always a tendency to consider people that don't know what you are talking about as stupid (even if you don't understand what they're talking about either).
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 25th, 2007, 7:47am

on 06/24/07 at 12:11:50, towr wrote:
Seems like a bit of a prejudice..

Of course the really intelligent people are in artificial intelligence; cause that's where I am. ;)

The difference between AI and theoretical computer science is that the former deals with heuristics while the latter requires provable guarantees.

So is it harder to come up with good heuristics than  it is to come up with provable guarantees (for presumably simpler problems)?

It's hard to say.

There are theoreticians who dabble in AI of course, complicating things even more:

http://www.it.jcu.edu.au/Subjects/cp4000/2003-1/resources/seminar/p181-paterson.pdf
Title: Re: Ranking search results based on IQ
Post by towr on Jun 25th, 2007, 9:07am
That really wasn't the point of course.

And AI deals with rather a bit more than heuristics. e.g. The particular subject I'm writing my thesis on is a theorem prover; pure logical reasoning.
And if one were to write a program that can automatically prove the validity of programs; that would be AI applied to theoretical computer science.
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 25th, 2007, 9:16am

on 06/25/07 at 09:07:06, towr wrote:
That really wasn't the point of course.

And AI deals with rather a bit more than heuristics. e.g. The particular subject I'm writing my thesis on is a theorem prover; pure logical reasoning.

Theorem provers generally employ heuristics.  The problem addressed is usually undecidable.
Title: Re: Ranking search results based on IQ
Post by towr on Jun 25th, 2007, 9:25am

on 06/25/07 at 09:16:58, amichail wrote:
Theorem provers generally employ heuristics.  The problem addressed is usually undecidable.
Theorem provers also generally have provably correct results; which is rather the point. So any 'heuristic' that is used cannot be allowed to introduce any uncertainty about correctness of the result.
If you find a proof, the proof must be correct; if you find there isn't a proof, there mustn't be; if you can't decide within reasonable time, you don't.
Title: Re: Ranking search results based on IQ
Post by amichail on Jun 25th, 2007, 9:29am

on 06/25/07 at 09:25:19, towr wrote:
Theorem provers also generally have provably correct results; which is rather the point. So any 'heuristic' that is used cannot be allowed to introduce any uncertainty about correctness of the result.
If you find a proof, the proof must be correct; if you find there isn't a proof, there mustn't be; if you can't decide within reasonable time, you don't.

Evaluation though cannot consist of only proofs (as in theory).  You need to measure performance using experiments.

Ensuring correctness may of course require some non-trivial proofs.
Title: Re: Ranking search results based on IQ
Post by rmsgrey on Jun 25th, 2007, 10:11am

on 06/25/07 at 01:15:26, towr wrote:
Yes, I suspect there is always a tendency to consider people that don't know what you are talking about as stupid (even if you don't understand what they're talking about either).

"And then, like, Britney was dating, like, Justin, and I, like, heard..."
Title: Re: Ranking search results based on IQ
Post by towr on Jun 25th, 2007, 10:12am

on 06/25/07 at 09:29:57, amichail wrote:
Evaluation though cannot consist of only proofs (as in theory).
That holds for every applied science.
Theoretical computer science may give wonderful algorithm with great theoretical time complexity that when applied are entirely worthless in any real situation.

And also AI, like CS, isn't necessarily applied; there is a lot of theoretical work being done (like inventing logics and protocols).
Title: Re: Ranking search results based on IQ
Post by Sameer on Jun 25th, 2007, 1:11pm
I think it would be important to figure out how a person with high IQ (we are as yet to define IQ without bias) got the way he/she is now. My imaginative/visual brain tells me as a person growsn and learns he develops a neural pathway, kind of like a cache line/circuit in microprocessor .. dynamic ain't it? ... witht that and his personal physiology will dictate how many pathways the person can accomodate and how fast can he generate/retain those.. like the speed of the processor ;-) ... anyways.. this will ultimately result in a fast thinking creative indivdual who we perceive as a high IQ person. Now where have I heard this before.. yes.. I think this does apply to AI/neural/fuzzy logic and if we come up with a good adaptive model (as I suggested before) ... we should have a good ranking algorithm!!!
Title: Re: Ranking search results based on IQ
Post by JiNbOtAk on Jun 25th, 2007, 10:37pm

on 06/25/07 at 10:11:52, rmsgrey wrote:
"And then, like, Britney was dating, like, Justin, and I, like, heard..."


Uh, yeah, and I was like, whatever..  

p/s : Couldn't resist.   ;D
Title: Re: Ranking search results based on IQ
Post by Barukh on Jun 26th, 2007, 9:57am
On a related note: when I googled "Cyclotomic Polynomial" (an important concept in number theory), I expected the articles from Mathworld and Wikipedia to get very high ranks. The latter is indeed the first, but where is  the former?..
Title: Re: Ranking search results based on IQ
Post by towr on Jun 26th, 2007, 10:27am

on 06/26/07 at 09:57:48, Barukh wrote:
On a related note: when I googled "Cyclotomic Polynomial" (an important concept in number theory), I expected the articles from Mathworld and Wikipedia to get very high ranks. The latter is indeed the first, but where is  the former?..
http://mathworld.wolfram.com/CyclotomicPolynomial.html ?
Or do you mean amongst google's results?
Title: Re: Ranking search results based on IQ
Post by Barukh on Jun 26th, 2007, 12:14pm

on 06/26/07 at 10:27:48, towr wrote:
Or do you mean amongst google's results?

Yes, of  course!
Title: Re: Ranking search results based on IQ
Post by Sameer on Jun 26th, 2007, 1:12pm
I notice that lot of results are from .edu website.. Maybe google does give them a higher rank than wolfram?

THere was one thing to mention on that. Long time ago I had opened up my website on a freeserver. I wouldn't see that on the top of google results. So in order to improve it I added the site to google groups and such... and voila it was number 1 in my search result. Now, I used to update it regularly and it stayed on the top, but slowly it started showing results for me and people with similar name from edu site on higher rank. As for e.g. a website with my name on a paper in the citations would show up higher than my website. Lately I haven't updated my website at all as I feel very bored in doing so.. my information is at least 2 years outdated... the result appears now somewhere way down...  


Powered by YaBB 1 Gold - SP 1.4!
Forum software copyright © 2000-2004 Yet another Bulletin Board