Wolfram|Alpha

[kandrathe]

I came across this gem of a website and thought I would share it.

http://www.wolframalpha.com

[swirly]

Leo Laporte did an interview with Stephen Wolfram about this site awhile ago. It's an interesting site, though I've not found a real use for it. I think the interview is on the end of this TWiT if people would like to hear it. There is also a transcript up here of the whole episode of TWiT (including the interview at the end) if people would rather read it. Just a little more information on the site.

[LennyLen]

I used it for a while when it was still in beta testing and just after it was released, but abandoned it after noticing that a lot of the results were plain wrong. You can also get better info on most searches off google. It's good for a few calculations though, but not for much else.

[swirly]

I used it for a while when it was still in beta testing and just after it was released, but abandoned it after noticing that a lot of the results were plain wrong. You can also get better info on most searches off google. It's good for a few calculations though, but not for much else.

I think the issue is that this isn't really what it is meant for. It isn't really a search, but rather a program trying to look at data and draw comparisons or correlations. I freely admit that I don't really understand it myself. From hearing Stephen Wolfram talk about it though, it really isn't meant to be used to find data that you can find just by using google. Let me quote him. (taken from http://wiki.twit.tv/wiki/TWiT_195/Transcript)

right now, Wolfram|Alpha is mostly making use of kind of old kind of science facts, it’s making use of the models and methods and so on, that have been developed in science and physics and financial analysis and lots of other areas over the course of a few – of the last few hundred years. What this kind of – what the ultimate sort of direction that one can go in is not just to say, here’s a question we have to ask which can answered by existing science and now Wolfram|Alpha can actually answer it quickly for you on the web type thing, but say here is a question that I have for which I have to sort of invent a new model, something that science has never produced for us, to go out and discover sort of a new algorithm. Now can we do this thing of going out and sort of, going and sampling this – mining the computational Universe to find something to sort of invent on the fly.

Here’s a fun thing that I just was really surprised by. Given you can know like from birth record, you can know how many people named I don’t know, Stephen or something, were born in each year for the last 100 years or so. But then you can use kind of mortality data to figure out, so just how many will be surviving at this time and so on. And from that you can get kind of a distribution of what the expected ages of people will be. And it’s really bizarre because an awful lot of people, particularly with slightly unusual names, you can basically predict from their name roughly how old they are.

Now this information may not have any real value or meaning, but I think it is more the purpose of the site. To find areas of relation and attempt to see what can be figured out by it. Which may result in useless information, but it might be interesting as well. Like I said in my first post though, I've yet to find a real use for it myself. I just don't think people should be going to it trying to view it like a google.

[--Pete]

Hi,

I think the issue is that this isn't really what it is meant for. It isn't really a search, but rather a program trying to look at data and draw comparisons or correlations.

I think you've got the basic idea here. This stuff is very interesting and exciting. Sort of a scientific expert system.

The way I understand it, there are (at least) two kinds of knowledge. There's the knowledge of facts (such as the date of Magna Carta or the formula for calculating the circumference from the diameter) and the knowledge of abilities (such as knapping a spearhead or actually multiplying by pi). Search engines are good for finding the facts, including the facts of *how* to do things, but they do not help much when it comes actually doing things. Doing a restricted set of things is what this program is about.

To give an example. If you're interested in the gravitational field in a binary system according to general relativity, you could search for it. Unless someone has done the calculations and posted the results, the best you would find is articles on general relativity from which you could do the calculation yourself. Unless you are one of the relatively few who can actually do math at that level, you would be stuck at this point. You would have all the facts you needed, but not the ability to use those facts to answer your question. This site is an attempt to take that last step, to use the facts to solve the problem.

I tried a few things, some worked, some gave an incorrect answer, some gave no answer at all. It's a small step toward computers that actually solve problems for you, rather than just running the program you wrote to solve the problem. Very exciting.

--Pete

[Lissa]

Hi,
I think you've got the basic idea here. This stuff is very interesting and exciting. Sort of a scientific expert system.

The way I understand it, there are (at least) two kinds of knowledge. There's the knowledge of facts (such as the date of Magna Carta or the formula for calculating the circumference from the diameter) and the knowledge of abilities (such as knapping a spearhead or actually multiplying by pi). Search engines are good for finding the facts, including the facts of *how* to do things, but they do not help much when it comes actually doing things. Doing a restricted set of things is what this program is about.

To give an example. If you're interested in the gravitational field in a binary system according to general relativity, you could search for it. Unless someone has done the calculations and posted the results, the best you would find is articles on general relativity from which you could do the calculation yourself. Unless you are one of the relatively few who can actually do math at that level, you would be stuck at this point. You would have all the facts you needed, but not the ability to use those facts to answer your question. This site is an attempt to take that last step, to use the facts to solve the problem.

I tried a few things, some worked, some gave an incorrect answer, some gave no answer at all. It's a small step toward computers that actually solve problems for you, rather than just running the program you wrote to solve the problem. Very exciting.

--Pete

I wonder what it would do if you tossed in Boltzmann's Transport Equation...hopefully it would throw up it's hands rather quickly and say unsolveable (as Boltzmann's Transport Equation is in its raw form without assumptions being made).

[--Pete]

Hi,

I wonder what it would do if you tossed in Boltzmann's Transport Equation...hopefully it would throw up it's hands rather quickly and say unsolveable (as Boltzmann's Transport Equation is in its raw form without assumptions being made).

I tried something much simpler -- the orbit of Earth's Moon around the Sun using classical mechanics. Either I didn't frame the problem correctly or it just couldn't do it. I suspect that anything having to do with plasma dynamics, especially an indeterminate set of tensor equations, might just give that program a migraine. ;)

--Pete

[Jester]

I came across this gem of a website and thought I would share it.

http://www.wolframalpha.com

Neat. I might actually be able to use this. It would make a great econ history lecturing tool, so long as you knew the source of this stuff.

Wanna see the demographic transition in action? Just try GDP per capita vs. fertility rate. Or life expectancy vs. fertility rate. Hey presto, it's right there.

-Jester

[kandrathe]

Neat. I might actually be able to use this. It would make a great econ history lecturing tool, so long as you knew the source of this stuff.

Wanna see the demographic transition in action? Just try GDP per capita vs. fertility rate. Or life expectancy vs. fertility rate. Hey presto, it's right there.

I also find things like "quadratic (-2,3,-5,4,-9,12,7,30)" interesting.

Try also,

• Poison Ivy<>
  
• standard deviation (90,120,145,124,234,80,75)<>
  
• University of Pennsylvania<>
  
• Hexaflourobenzene<>
  
• Deer<>
  
• Mandlebrot -0.5,0.56667<>
  
• 64x^6-96x^4+36x^2-3<>
  
• Dimethylsulfoxide<>
  
• lunar eclipse<>
  
• Ursa Major<>
  
• Rydberg formula ni=6, nf=7<>
  
• Chicken McNugget<>
  
• Cow Dung<>
  [st]

[LennyLen]

I think the issue is that this isn't really what it is meant for. It isn't really a search, but rather a program trying to look at data and draw comparisons or correlations.

One of the things I was using it for was for comparing incidences of automobile accidents in different areas to the relative population and population densities of the respective areas. However, the figures it was quoting for population and density were outdated, which made any results it gave meaningless. It was these figures that I was able to obtain more accurately using google and/or Wikipedia.

[kandrathe]

One of the things I was using it for was for comparing incidences of automobile accidents in different areas to the relative population and population densities of the respective areas. However, the figures it was quoting for population and density were outdated, which made any results it gave meaningless. It was these figures that I was able to obtain more accurately using google and/or Wikipedia.

Yes, the internet (and publishing for that matter) in general is better with static knowledge. There is a "freshness date" on many statistical reports, and the problem of comparing proper data sets.