An Update on Wolfram Alpha
Last month I (along with half of the blogging world it seems) reported on a new project from Wolfram Research called Wolfram Alpha. At the time I didn’t know much about it but since then much more information has been available.
Rather than write more about it myself, I will simply refer you to the information I have seen. First of all, the video below is a recording of a talk given by Stephen Wolfram about the project a couple of days ago. Click here if you prefer to watch this directly on YouTube.
Secondly, a new blog has started up called (predictably enough) The Wolfram Alpha blog. The first article is called The Quest for Computable Knowledge: A Longer View.
Many people are comparing Wolfram Alpha to Google which is completely wrong as far as I can tell. It’s a bit like comparing a table of logarithms with a computer algebra system (CAS) that can calculate any logarithm you like, in any base along with the ability to plot a graph of any function that contains a logarithm. The CAS can also differentiate, integrate, and simplify equations containing logarithms and so on. The table of logarithms contains pre-computed data that you can search whereas the CAS can do computations based on your ‘searches’ on the fly.
Let’s take an example from the talk that Stephen gave. Say you enter
6000C
(I assume this is what he entered, the video wasn’t so hot on showing the overhead projector connected to his laptop. I’d give good money to get some slides right now)
Wolfram Alpha tries to tell you something useful about this input drawing from the data-sets that it has access to. It also COMPUTES potentially useful things from these data sets. It’s the computation part that makes it fundamentally different from google (NOTE: I said different from, not better than or a competitor to – the distinction is important in my opinion).
Wolfram Alpha will say something like ‘assuming that C is a unit and assuming that unit is Celsius then I can tell you the following about 6000C’ It will then go on to tell you various conversion factors (such as, I guess, what it would be in degrees Fahrenheit or Kelvin). It might plot the black body spectrum at 6000 degrees C and show what colour a black-body object at that temperature would be.
All of these results will have been computed from scratch based upon various mathematical formulae and models that Wolfram Alpha has access to. Google, on the other hand, would only show the black body spectrum at 6000C if someone had calculated it in advance and put it up on a webpage.
Of course Wolfram Alpha knows a lot more than just temperatures. It knows about weather statistics, economic data, the human genome, properties of materials, symbolic integrals (and how to do them by hand) and loads more.
Looks like an exciting project! I wonder how Wolfram will make money out if it (or even if it intends to)?
They do plan to make money. If you subscribe to the premium version of Alpha you get access to the raw data it generates.
As I see it it is, among other things, a cut down version of their Mathematica product. By ignoring the multitude of lazy mistaken comparisons with Google, they have generated an amazing amount of interest in a product only a few people knew about.
As some of the useful equation simplification tools only go into a finite amount of depth it will encourage people to either subscribe to Wolfram Pro or buy a their software for more complex solutions.