The Power & the Problems with the New Wolfram Alpha Pro

Wolfram Alpha unveils a "pro" version of its computational search engine today, with a new set of tools that promises to make it easier for anyone to tap into this powerful technology.

Although we've come to associate searching for information online with "Googling," Wolfram Alpha offers a different sort of "search." Rather than querying the Web for pages that might have the information you're looking for, Wolfram Alpha queries its own massive database, containing hundreds of datasets (everything from weather data to sports stats to birth records to box office records). If you ask Google "What is the meaning of life?" you'll get pages of links -- to Aristotle, to reviews for the iPhone 4S, to Zen Buddhist sites, to WikiAnswers, and so on. Ask Wolfram Alpha, and you get the correct answer, along with the source of the information.

The Power of Wolfram Alpha Pro

For $4.99 per month ($2.99 for students), Wolfram Alpha Pro will give you access to more of the computational power "under the hood" of the site, in part by allowing you to apply the technology to your own datasets.  Rather than just offer an input box where you can type your queries, you'll be able to upload images and files that Wolfram Alpha will in turn analyze for you.

This includes text files (Wolfram Alpha will respond with the character and word count, an estimate on how long it would take to read aloud, most common word, average sentence length and more); spreadsheets (it will crunch the numbers and return a variety of statistics and graphs); and image files (it will analyze the image's dimensions, size, and colors as well as let you apply several different filters). Over 60 different file types can be uploaded, and you will also be able to download the analysis as well (and not just from your own uploads, but from all the data related to your particular search query).

There's also a new extended keyboard that contains the Greek alphabet and other special characters for manually entering data. (Side note: Read this post by Dan Meyer on math education and Silicon Valley, and stew on how Wolfram Alpha's tools fit in to his argument. Or at least that's what I've been doing.)

The Problems

I've been tinkering with the new tool since I got access to it earlier this week, and I confess, even before I watched Stephen Wolfram demo Wolfram Alpha Pro, I had my credit card out and ready to pay the subscription fee. Why? Well, as I've written previously (rather, as I write regularly), data is becoming increasingly important (and politicized, no doubt) in education. But data isn't valuable unless you can extract meaning from it, and many of the tools for analyzing data (particularly large datasets) remain out of the hands of non-data-scientists.

My friend Pete Warden has been at the forefront of helping create just these sorts of tools. His Data Science Toolkit offers open data sets and open source data tools. Note the adjective there: "open." That's something that the Wolfram Alpha product definitely is not (See: its CDF file format).

But that's just one problem that I have with the new tool -- and a concern that I'm not sure others share. My other beef isn't with the tool itself but with what Wolfram Alpha Pro quickly reveals: the shoddy state that a lot of datasets are in.

The first place I went to test-drive Wolfram Alpha Pro was ed.data.gov, the federal government's collection of education data. There are 110 datasets (purportedly) available there, but I struggled to find a single one that would work. There were lots of "files not found" and incompatible file formats -- quite a feat considering the tool accepts 60 different types. In some cases, to make the upload work just meant opening up the CSV files and deleting a few columns -- a good reminder of the difference between machine-readable and human-readable spreadsheets.

With the new tool, "you take the data, throw it at Wolfram Alpha Pro, and see what it has to say about it," says Stephen Wolfram.  Unfortunately, in a lot of cases, Wolfram Alpha Pro won't be able to say much.



Tags: , , , , ,