Wolfram Alpha set for launch, first look unveiled
May 14, 2009 6:15 PM CT
Wolfram Alpha is set to launch Friday, and creator Stephen Wolfram has published a great screencast that demonstrates the strengths of his new tool. I've watched the entire screencast, and I've read Wolfram's latest blog post, so what follows are my preliminary thoughts and impressions of Alpha what I've seen so far.
First, the juxtaposition of a fantastic software demo with a blog post of mind-bending bogosity recalls my very first impression of Alpha when its existence was unveiled, to wit: "I leave evaluation of Wolfram Alpha's prospects as a 'Google killer' to Web 2.0 pundits, and, ultimately, to the marketplace. (Regarding the latter, Wolfram Alpha's commercial prospects are considerably brighter than its singularity-inducing metaphysical ones, since Wolfram does know how to produce great software.)" But I'll leave off any more discussion of Wolfram the man and focus on Wolfram the software company, instead.
I have my credit card out
I already have my credit card out to sign up for the Pro version, and this will even give me an excuse to renew my Mathematica license. Nerd that I am, I loved using Mathematica so much that I continued to look for excuses to play with it long after I left engineering, but I eventually gave up in my second year of humanities graduate work.
I've even cleared a spot on my second monitor for a permanently open Mathematica notebook that I can use with Alpha—I imagine it will be that useful for journalists, students, and researchers in general—both the good among them, and the evil (more on that in a moment, though).
If you're someone who spends a lot of time online reading product and movie reviews, shopping, keeping up on the latest news, vanity surfing, cyber stalking, hunting for recipes, trolling for stock tips, reading celebrity gossip, and generally immersing yourself in the vast sea of largely unstructured information that Google makes instantly available, then you may not find much to interest you in Alpha.
In a nutshell, Alpha seems to have the following going for it as a research tool:
* A very good natural language parser
* Mathematica's formidable prowess at calculation, unit conversion, and chart and graph generation
* A back-end team of dedicated humans who have carefully structured a large body of data
* Access to premium data sources, and a system for automatically structuring the data from each source so that it works with Alpha.
Mix all of that together, and you have a product that can answer "what is the price of tea in China" by producing an up-to-date chart showing the prices of various Chinese teas in major world currencies, plots of the average price of Chinese tea over time, the geographic distribution of tea-growing regions, and so on.
I can envision replacing between 10 and 20 percent of my Google queries with Alpha queries, and I can also see where I'll end up doing Alpha queries that would never have occurred to me to try via Google.
But even so, a Google-killer this is not. And it clearly doesn't aim to be, either, despite the inevitable comparisons.
Structured vs. unstructured, facts vs. web pages
For all its strengths as an answerer of questions, Google is still a search engine that takes in search terms and returns a list of relevant webpages. Some of those webpages contain facts and charts and plots, so that if you're looking for such things you can find them via Google. But ultimately Google is about finding web pages, in the same way that the card catalog is about finding books and articles.
Alpha, in contrast, skips over the web pages part and gives you data, especially data that can be quantified numerically. Alpha seems especially useful for producing comparisons of how different data series have fluctuated over time. For instance, near the end of Wolfram's screencast, he shows a graph that reveals a curious inverse symmetry between the popularity of the names Andrew and Paul over the past few decades.
So if you're someone who spends a lot of time looking at graphs and tables in your day job, Alpha will have a ton of appeal. In contrast, if you're someone who spends a lot of time online reading product and movie reviews, shopping, keeping up on the latest news, vanity surfing, cyber stalking, hunting for recipes, trolling for stock tips, reading celebrity gossip, and generally immersing yourself in the vast sea of largely unstructured information that Google makes instantly available, then you may not find much to interest you in Alpha.
With great power comes great responsibility
It dawned on me at the moment in the screencast where Alpha produced a scatter plot of GDP vs. railway length in Europe—the two do appear to correlate—that the software's powerful ability to produce time series comparisons of any two or more things based on any number of quantifiable factors will be both a great blessing and a great curse. Specifically, I paused the screencast and imagined George Will using Alpha to do "research" on Antarctic sea ice levels for his next column.
Depressingly, I think that Alpha will open up a whole new world of possibilities for cranks and hacks of every flavor, giving them the power to easily and prolifically produce what former Ars editor Julian Sanchez has usefully dubbed "one-way hash arguments":
Certain bad arguments work the same way—skim online debates between biologists and earnest ID afficionados armed with talking points if you want a few examples: The talking point on one side is just complex enough that it’s both intelligible—even somewhat intuitive—to the layman and sounds as though it might qualify as some kind of insight. (If it seems too obvious, perhaps paradoxically, we’ll tend to assume everyone on the other side thought of it themselves and had some good reason to reject it.) The rebuttal, by contrast, may require explaining a whole series of preliminary concepts before it’s really possible to explain why the talking point is wrong. So the setup is “snappy, intuitively appealing argument without obvious problems” vs. “rebuttal I probably don’t have time to read, let alone analyze closely.”
I anticipate that pundits, think-tankers, and anonymous posters to Internet discussion boards and blogs will use Alpha to build one-way hash arguments that ape the authoritative appearance of sound science with near-perfect verisimilitude, all to devastating effect on public discourse. Truly, Sting had it right when he wrote: "I never saw no miracle of science that didn't go from a blessing to a curse."
But back on the positive side, I look forward to finding out that Alpha contains roadmap information and specs for every CPU and GPU of the past two decades, at least. And if it doesn't, I look forward to entering some of that info into the Pro version for my own reference.
Also, if the Wolfram Alpha team is looking for a humanities dataset to ingest, the Perseus Project has a tagged, formatted, marked-up archive of texts, maps, and images from antiquity that they clearly have neither the funding nor the software engineering capability to do justice to. That project has stagnated for years (it always was a travesty from a UI and performance perspective), and the underlying dataset is crying out for at least the Web 2.0 treatment, if not the Alpha treatment.