While visiting Slate recently, an article on a ’15-minute per day’ exercise routine caught my eye (check out the video associated with the article, below). I am no lover of exercise in general; I can abide and tolerate it, but I’ve never left a workout exhilarated or anything. Having said that, I recognize its importance, and so want to find ways to make it part of my routine. Fifteen minutes a day sounds do-able. Here’s the thing about this routine: it was written by a great Dane in the early 1900s. It involves a lot of self-rubbing and contortions. There’s no associated exercise video.

Like many people that read the Slate piece, I was intrigued and interested in trying the routine out. The book, though freely available, is a bit on the dense (and old-timey) side. So, I decided to start summarizing the exercises. More than that, I opened the editing to other readers of the Slate piece. Turns out that there was really only one other person invested in the summarizing task, so between the two of us, we got it done. It’s quicker than trying to read all of the book’s instructions, but not as good as a video would be. Still, it’s a start.

Summary of J.P. Müller’s “My System”

From the Slate article:

This is an article that I co-wrote with the director of McMaster’s Integrated Science program about the some of the new ideas and collaborations we’re trying. It’s great to be able to share our ideas with a wide audience. The article was originally published in EDUCAUSE Review.

Colgoni, A. & Eyles, C., January/February 2010. A New Approach to Science Education for the 21st Century. EDUCAUSE Review. 45(1):10-11.

[This is an article I wrote for McMaster's Library FYI as a member of our Teaching with Technology committee, highlighting tools and technologies that some of our staff may not know about]

“It is an intellectual travesty. The idea of jumping from concepts to numbers without any attention to how the numbers are produced and how they can or can’t be compared is simply anti-intellectual.”

Rory Litwin, Library Juice

There hasn’t been much dialogue or discussion in the library community about the value (or problems) of Wolfram|Alpha (W|A). A few have harshly criticized it (like in the quote above), while others have noted that it could be a boon for the reference desk. The media broke the unveiling of W|A as either ‘a Google/Library Killer’ or ‘mostly useless’. W|A is not a killer of anything (including libraries), nor is it useless; it is simply a tool. It is what the user makes of it, whether it provides valuable fodder for deeper thinking, or simply does a calculation.

What is Wolfram|Alpha?

Wolfram|Alpha is a web-based product from physicist Stephen Wolfram and his namesake company, Wolfram Research. Wolfram is the author of “A New Kind of Science” and creator of Mathematica, a computer program for mathematical analysis and visualization. It is easier to describe what it’s not, however: it is not a website search engine. Articles dismissing W|A because it can’t find the lyrics to the latest song miss the point. Google doesn’t ‘know’ the answer, either, but it can point you to a page that does. The difference is that W|A doesn’t search the internet at all. Where Google crawls every webpage, W|A takes many specifically curated datasets, and returns both that raw data and also quantitative analysis of that data (by tapping into the computational analysis power of Mathematica). W|A does not offer any significant qualitative analysis, nor interpretation of that data. That part is up to the user.

So, then, what can we call this kind of search tool? W|A bills itself as a ‘computational knowledge engine’, but that is not a completely accurate description, for W|A does not serve up knowledge, but rather data, statistics and other bits of information. Perhaps a more apt description is a ‘data and statistics computation engine.’

What is Wolfram|Alpha good for?

Wolfram|Alpha is particularly useful for finding out information that has a quantitative component to it. That is, numbers, statistics, graphs, etc. So, it’s great for looking for information on chemical compounds, astronomical bodies, populations of places, weather trends, GDP comparisons, and so on. These all have measured numbers that fit into W|As databases.

Wolfram|Alpha also contains names of books, movies and people (and their dates of birth). What else? BMI, intersections, surfaces and shapes, human genome DNA sequences, musical notes, scales and chords, calendars, colours, lat/long, nutrition data, scale conversions, sets and sequences, and tides. The list goes on, so I encourage you to explore the possibilities, to discover how you might be able to use it. New features and data are added continually. (For added inspiration, check out the W|A blog, and on that same page, the ‘Interesting Inputs’ and ‘Output Galleries’. See also the new Wolfram|Alpha for educators).

As an example of how Wolfram|Alpha can be used, technology thinker Jon Udell wondered how much nine Watts actually ‘costs’ in real-world terms. Not able to perform the conversions in his head, he used W|A to his advantage. W|A will offer useful references for some measurements, as well as perform math on those measurements. 9W, for example, is about half the energy expended by the human brain, while 9W * (30 *24) (if you drew 9W of electricity 24 hr/day for a month) is about half the energy released from combustion of a kilogram of gasoline. ( 1 kilogram / density of gasoline ) / 2 produces a numerical result, because W|A replaces ‘density of gasoline’ with the appropriate number. In the end, the dollar cost to Udell was negligible (about $1 a month), but relating 9W to gasoline (and driving distance) helped him understand the energy cost.

What are the issues?

Wolfram|Alpha is not without its problems, however. First of all, it’s unclear what kind of data it contains without actually probing it to find out. The examples that W|A provides on its website indicate that it has information for dates, places, stocks, performing math functions and calculations. During the screencast that Stephen Wolfram hosts, he shows that W|A contains some data from medical research, like LDL cholesterol levels in the population. Given that information, it would still be difficult to predict what else it contained. For example, now that we know there is some medical information, we might try a search for autism or Crohn’s disease, but there would be little information returned. This is a function of what numbers and datasets that W|A has included, and also not being able to predict what those are.

It can be difficult sometimes to know the exact way to structure a query in Wolfram|Alpha. For example, if one wished to see how much larger Germany’s GDP is compared to France, the natural language query, “How much bigger is Germany’s GDP compared to France’s” would not compute. To ask this question, the user would need to know to input “GDP Germany/France”. A restructuring of the natural language query to “Compare Germany’s GDP to France” would also produce a (different) result.

However, in some cases, W|A is improving its ability to answer natural language questions. Back in March of ’09, one review of W|A said,

“If you ask “How fast does hair grow?”, it can’t parse or answer that query. But if you type in a speed, say “10cm/year“, it gives you a long and quite interesting list of things that happen at about that speed, involving glaciers melting, tectonic shift, and… hair growing.” (Lenat 2009)

Fast forward to 2010, and W|A now produces an answer (0.4mm/day, in case you were wondering).

Another problem, of more concern for information literacy, is the sourcing of the provided information. Particularly, the references used for the result are not concrete, not allowing the user to directly determine the source. For example, a search for Earth’s Orbit produces details about orbital properties, an illustration of earth’s orbit, and below that (like on every page), a sources link. Clicking that link produces a pop-up window that reveals that the primary source as “Wolfram|Alpha curated data, 2009″ and “Wolfram Mathematica AstronomicalData”. Below that is a lengthy list of ‘background sources’, which could be very useful to a researcher, but W|A is explicit that these sources may or may not have contributed to answer provided. Having said that, the provision of some references makes it preferable to completely unreferenced information.

Furthermore, there is the issue of trust in the accuracy of results that W|A produces. While much of the data provided is valid, it cannot be considered totally accurate (particularly with mathematical derivations), and a visit to the ‘bugs’ forum can demonstrate where errors are ocurring. W|A must then be evaluated as any website should be, and depending on how the information is being used, be verified using another source.

Conclusions

Wolfram|Alpha is an excellent resource for quickly checking factual data, and for automating calculations that would be time-consuming otherwise. It is a treasure-trove of interesting information, but lacks a way to predict what information the system contains or the best way to manipulate that information. It is still a new resource, and will become more useful and valuable as more data is added and bugs are ironed out. I encourage you to try it out for yourself (also available as an iPad/iPhone app), and see if you think it might a valuable addition – either personally, or on the reference desk.