just an asterisk

Ok so i think we’ve established that I am a geek and proud of it.  I love data, tools that make visualization easy and just generally, the power of knowledge.  I love that a slow but steady direction for the web is the improvement of search, both with improvements to traditional sites like Google and with interesting newcomers to the seen like WolframAlpha and Cuil. 

Recently, I’ve read  several articles on the brilliance of Wolfram Alpha, a new search tool from the computational giant Wolfram (makers of Mathematica).  The trend is obvious – we are aiming for natural language search that combines data from many sources into one view.  What does this mean?  Let’s do an example to explore a bit.

Tylenol vs. Aspirin vs. Ibuprofen

Here is the test: I simply type Tylenol vs. Aspirin vs. Ibuprofen into several search engines and record the result as of 8/8/2009.

Let’s start with plain Google.   Here is the result.  I received 715,000 hits.

The top 3 results from Google (for me) are these:

Not particularly enlightening – it does seem like result number 3 is pretty good, but I’d have to go somewhere else (away from Google) to know for sure.

Now let’s try the exact same search in Wolfram Alpha (don’t worry, we’ll give Google another chance in a bit).  Here is the result.

This is looking a bit better.  It figured out that I’m not a doctor and translated Tylenol to acetaminophen.  It is also showing data side by side, providing comparison – albeit of data that I’m probably not interested in if I’m just trying to figure out what to take for my hangover.  What is missing here is a filter, some sort of way to go a chemical/medical direction, as shown above or to go a average Joe direction, as in the (partly) fictional case of a hangover.

Now let’s try a new product that (despite claims of Google to the contrary) is supposed to be a rival to Wolfram Alpha.  It’s called Google Squared.  I put the same search in and here is the result.

Not super inspiring but there is hope.  I now spend a few seconds putting my queries into the separate boxes and hitting “square it”.  Here is the improved result.  What is cool here is that we’ve got a variety of results.

 

And now things get interesting, because you can add and delete columns on the fly.  After another minute of playing, I get the exact comparison that I want:

That’s all I really wanted – the dosage, a bare bones description and a picture.  It’s still not perfect.  I would love to type “hangover” in as a new column and get a recommendation (yes, yes, no) on which one to take.  You can also click on any result to see Google’s confidence in the result (it always seems to be low…).

Now that we’ve gone through all that – what worked and what didn’t?

Limiting results – I don’t need to see 715,000 hits.  I was really just looking for 3, with some data on each.  Traditional Google fails because it’s overload.  Even if the best results are on top, I still have to comb through other websites (which takes me away from their advertising and loses them money, as a side note).  Wolfram Alpha goes in the other direction, limiting results to the point of mere scientific curiosity.  Google squared limited results but allowed me to add more data to my view and I think this is why I was happiest with the result.

Categorizing data – Plain Google didn’t categorize my results.  Wolfram Alpha chose the categories for me.  Google Squared let me attempt to categorize data as needed.  Again, this flexibility was what provided the best view of results for my query.

Magical data categorization, based on the information gleaned by watching me over the past year, could easily have been included in my search results for Google.  Although you can read my bio (AB), none of these sites knew of my background in Chemistry, and the the result sets were primarily scientific in nature.  The reason for this is quite obvious, data is easier to manage than opinion (the foundation of probably 99% of the web).  Chemical data is the easiest thing grab when someone searches for a chemical, but where was the insight into me?  I would have loved a scientific paper on my search or a list of hangover remedies (although I would have been ashamed if the hangover remedies were proposed first).  Admittedly, it didn’t take much more Googling to find these results, but I wish that categories were proposed based on me.  It would make me feel like all this information on my habits was worth having in the cloud, rather than a concern for my privacy.

It seems like my wish for a mind-reading search tool is still nowhere close to being granted.  Given that so much of my data is publically (and frighteningly) accessible, this is disappointing.  However, the future is bright –new tools like Wolfram Alpha and Google Squared are going in the right direction – they just aren’t there yet…  Certainly, these tools are making search easier, but there is room for a brilliant product, that uses your history in a positive way to make accurate suggestions, categorized by relevant categories and easily adaptable to the whim of the user.  When this arrives and once the privacy concerns have been answered, everyone is going to take notice. However until that day, these incremental improvements are not going to make waves.

~ab

This entry was posted on Saturday, August 8th, 2009 at 6:21 am and is filed under future post. You can follow any responses to this entry through the RSS 2.0 feed.

We would love to hear what you have to say! You can leave a response, or trackback from your own site.