Friday, August 13, 2010

The tale of a difficult search (hard for me at least)

In the category of hard search problems caused by the way your mind works…

I wandered into my manager’s office yesterday and mentioned that I read on article about how mathematicians have a new wiki-like website to jointly solve big, hairy, complicated problems.   

“It’s incredibly cool,” I said, “they post problems, math-geeks work on them collaboratively and collectively real work gets done quickly.” 

He was skeptical.  “I ran something like that back in the day…and it petered out after a while.  Still, what’s the URL?”  He turned to the keyboard waiting to hear. 

And I couldn’t remember.  So, being a good researcher, I started searching and found that it didn’t pop to the top.

Here’s what I knew at the moment: 

  •  It was an article I’d read within the past couple of days.
  •  I thought I’d read it in Newsweek.
  •  I knew it was about math, and probably had the words Wiki or Blog in it since it was about working together.
  •  The article was about mathematicians working collaboratively.

I just wanted to find the article.  Should be simple, no?

Well… it wasn’t.  Let me recap my search for you. 

Here’s where I started—searching for a news article within the past week (limited by the left-hand side nav restrictions)

[ math group proof ]   -- I set the left-hand nav to search“Everything” and “Within last week” 

Nothing.  Many of the results were invaded by the recent announcement of an attempted proof that P ≠ NP.  Still, I tried again, altering  “proof” to “proves”…

[ math social group proves ]

Then adding in the idea of “social” proof…

[ math social group prove social ]

And taking a new tack, tried adding in “blog”

[ math group proof blog ]

I had a half-baked notion that the wiki/blog/thing I was looking for was called “polymath” (it would be a clever name), but the results looked unpromising.  So..

[ polymath wiki Newsweek ]

I’d searched for about 15 minutes at this point and… I gave up for a bit.  Just to let my brain stew on this for a bit.  This is sometimes a good idea. 

I came back to the search after an hour with a new search term in mind:  “crowdsource”! 

So I tried the obvious first search,  using the new search term and adding in Newsweek to restrict the search.  Again I limited my search to the last week. 

[ crowd-sourced wiki Newsweek  ]   

(I didn’t use a SITE: restrict because I didn’t want to overly constrain the search to JUST Newsweek.  I thought that perhaps a newspaper might mention “In the recent Newsweek article…”  And I’d be good to go. 

But this wasn’t good either.  But I figured maybe if I just added back in the MATH term, that would be even more focused.  So I tried:

[ crowdsourced wiki Newsweek math ]

Which showed me a link to the Interceder site (a news aggregator site) that collected information on (in this case) “crowdsource” article. 

I scrolled through their collection of articles, and finding it long (and tiny font!), I used Control-F to look for “math” in the text. 

FINALLY I got a break.  About half-way down the page there was a link to an article on Dr. Dobb’s Journal  (a pretty well-known computer / technology magazine) with the intriguing title: “Massively Collaborative Math ” (Aug 9, 2010)

Paydirt!  The Dr. Dobbs article then pointed me to an article in the San Jose Mercury News.  (Which is my local newspaper.) 


Now... what’s odd here?  Why was this search so hard for me? 

1.  I didn’t remember that it was a site at UC Berkeley… or that Stanford was also involved.  (I live close to Stanford and have even taught there!  How could I have NOT remembered that about the news item?) 

2.  I misremembered the source.  I would have sworn it was Newsweek and NOT the Mercury News.  But I should have known better.  (When I got home that evening I found Newsweek on my desk—it was just the most recent thing I’d read, NOT the actual source of my initial read.) 

3.  I correctly remembered that I’d read it recently, but I did NOT remember that it was that same day!  (In fact, just 5 hours earlier.) 

Retrospect:  Looking back, what could I have done differently? 

It’s clear that using the word “crowdsource” was the key breakthrough.  Since the article was very recent, a very simple query [ crowdsource math ] brings up 3 perfectly good hits on the MathOverflow site.  (Although I recognize that this search probably won’t work correctly in a year or so…) 

And, in the end, all of my floundering around with terms like “Newsweek” and “wiki” were not productive.  The article I really was trying to find (the newspaper article) wasn’t on Newsweek at all.  I should know better:  Let the search engine search… that’s what it’s good at. 

I should have focused my efforts on looking for better terms to describe the thing I was looking for.  “Crowdsourcing” turned out to be the key term that got me to the target.  

Sometimes, I have to take my own advice. 

1 comment:

  1. Now the trick is to go through this research process with someone else. I work in a library where we get requests for articles where the source, date, title and/or author are remembered incorrectly by the requestor. Usually they can remember the topic but not actual keywords. I haven't put "mind reading" as a skill on my resume yet but might consider it. :-)