Tuesday, November 8, 2011

Alexis Madrigal's research process

In his recent Google+ post, the Atlantic Monthly writer Alexis Madrigal noted down a recent research task that he did.  

I'm copying most of the text of his post here so I can comment on bits of it... 

He wanted to do some background research on "how people want to tow icebergs for fresh water supply."  (My quick characterization of what he wanted to find out.)  

What he did to research this topic was pretty interesting, and since he wrote it down we can take a look at his process.  

From his post: "... I knew that "iceberg towing" was a unique phrase that could be pumped through every search engine imaginable to find interesting information. And that's really what enabled me to write a comprehensiveish post about the phenomenon in a short amount of time. Here was my rough order of operations, although given tabbed browsing, many of these things were going on simultaneously..."  

Madrigal then goes on to list the things he checked.  

He's right to notice that a phrase like "iceberg towing" is so perfectly descriptive of what he wants (and wonderfully uncommon) that almost anything he finds is going to be just about on-target.  

My comments come after each of his steps.  

1. Google Books (first with only free books shown, then later with all previewable books shown)

This is actually a great first move, as Google Books has not just books, but also many popular magazines.  Popular Mechanics (1978, vol 149, no 1, pg 47) has a wonderful picture of a ship nearly being sunk by shards of ice falling from the iceberg.  That's one of the values added by popular magazines--images and "inside stories" that are often missed by more academic monographs and technical articles.  

2. Standard Google search, followed by searches just within .gov sites and then looking only for PDFs

Using searches like [ "iceberg towing" filetype:pdf ] and [ "iceberg towing" site:.gov ].  These limit the search results to PDFs (usually articles from academic sources) and to government sites (.gov).  Another useful trick might have been to try [ "iceberg towing" site:.edu ] 

3. The Chronicling America collection, late 1800s and early 1900s newspaper accounts as well as the other Library of Congress databases, though almost all of them came up empty. 

I have admit I'm not surprised by the lack of content here.  Still, it's a smart move to check these kinds of special collecitons, and it doesn't take long to check.  

4. YouTube

I think I would have checked on Google Video, as it's a superset of the YouTube results.  One search would do both sets of content.  But Alexis is following advice I give in all of my classes about being a good searcher:  Check other kinds of content beyond just web-search.  

5. Flickr Commons, then Flickr

This seems redundant as the images returned from both sets are the same.  On the other hand, you can quickly figure out that perry709 was the person who took those photos, and by doing the obvious searches, you can find out that he even has an iceberg tattoo and clearly works in marine research.  But we digress.  My point is that a little bit of chaining from a known source can lead to even more discoveries and confirmation about the credibility of a source.  (And I found a few photos that perry7009 hadn't put onto Flickr, but were under his Imgur account--images that hadn't appeared anyplace else.)    

6. Here, I pinged Twitter, which brought Project Habakkuk and Brewster's Millions to my attention. Twitter is great for testing what people first think of when they hear some phrase like "iceberg towing." It lets you play defense against common assumptions.

I completely agree.  Even though I work at Google, one of the great things I've seen in truly power searchers (what I call "ultra searchers") is that they always mine their social networks.  Twitter is a great, fast way to tap into a large community, including people you might not know directly, but who can point you to places you would have never found in any other way.    

7. Google Scholar, followed by searching in the footnotes of the papers found there, which led to the discovery that there were two major papers put out in 1973, one in the Journal of Glaciology by Weeks and Campbell (later published in modified form in The Bulletin of Atomic Scientists) and another by RAND researchers J.L. Hult and O.C. Ostrander. Both were gold mines of information.

Google Scholar is one of the great, but underused, repositories around.  For a research topic like this that skirts around in the academic literature, this is a great place to go.    

8. By now, many signs pointed to John Isaacs at Scripps Oceanographic Institute as the key figure in the midcentury, so I went to Scripps' archives and found a bunch of photos of the guy as well as a book about him, which was previewable in Google Books

Within a few clicks you'll arrive at the Scripps Archive, which is (apparently) not indexed by Google!  Discovering dark-web repositories (as is often the case with special-library archives that run their own search engines of their content) is often a way to find things that other, less motivated searchers, will never discover.  If you're looking for a new story or a scoop, this is a great technique for finding good stuff.     

The other thing he could have done was to use the UCSD "Ask a Librarian" service.  Many libraries run a special email (or live chat) answer service.  I think I might have checked out the "Ask a Librarian" about what resources Scripps has on "iceberg towing" -- you never know what's not in any online archive and just sitting in an archive box on the librarian's desk.  

9. Meanwhile, the RAND report led me to a slough of other sources and filled in the deep history of iceberg towing from South America.  

Makes sense to me too. 

10. I decided to follow up on an early hit from TIME Magazine's online archives about a conference sponsored by Prince Faisal of Saudi Arabia at Iowa State University (!) about iceberg towing. Turns out, Iowa State maintains a small archival collection about the conferences, which began in 1979 and ran for a few years.

This is a fantastic find.  (Check out the site: http://www.lib.iastate.edu/arch/rgrp/0-4-4.html)  I agree with Alexis, this was pretty much unexpected, so it's another great resource to draw on.  

11. That was pretty much it. I did some more extensive searching around "iceberg towing + Labrador Sea" and that yielded a bunch of interesting information about the business and science of towing bergs for the oil industry. Specifically, a 2003 Smithsonian mag profile surfaced and an excellent document from Grand Banks Iceberg Management from 2007, which pretty much tells you anything you need to know about how to actually get the practice done.

Not to be picky, but a search like [ "iceberg towing + Labrador Sea" ] won't actually find you much.  (The + was ignored, and so the query is really just [ "iceberg towing Labrador Sea" ] -- which is okay, it finds a few results.  

But it makes me think that a REALLY interesting search might be to use this strategy more generally.  Example:  use the "fill in the blank" (aka "wildcard") mechanism.  Like this: 

[ "iceberg towing" "* sea" ]  -- which finds a video I hadn't seen before 


[ "iceberg towing"  "* ocean" ] -- which finds even more results

The strategy there is to try and find new kinds of specific results (e.g., "Arctic Ocean") rather than just the Labrador Sea.  

All in all, I was pretty impressed--Alexis showed a real mastery of the art of search.  Kudos to him for documenting a longish, deep and determined search strategy and sharing it with everyone on the web. 

Search on, Alexis! 

No comments:

Post a Comment