Jon’s Radio Comments

December 4, 2006

Hunting the elusive search strategy

Filed under: Uncategorized — jonsradiocomments @ 4:40 pm

The original item is here.



  1. If you are serious about this, turn on your screen capture tool of choice, describe into the microphone what you are thinking as you click away. Then, afterwards replay and condense it down to some bullet points for posting.

    Comment by Chaim — December 4, 2006 @ 5:48 pm | Reply

  2. Its so true that many times I’m trying to pin down the best search term itself to use in Google, for successful results. There has been occasions where I browse an online (or perhaps mental) thesaurus in an attempt to diversify and find the successful search time.

    As a side, all this talk about searching through data and meta data to find what we want reminds me of the character called Laney from William Gibson’s trilogy set in San Francisco and Tokyo (Virtual Light, Idoru and All Tomorrow’s Parties) who could literally ‘feel’ his way through data (Anybody know what I’m talking about?!). Perhaps, as we are refining our search techniques, we are developing an intuition of sorts.

    Comment by Sam — December 4, 2006 @ 6:07 pm | Reply

  3. The example provided in comment #1 by Chaim reminds me a lot of work done here at our library. A unit within the library had been charged with putting together a digital repository based on MIT’s Dspace software. After implementing the repository they wanted to increase the participation of faculty in not just using the repository, but storing their work in it as well. This led to the webmaster, Dave Lindahl recalling work he had been involved in at Xerox. The research he had helped with is known as ‘Work Practice Studies’ and was meant to help software engineers refine designs for document management systems custom tailored to each client. It’s an interesting topic, and not widely known:

    Comment by Eric Likness — December 4, 2006 @ 7:30 pm | Reply

  4. There is an art to searching on-line. I taught math for a hot minute, and found that there are some methods to the madness. If you look at the process of solving a math problem, the first thing you do is gather your tools and put them in the tool box. The example I’d like to try is that of solving a complex trig identity.

    Before going to the search engine: what is the information I’m looking for? Is that REALLY what I’m looking for? If not, what else sounds like what I might like to find? This process takes only a few seconds, once you’re “in the zone,” but if you want to develop a consistent strategy, checklists are the way to go.

    Once you get your first batch of results, click through the first three search result screens to see whether you are in the right ballpark,– SEO marketing actually can hamper good searching, because the people who pay for placement (one way or another) don’t always have what you’re looking for.

    The analogy is to running through the formulae in your head and making sure you have at least two identities: one containing the place you are starting and the other containing the place you need to end up (the thing to be proved). then, to be safe, you throw in other identities that contain the two words…like tag clouding (brainstorming).

    Then, make some quick decisions about which links to click. The search engines sell placement, but they can’t sell thought patterns for dedicated searchers. skepticism will get your results a lot faster.

    The analogy is to play (in your head) with a few steps in, to see if there is a quick path to solving the proof, or whether there needs to be significant manipulation one way or the other. Sometimes, it’s easier to solve a proof backwards.

    Finally, once you have a better idea of the universe, you can narrow down your search by using “hacks.” In one of your examples, you probably could have saved a lot of time by putting quotes around “Atlantic Monthly Panel” because “state of the union” (at least in the US) is MORE likely to pull up government stuff than it is a forum panel speech.

    This might correlate to substituting known short cuts,–walking through a series of known proofs to see how they attack similar problems. Conversely, maybe what you find out is that you don’t have the right tools and you need to add something to the toolbox (e.g. another search engine).

    Consumers pay lawyers much more to do exactly this sort of thing everyday, –$100 an hour is cheap if you have to have the information fast!

    Comment by R. Mullen — December 4, 2006 @ 9:27 pm | Reply

  5. Here’s one for the collection. As if on cue, my wife just walked in having failed to locate an address and phone number on Switchboard. She’d spelled the last name correctly, and knew the exact town of the target individual as well: Montclair, NJ.
    I tried and, sure enough, nothing. So on a hunch I reduced the query to:
    And that worked like a charm. If the name had been Smith it would have been hopeless. In this case, the name is actually unique within the state of New Jersey, and while I didn’t actually know that, I did know it was at least uncommon enough to be worth a shot.
    Somewhere between those two extremes there’s the more typical situation. If a name yields a dozen screens to scan for some other clue — like a street name that rings a bell — it’s worth doing. If the name yields 50 screens it’s probably not worth doing, unless you’re /really/ desperate. How much of this calculus is unconscious, or barely conscious? A lot.
    The really counter-intuitive twist to this one, of course, is the town name. If that part of the query is correct, and I know it’s correct, what intuition leads me to generalize the query anyway?

    Comment by Jon Udell — December 4, 2006 @ 11:28 pm | Reply

  6. It’d be interesting to study this enough to break the main types of searches and techniques into distinct categories. This particular type of question is often most easily answered by search types b (use of “”) and c (filetype limiting) employed in conjunction or something like that.

    It’s the kind of thing that I, as a teacher, would like to see focused on a lot more in school. I don’t see nearly enough focus on search strategies or how to use boolean operators.

    Comment by Tom — December 5, 2006 @ 1:07 am | Reply

  7. Here’s a really nice example courtesy of Michael Rogers:

    Hi Jon,
    I recently got a search question from a former lab colleague, here it

    Can you help me out and use your superior searching
    skills to find a paper for me?? I know there is a paper out there,
    published in late ’90s I think, where some guys set up thermal imagers
    in an incubator and measured heat production in cells expressing UCP,
    in a 96-well format. It was work done in a pharma company, don’t
    remember which one. I cannot find this reference anywhere, and I need
    to. I could be remembering parts of this wrong, but the general gist
    is correct.

    background info: “UCP” is uncoupling protein, and the context is
    diabetes/fat metabolism research. I found the paper in about 10
    minutes after reading my email late one night, prompting him to ask me
    how I did it. Note: I used Pubmed a bit, but found the answer with

    (UCP | uncoupling protein) thermal-imaging 96-well
    (it’s good to use their boolean & phrase search features sometimes!)
    First reference had this:
    TSA was first explored by Mark Paulik, Ph.D. of GlaxoSmithKline as a
    method to measure the effects of drug compounds targeting metabolic

    So I searched Pubmed for Paulik [AU] and imaging
    and found the reference my friend had in mind:
    Development of infrared imaging to measure thermogenesis in cell
    culture: thermogenic effects of uncoupling protein-2, troglitazone,
    and beta-adrenoceptor agonists.

    Comment by jonsradiocomments — December 5, 2006 @ 3:44 am | Reply

  8. I try to do this at my website: Answer My Searches. However your post has inspired me to try switching to a format like this post where I document and narrarate the search process from start to finish instead of focusing on just on the answer.

    Comment by Greg — December 5, 2006 @ 4:19 am | Reply

    I’ve found it more valuable lately to put the whole damn thing in and Voila!

    Comment by raised by wolves — December 5, 2006 @ 7:28 am | Reply

  10. Funny how the Internet brings stuff to you when you’re thinking about it! I was just reflecting on this earlier today, how so many people I know get frustrated finding the hay in the needlestack, when it seems so much easier for me simply because I use boolean search tricks and quote searches. But then, I thought, there must be something more to it than this, it must be a way of thinking about information that floats the search results to the top of the list, and an increasing familiarity with the way search engines work. For me, this comes naturally, or at least it comes after 10+ years of using the ‘Net to find useful information.

    Which makes me wonder whether initiatives to provide natural language querying will ever truly be successful and useful. These projects attempt to reverse engineer what you successful searchers are doing with your queries, but algorithms cannot divine what the user does not know — and it’s the user’s knowledge that best refines the queries.

    Anyhow, interesting article.


    Comment by Rich Tatum — December 5, 2006 @ 11:32 am | Reply

  11. Jon, this is an interesting post. The first interesting point is that someone might be prepared to pay someone else to find/recommend an item for them using a search engine. This isn’t such a mad idea – after all, this is what reference librarians exist for: essentially they are adding the missing element in a search query (other than recall and precision), namely “authority” – i.e. “I, as a qualified X, have run this query for you, and consider the following results to be trustworthy.”

    The second is around searching strategies – and, I agree that there are a number of abstract stages in refining a search which would be useful learning points for any surfer. However, the first point of reference for anyone in trying to improve his/her search is to ask the question – what typ of search am I conducting? Am I looking for an item I know exists – in which case, do I have any data regarding that item (title, author, location, publisher etc.)?; Am I looking for an item which I feel may/must exist (but do not know for sure)?; Am I looking for an item which may not exist – I simply don’t have a clue?; Am I looking for information about a subject, rather than an item…? Each of these different types of search may require a different starting strategy and a different set of refinements.

    Comment by Mark Thristan — December 5, 2006 @ 12:57 pm | Reply

  12. One way to keep track is a FireFox plugin called Slogger –> it’ll keep a clickpath and the full contents of each page you visit so you can go back & review how you got there.

    A couple of years ago I did something similar called Dowser ( that searched multiple engines at once, combined, deduped & ranked the results, and saved the contents of each hit plus how you found it.

    Comment by aristus — December 5, 2006 @ 5:34 pm | Reply

  13. Right after reading this entry, I came across this entry: from Dumb Little Man – Tips for Life.

    Comment by fenux — December 5, 2006 @ 9:51 pm | Reply

  14. Very interesting post Jon.

    I think that after using Google for a while, not only do you develop search patterns and techniques, but you also train your brains to properly scan and filter out the good from the bad.

    Many of my searches are about finding some help on a Linux issue, and it really helps when you can, at a glimpse of an eye, determine whether the result leads you to a forum, a mailing list, or a useful answer.

    Comment by Rami Kayyali — December 6, 2006 @ 2:02 am | Reply

  15. Your point about the vocabulary problem, which in your second-to-last entry you call: “discover tags, assigned by domain insiders, which differ from tags I had assigned as a domain outsider” is incredibly familiar to me. I see it all the time in enterprise and site search log analysis. I first learned about it back in the late 80s, from Marcia Bates of the UCLA Library School, who talked about user vocabulary vs. domain vocabulary, and the importance of creating thesauri to connect the two.

    It’s so powerful that people can skip right over the answer to their question, if it’s not couched in familiar words. Marketing-speak, in particular, can be positively revolting. I’ve successfully reworded page titles and search suggestions to conform to user vocabulary, and seen significant increases in clicks in site search results. In another case, I convinced a site content creator to make a whole page oriented around a generic term that they didn’t use in their product line. As far as I can tell, people are finding it, reading it, and buying the product — a small chunk of purchases completely lost without that particular term.

    Sorry to go on about it, but this is how it feels from the search admin side of things.

    Comment by Avi Rappoport — December 6, 2006 @ 11:58 pm | Reply

  16. Interesting thoughts but as noted by others this is a common problem. I think it is actually on of the reasons MS Help files are claimed as “unhelpful”.

    Think about it, your knowledge doesn’t match the developer/help file builder knowledge and your “tags” don’t tie either. That makes it hard.

    Like anything else you have to start understanding the method to the madness before you can succeed quickly. That may be worth $100 per hour if you have to focus on marketing, sales or delivery tasks that have an equal or greater worth. Outsource where it makes sense is the rule of thumb.

    Comment by Jim — December 7, 2006 @ 2:26 pm | Reply

  17. I’m glad a couple people noted the reference librarian connection.

    In my view, the reference librarian can offer four avenues of help in the search process. The first, possessed by many but by no means a majority (IMNSHO), is “expanded cultural literacy.” This is a more exact understanding of what the seeker wants than what the seeker actually possesses. This leads to better search terms.

    The second, rarer and hard to find given how libraries silo staff members, is actual subject expertise. If you’ve got a legal question, go to a law librarian. If you can find one available.

    The third is searching expertise. Readers here no doubt can google on their own, but there are always “tricks of the trade” and an understanding that some search engines are better than others for a particular purpose.

    The fourth is access to proprietary databases. Lots of good stuff … often the best stuff … simply isn’t in the spiderable public domain.

    Comment by Bob Watson — December 7, 2006 @ 4:12 pm | Reply

  18. Interestingly I work for a large enterprise-software company in technical support – a good deal of our support, in the WS-* space particularly, and third-party products in general, is also largely comprised of effective Googling. This particular skill is widely variable, even within the support department.

    From the first inception of the web, I always maintained that what it really needed was a few good librarians..

    Comment by Doug Kretzmann — December 11, 2006 @ 11:45 pm | Reply

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Blog at

%d bloggers like this: