In defense of irrelevance

Many thanks to Latasha, Amy, John, and Adam for bringing me back to these articles again, and again, and again…

But first, two points about those articles which have little to do with my main point, because, well, irrelevance.

First point:  Personalizing search results based on peer searches?  Previous searches?  Assignment info?  It smacks of a peculiar sort of idiocy that only brilliant people tend to get.  The notion is correct about one thing: If your average undergraduate could type the search, “Can you give me the information I need to ace this paper?  And maybe the paper itself?” they totally would. There’s so many reasons not to go this deep in personalization, including privacy issues.  But my reasons include:

• Basing the searches off your peers assumes that your peers have been working harder than you, which, if they were, would make them not-your-peers.
• Basing it off of your previous searches assumes a continuity in undergraduate searching that doesn’t exist.  If I was working on one class an hour ago and am working on something else now, STOP USING THE OLD DATA!
• It sets up a method of finding information that is not relevant to the lives of these students, whether they continue to be researchers or not.  It’s different sides of the spectrum:  If you’re faculty, your main research depends on certain researchers and journals (and what your advisor in grad school did); in essence, you depend on the same searches (or PLN), tweaking them once in a while.  If you don’t get an advanced degree, well, then college was the last time that someone gave you a textbook, or a search engine that accessed academic information, or a search engine that would know anything more than your past searches (or would use that information to do anything more than make purchase suggestions), or (at least at work), the last time anyone will let you choose your own topic (and then say nothing more than, “Start researching!”).  And college was also the last time that someone said, “This presentation/proposal/idea is good, but it’s not LONG enough!” but that’s not relevant (to this irrelevant point).

Second point:  Reference desks and academic search engines are things that, the better they work, the less time people spend using them, and the fewer results people need.  This is in complete conflict with many metrics of their performance.  Discuss.  Irrelevant point (to this irrelevant point): I am 41.  I have asked an actual reference librarian an actual question once.  I was required to for an MLIS class.  I learned something cool; I have yet to use that learning since.  I understand that some people think the last sentence contains a contradiction.

But enough of that: here is my relevant point (which is about irrelevance): I work here with Adam’s article:

 Furthermore, I really think the author is on to something when they bring up the value of serendipity in the stacks. I feel like this is an idea that very quickly gets overlooked when all the too brilliant, creative, and innovative minds out there start fantasizing about all the ways computers and data can “make life easier” for us. While I don’t necessarily always agree that the hard way is the only good way, knowledge and understanding are certainly amplified when they are worked for and not passively received.

It’s a little tricky to understand: How can it be bad to give people the information they want, and only the information they want, once you know (through asking or awesome computer goodness) what they really want to know? After all, that’s one measure of search effectiveness, there! What’s that voice in our head that starts saying “whoa” when articles talk about personalizing results with outside context?

Let’s move to a smaller domain: geography. If you want to know how to get from where you are to where you’re going, A GPS will tell you that. You type in your destination, and it tells you exactly what to do, and you get there. And I promise, someday, you won’t even need to do anything; you’ll type where you want to go, the car will drive you there, then you’ll get out, and another voice (in your ear) will tell you where to walk. In fact, some have posited a similar future for all knowledge: People (or computer programs) that whisper the answers in your ear. But this geographical relevance comes at a cost: You really do lose your ability to navigate. Seriously, read this article.

Now true, you’ve lost your ability to recite epic poems from memory, and do long division, and no one’s really complaining. But, seriously, navigation: That’s a pretty basic skill. Let’s break that down: You’re losing the ability to stay out of danger when the power goes out, or when the GPS is just wrong (see the article’s notes on hikers). You’re losing the ability to see when the GPS is doing something technically correct, but nonsensical, much less formulate an alternative. You’re losing the ability to know if something’s there unless someone tells you it’s there. Seriously, have you ever used a GPS to get somewhere, and your passenger says later, “We should go to the place I saw on the way here,” and you said, “What place?” because you only saw a GPS street grid the whole time? In essence, you’re losing the ability to say that you know your part of the world, your geographic “common sense”. All of these things are especially painful to people who are “natural navigators,” who never forget how to get somewhere they’ve been, who seem to be able to point at their destination, even when it’s miles away, who spend hours zooming in and out on Google Maps because it’s fun, who suddenly turn right while driving in their own town because they’ve never been there, who know, just know, that there’s the shortest way, and there’s the quickest way, and then there’s the BEST way, and those are often three different ways, changing with traffic, weather, and the flavor of the day at Culver’s.

Then again, some people were never “natural navigators”; they got lost a lot, maps were these horrible messes that never folded back up, they’ll never go hiking BECAUSE it’s a navigation nightmare. To them, GPS is a godsend, and they are looking forward to never being lost again, and they’ll call their mom or their friend if the GPS is wrong.

This example maps perfectly to knowledge in general. To “natural researchers,” information is simply a waypoint to knowledge, to authority over information. They want to know how things connect, how to examine “information maps” and navigate where they’ve never been, to be able to say where no one’s ever been. And while they have envisioned, and partially made, the system that says, “Oh, here are the best ten papers for your assignment,” or, “this is the tallest building in the world,” or, “here’s your A paper,” and they get the bills paid because other people really WANT that system, they long for the experience of standing in front of a library shelf that whispers, “Oh, this book next to the one you want? It’s kind of related, but not really. You should grab it anyways.” Not to mention the bookshelf on the other side of the room yelling, “I have useless stuff about Bill and Ted’s Excellent Adventure! Jam it into your paper because you CAN!”

In short: We are working towards two worlds of knowledge searches, one using metadata to increase relevance and shorten the path to results, and one, likely always requiring human intervention (or actual machine intelligence) to use, that lays bare the metadata used to define relevance, to go beyond relevance, to allow users to find the best results, “best” depending on the user, the context, and the flavor of the day at Culver’s.

Oh, and if you think about it, all of the irrelevant stuff? It wasn’t really irrelevant, was it?