RDA-L Some discoveries of search engine results

Posting to RDA-L

On 8/21/2015 10:07 AM, Bernhard Eversberg wrote:
Well, “relevance” the metaphor, not the concept.
The googles, like Humpty Dumpty, choose the meaning that fits their purposes, but neglect to tell what their meaning is and what their purposes are.  A strong “caveat emptor” is appropriate here, as always, but it would take an awareness on the side of the customers that they are customers.

Should libraries, for their catalogs, adopt this attitude of the googles? I’m still inclined to say, as things go, it amounts to deception.

Relevance as a metaphor? I like that.

I think that “relevance ranking” is a fact of informational life that catalogs have to accept–until something better comes along. And yes, it is a deception, but there are many deceptions in the world that we all silently agree to. I am sure I am not the first to say that if deceptions were abolished, human society would disintegrate.

That said, the solution is to build something that gives people a choice other than relevance ranking, and I think library methods could go a long way in providing the directions the new tool could take.

Until that happy day, I am a believer in “telling the truth” and that could be done simply enough with a link next to “Rank by relevance”. The link could say something like, “What does this mean?” and a short explanation would pop up, hopefully with a link to a page that discusses the issue in more detail.

Of course, I realize that almost nobody would click on it, but I am sure somebody would and if they found the page interesting, they might tell others about it. In my experience, I have found that some people like knowing what is really going on, in spite of the complexities.

I just saw the message from Kathie Goldfarb:

Instead of decrying the use of keyword searches, we ‘just’ need to have the ability to define what is relevant.

The point is that while we can tweak relevance, it will always lead to a poor result–much worse than anybody will get in the Googles. Why? First, the Googles have vast amounts of personal data that we can never have and that is one of the primary drivers of why their relevance ranking works as well as it does. Second, our catalogs have very little data to work with while the Googles have full text. Our relevance results will always be worse.

But the main reason is that the catalog was designed on completely different foundations from what the public experiences today. In those earlier (physical) environments, there was a genuine power in the catalog if you used it correctly, but keyword searching (modern environment) has hidden it. As a result, the catalog must be redesigned for today’s environment. That does not mean that the records themselves need to change. Changing records is irrelevant: you can change all of the records and still not make a difference in how the catalog works.

It is similar to how dictionaries function online. The individual dictionary entries didn’t need to be changed–they had to find a way–or several ways–for modern people to search dictionaries. They could have changed all of the entries and built a tool that had everyone to do alphabetical searching like in a printed dictionary.

That would have been silly but it is what I think has happened with the catalog. We can change the records as much as we want but the catalog itself remains broken. We need to design it so that people today can see the power of the catalog in a keyword environment. I have ideas in how this could be done, but my ideas are probably wrong. Still, until we make it function in a keyword environment, the users will see only inferiority.

The current direction of information is developing at a rate that I find almost unbelievable. There is a great talk of Evgeny Morozov “Big data, Small politics” where he discusses some of these directions and draws political implications. In one part he discusses Google Now, which I have never really used and he mentions “autonomous search”. Perhaps I should find out about it. Here is a link to that part of the talk. https://youtu.be/Ba0rIaEftKU?t=22m

Without using the term “autonomous search” I discussed this in a podcast and I called it a nightmare (https://blog.jweinheimer.net/2010/12/cataloging-matters-podcast-no-7-search.html). I didn’t know it would appear so quickly. I am sure however, that it will be very popular with lots and lots of people.