Comments on: Google is not enough http://dubroy.com/blog/2007/05/09/google-is-not-enough/ on programming, usability, and design; by Patrick Dubroy Wed, 20 Aug 2008 13:41:46 +0000 http://wordpress.org/?v=2.0.4 by: Dubroy.com/blog » Greatest Misses: Popular posts from the last year http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-4408 Mon, 15 Oct 2007 16:46:35 +0000 http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-4408 <p>[...] Google is not enough [...]</p> […] Google is not enough […]

]]> by: Patrick http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-736 Tue, 22 May 2007 14:16:06 +0000 http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-736 <p>True, this study did rely on the people to report what they did, rather than being directly observed. It's not the most scientific study. But if people were going to "stretch the truth", I would think they would do it towards making their behaviour seem more perfect, e.g. "Oh, I was looking for Jack's phone number, so I searched for `Jack Tripper University of Toronto phone number'". In the paper, many of the people are apologetic or even embarrassed about how they actually performed the search.</p> True, this study did rely on the people to report what they did, rather than being directly observed. It’s not the most scientific study. But if people were going to “stretch the truth”, I would think they would do it towards making their behaviour seem more perfect, e.g. “Oh, I was looking for Jack’s phone number, so I searched for `Jack Tripper University of Toronto phone number’”. In the paper, many of the people are apologetic or even embarrassed about how they actually performed the search.

]]> by: MH http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-735 Tue, 22 May 2007 14:02:47 +0000 http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-735 <p>But what about the idea that you rely on what users do, versus what they tell you they will do? Does that not apply here?</p> But what about the idea that you rely on what users do, versus what they tell you they will do? Does that not apply here?

]]> by: Patrick http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-640 Wed, 16 May 2007 15:32:50 +0000 http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-640 <p>I guess that's a fair criticism. In one paper I read yesterday, the analogy they used for that kind of conclusion was "it's like observing a family listening to the radio, and concluding that they don't want pictures". But still, I think there is a certain cognitive difficulty of specifying exactly what you are looking for, and it's often easier to follow a slower path and orient yourself using the context.</p> I guess that’s a fair criticism. In one paper I read yesterday, the analogy they used for that kind of conclusion was “it’s like observing a family listening to the radio, and concluding that they don’t want pictures”. But still, I think there is a certain cognitive difficulty of specifying exactly what you are looking for, and it’s often easier to follow a slower path and orient yourself using the context.

]]> by: e http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-638 Wed, 16 May 2007 15:14:26 +0000 http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-638 <p>I did get around to reading the paper. The basic gist seems to be "hey! people orienteer!" which I'm find with; but I think part of the reason for some orienteering is due to the crapitude of keyword search. </p> <p>There are good reasons to orienteer: finding context, searching for unknown information, or performing a survey of information. However, I would argue that orienteering for known information (ie, something like a phone number or a meeting time that the user has seen in the past, but doesn't know the location of) indicates a failure of the search facilities. That isn't to say that the user shouldn't be given the option of orienteering, and that tools shouldn't be built with orienteering in mind, but that search should be improved so that orienteering isn't necessary.</p> <p>Thanks for the article citation. It's nice to dip into academia again. =)</p> I did get around to reading the paper. The basic gist seems to be “hey! people orienteer!” which I’m find with; but I think part of the reason for some orienteering is due to the crapitude of keyword search.

There are good reasons to orienteer: finding context, searching for unknown information, or performing a survey of information. However, I would argue that orienteering for known information (ie, something like a phone number or a meeting time that the user has seen in the past, but doesn’t know the location of) indicates a failure of the search facilities. That isn’t to say that the user shouldn’t be given the option of orienteering, and that tools shouldn’t be built with orienteering in mind, but that search should be improved so that orienteering isn’t necessary.

Thanks for the article citation. It’s nice to dip into academia again. =)

]]> by: Patrick http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-546 Thu, 10 May 2007 18:47:49 +0000 http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-546 <p>You're right that to some extent people do this because they are used to being disappointed by search results. If a "perfect search" did exist, people might be more inclined to do a direct search. They didn't ignore that in the paper though; I just didn't choose the best example. Here's what the paper has to say about why even the perfect search is not enough:</p> <blockquote> <p>However, even when a person knows exactly what they are looking for, the perfect search engine might not be enough. Consider Rachel, ... she attempted to locate a document that she knew existed in her file system. Although she knew exactly what document she was looking for ... she could not describe the document, its contents, or its location in advance:</p> </blockquote><blockquote> <p>I don't know how I could have the directory [the document was in] in mind without knowing its name, but I felt sure which it was</p> </blockquote> You’re right that to some extent people do this because they are used to being disappointed by search results. If a “perfect search” did exist, people might be more inclined to do a direct search. They didn’t ignore that in the paper though; I just didn’t choose the best example. Here’s what the paper has to say about why even the perfect search is not enough:

However, even when a person knows exactly what they are looking for, the perfect
search engine might not be enough. Consider Rachel, … she attempted to locate
a document that she knew existed in her file system. Although she knew exactly
what document she was looking for … she could not describe the document, its
contents, or its location in advance:

I don’t know how I could have the directory [the document was in] in
mind without knowing its name, but I felt sure which it was

]]> by: e http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-544 Thu, 10 May 2007 18:22:03 +0000 http://dubroy.com/blog/2007/05/09/google-is-not-enough/#comment-544 <p>Regarding orienteering: I haven't read the paper, but is it possible that the authors are overlooking that the search isn't perfect? Keyword search allows us to look for instances of words that we already know. When we're looking for a phone number, <i>we're looking for a word that we don't know</i>. Keyword search isn't designed to do that. It forces the user to search for words that probably appear with the target word, but are not guaranteed to occur with it. The example that you gave is outside of the design parameters of keyword search. </p> <p>The problem with phone numbers is that the user is expected to understand what they are through context. When listing a phone numbers on web pages, people usually just put the phone number down. The syntax and context is enough to identify what it is: ###.#### or (###)###-####. It isn't necessary to use semantic waypoints (such as the word "phone") on the page; meaning that a keyword search is almost impossible, unless you already know the phone number. </p> <p>We can get around this by making our search engines aware of the syntax of common strings, and index instances of those strings specially. ###.#### or ###-#### is something that is usually a phone number, so store it in a magic index of phone numbers, and allow the user to search for pages that contain an item of that type, without having to know the exact instance that occurs there.</p> Regarding orienteering: I haven’t read the paper, but is it possible that the authors are overlooking that the search isn’t perfect? Keyword search allows us to look for instances of words that we already know. When we’re looking for a phone number, we’re looking for a word that we don’t know. Keyword search isn’t designed to do that. It forces the user to search for words that probably appear with the target word, but are not guaranteed to occur with it. The example that you gave is outside of the design parameters of keyword search.

The problem with phone numbers is that the user is expected to understand what they are through context. When listing a phone numbers on web pages, people usually just put the phone number down. The syntax and context is enough to identify what it is: ###.#### or (###)###-####. It isn’t necessary to use semantic waypoints (such as the word “phone”) on the page; meaning that a keyword search is almost impossible, unless you already know the phone number.

We can get around this by making our search engines aware of the syntax of common strings, and index instances of those strings specially. ###.#### or ###-#### is something that is usually a phone number, so store it in a magic index of phone numbers, and allow the user to search for pages that contain an item of that type, without having to know the exact instance that occurs there.

]]>