8 Responses to Are we asking too much of search?

  1. Ryan Pollack 2011/06/11 at 09:32 #

    I was listening to the webinar also, and I couldnt agree more. I like the idea of a help system offering up unknown unknowns, as they discussed, but there has to be _ some _ input into the system.

    We should be capturing search terms and making sure those terms are included in the relevant topics. But we’re not mind readers 🙂 it’s a fascinating problem though and one I hope people look into more.

    • Mark Baker 2011/06/12 at 22:58 #

      Hi Ryan,
      I agree that we should be paying close attention to search terms. Actually, I think we should go further than that in trying to understand the vocabulary of our users. There are many resources for doing this. Sales people and field staff in your company talk to users and know their vocabulary. And you can find our where your readers hang out on line (or in real life, for that matter) and listen to their conversations.

      • Ryan Pollack 2011/06/13 at 13:03 #

        Yes, I’ve been trying to get my writers away from talking with R&D about terminology discussions and more towards talking with support specialists, sales technicians, etc.

  2. Jonatan Lundin 2011/06/13 at 10:27 #

    I don’t fully agree in blaming users when not finding anything. Wasn’t it D Norman who said that you can never blame the user if a design fails, only the designer.

    When designing something you need to define and understand who you are designing for (the population). The design must be built to accommodate the knowledge level, behaviour etc of the population. Of course, if someone outside of the targeted population is using the design and is failing, it is not the designer to blame.

    Whenever you design something you should be explicit about the purpose of the design, the support it is intended to give and who the target audience is. The user is expecting a certain a level of support. The actual design can furthermore be measured from its actual support. If the actual support is way below the intended support it is not the user to blame, but the designer. My experience is that technical communicators are seldom explicit about the intended support.

    Is your manual built to always allow a certain user type find the answer when searching? If not, what is your intention?

    • Mark Baker 2011/06/13 at 11:09 #

      Hi Jonatan,

      I don’t disagree with anything you say. It was not my intention to blame the user for all search failures, but rather to point out that there is a limit to what we can reasonably expect from the best search engine working on the best content. As you point out, design should be done to meet the needs of a specific audience, and it is no fault of the design if it fails for someone who does not fit the user profile. Lance Armstrong and a five-year-old child may both enjoy their bicycles, but if they switched bikes, neither would be happy trying to ride the other’s machine. Design that is universal in its ambition is doomed to failure. All design must have a specific target.

      Good design is practical, and has specific, limited, achievable objectives. Allowing someone who does not know the vocabulary of a field to find information in that field is not an achievable objective. Still less is it an achievable objective to expect search to return people information they do not even know that they want.

      Cognitive distance is cognitive distance. Before we can understand, we must cross the cognitive distance between our current understanding and the new understanding we need. Search can conquer the physical distance to information, but not the cognitive distance to understanding. To say the crossing cognitive distance is an education problem is not to blame the user, it is simply to recognize that there is no panacea here. We should design content, and the search engine that works on that content to perform as well as possible for its target audience. We should not expect search to obliterate cognitive distance for the entire population of the planet.

  3. Yuriy Guskov 2011/07/12 at 10:42 #

    Actually, the precise search is what Google founders set as own goal. Ironically, they still has not made any advance in that direction. Since Google appearances, all enhancements were either visual or social, which, no doubt, made search better. But it still fails. Fails epically, and sometime funny (when the search returns evidently wrong results).

    The problem is we lack precise identification at the side of an user and at server-side. As well, some relations between items should be established. If not for the whole article, but at least for the subject of the article.

    I’ve described how it may be treated at:
    http://on-meaning.blogspot.com/2011/06/great-blunders-of-modern-it-and-their.html

    • Mark Baker 2011/07/12 at 22:53 #

      Hi Yuriy,

      Thanks for the comment. I haven’t had the opportunity to review all you material yet, but I do think it is unreasonable to say that search fails if it does not immediately provide the full and complete answer we were looking for. The word is, after all, search, not oracle. We have been searching for information (and many other things) for as long as we have been human. It has always been a long, painful, expensive, and uncertain pursuit. Google makes searching for information so much faster and more comprehensive than any previous method of search that it is, by any reasonable measure, an enormous success.

      It is very much looking the gift horse in the mouth to complain that it does not have the oracular omniscience of the Enterprise computer.

      • Yuriy Guskov 2011/07/13 at 02:41 #

        At this moment, Google is number one in search. That’s right. But seems like many one don’t believe there can be something more successful. Even though, Google just fails when you try to search something a little complex, than just one-two word. Try to search something which has complex relations like the situation, when your computer fails with rebooting. Google would return full and complete answer, yes, but you forced to look that answer in million references. It’s ridiculous. It is not full and complete. It’s like you come to a library, and a librarian said: “The book which you are looking for there, in that pile”. Who said we want a oracle? We want just more precise search. And this search is impossible while we are looking through plain text. And even natural language processing won’t help here, because usually information has a lot of implied or ignored information. That’s why I vote for methods which would help us to make information more precise (both in query and results), this way search may be more precise too.