In order to gain some familiarity with the Webgrid system, I decided to arbitrarily choose some topic to elicit. I wanted a topic that was general enough to have many elements and constructs, and that I am sufficiently knowledgeable of. After some deep soul searching, I settled on food. It's a topic that I'm faced with every day. Some food brings pleasure and satisfaction, while others bring great distress and merciless gut rot. So I thought back to the past, all the way to my childhood days, to recollect which foods I have noshed on in my life. I easily jotted down several dishes and beverages which I have enjoyed and been victim of and entered them into the system.
At this point, the Webgrid handed me a trio of elements, and asked me for something that made one of the elements distinct from the other two. Having avidly followed Sesame Street as a youngster, this was like second nature for me. After repeating this separating process several times, I was faced with some conflicts in my constructs. So all that I had to do was enter elements which were one of the constructs and not the other. It was that easy. However, I noticed that sometimes the system kept demanding new constructs and elements to make their union more unique. I could see how this might create cyclic problems when eliciting constructs for elements that were inherently quite similar in most aspects (my project topic, for example!)
In this case, the above process was essentially repeated using a pre-defined list of elements (the topics studied in class.) The first step was to develop the constructs based upon the topics using the triad method. This worked well in several cases, for example whether the topic was an application or an underlying technology. However, there were many instances where this method was sorely lacking. The particular circumstance I am thinking of involves the case where two items are fairly similar if described solely by adjectives, but have several different attributes. Specifically, the internet and broadband networks are quite similar in many aspects, but are obviously not the same thing. In other cases, some comparisons would simply be non-applicable to the construct chosen.
Another problem encountered was the situation when two constructs were judged to be similar. Since we are dealing with a fixed-size collection of elements, there was not really any way to differentiate the constructs by just adding elements.
Upon comparing the grid with another (in my case, the example by Dr. Gaines,) it was clear that many of the constructs were similar. However, some of the constructs present in his version were absent in mine, and the opposite was true as well.
My third experience with Webgrid involved an elicitation on my presentation topic, Virtual Reality. The elements I used were simple the section headers from the report. Since all of the elements were virtually identical except for the area they are used in, the value of Webgrid was questionable. The only constructs that really showed any difference in the graphs were the implemented-not implemented areas. I also had quite a bit of trouble in thinking up things that could distinguish, for example, educational applications from manufacturing applications. These applications would be the same except for where they are employed. I think that webgrid is more suited to areas where the topics are more discrete.