Search Quality Observations
Good computer security strikes a balance between security and usability. For example, a fifty character password might be more secure than one with fewer characters, but it would also impede usability. Not having a password could be easy, but it would also be unsecure. Rather than either extreme, a compromise should be made between what is secure and what is practical.
In an internet search, there is a similar dynamic. It involves usability and usefulness. There is an impetus for usability, but there is also a question, will this come at the expense of usefulness? This is an example of where it may:
North County, San Diego, is within the Public Education Zone of the San Onofre Nuclear Generating Station. Because of this, the Yellow Pages there have nuclear emergency information and include a diagram of nuclear evacuation zones. The diagram shows a map with several concentric circles emanating from the station, and each circle represents a different emergency response plan.
According to Wikipedia, 8.5 million people are living in the area and who may also wish to see this information. Some have already sought it out and put it online, such as Danny Sullivan; serendipitous. To find it, or to find similar information from other areas, one might do a Google Image search.
As of September 24, 2011, a Google Image search for the phrase [nuclear evacuation zones] yields no such results. Instead, the search results are thoroughly saturated with images of something else, a current event, the Fukushima Daiichi nuclear disaster. One hundred percent (100%) of the first page results are directly related to this, instead of anything else. It is as if it were the only nuclear event or nuclear evacuation zone ever to have existed. Nevermind that there are 8.5 million people who live within fifty miles of the San Onofre Nuclear Generating Station, who might be looking for different information.
It is at this point that Google may eventually become less useful. Where different search results would help expand and refine searches, they are now being funneled into channels and constrained to topics, automatically; which works well, except for when it doesn’t. Further attempts to refine the previous search continue to show this.
A search for [nuclear evacuation zones -japan], that is, the same search without references to Japan, still yields results from Fukushima, Japan. What’s more, is that the search quality seems to have been degraded. Page two becomes a no man’s land, not worth scrolling past. There are pictures of a dog, iconography, a stranger, and a comic book frame.
Perhaps this search needs to specify a location, like San Diego. A search for [nuclear evacuation zones san diego] yields fifteen pictures from Fukushima and one from San Onofre. Maybe San Onofre will work better as the location. Does it? Yes, it does. A search for [San Onofre nuclear evacuation zones] yields fifteen pictures of San Onofre, several of which are evacuation zones.
The initial search entirely related to the Fukushima Daiichi nuclear disaster, by default.
An unsuccessful attempt to revise the search and omit results related to Fukushima by excluding results about Japan.
The second page of a revised search is still showing results related to Fukushima and other results immaterial to the search.
A revised search almost entirely related to Fukushima yet done by specifying another geographic area.
In this scenario, the results were polarized by a particular topic. It took persistence to shape the search into something useful -beyond the issue; it was constrained too. This is one example of how the impetus to streamline search may be coming at the expense of encumbering some others. There are probably other examples too…
What’s good is that this is all happening faster. What’s terrible is that frustrated users are giving up more quickly. Would it not stand to reason that if Google provides this rich of search experience in 0.09 seconds then in 0.18 seconds, they might offer one twice as rich? No, but people might not mind waiting for better results.