The Problem of Search and the Aesthete

In the process of setting up this blog, I spent a considerable amount of time searching for a suitable theme. Initially I toyed with the idea of building my own. I calculated that my time would be better spend on other projects, however, especially with the the panoply of high quality themes already available.

Immediately I found myself awash in hundreds of potential templates. I spent weeks paging through thousands of themes. The vastness of offerings led me to believe the perfect theme might lie somewhere just beyond the “next page” button. The task quickly consumed far more time than crafting a custom theme would have. When I finally picked a theme, extensive modifications and fixes were still required. On the whole, my experience of searching for a theme was miserable.

I’ve had similar experiences in the past searching for fonts as well. Perhaps I shouldn’t have been so selective. I recently helped a friend setup a blog. He happily picked a theme by his second page of browsing. I feel strongly about the power of visual particulars, though.

Theme Searching

In theory, this problem should fall within the domain of search. Yet no search or filter helped me. I didn’t have any tangible requirements, such as the color or number of columns. While some offered more functionality or plugin support than others, I knew I could add anything I needed myself.

That isn’t to say I did not know what I wanted, though. While my requirements were conceptually abstract, they were highly specific. I knew the exact style of aesthetic I wanted, the questions of color, form and functionality aside.

Search Beyond Linear Text

This problem I encountered appears to be growing. Search in general is getting better. Most of my queries yield immediate answers. The majority of searches are becoming more efficient, but not all. Tragically, the areas of the web growing fastest fall beyond the reach of Google et al. User generated artistic content (such as themes, fonts, photos, texture), along with content generated with the context of a social graph is all understood differently from normal long text.

What does a search engine do with the comment “great idea!” unless it can understand that comments relationship to other entities within that context - a context which may span blogs, Twitter and Facebook. How does a search engine even know what I want when I look for a theme? There are no clear proxy variables.

Generating ad Hoc Proxy Variables

Recommendation engines may be able to solve part of the second problem. If a user likes the style of another person, he or she could weight recommendations based on what friends preferred. Just because I am looking for a professional, staid, open business theme one time doesn’t mean I will be looking for that in the future, however. Search engines will have to begin generating their own data from the start of every query instead.

First, present the user with panels of designs. An individual could begin to indicate the style he or she prefers. Such as system would have to be nonlinear, though. Frequently there are a few different sub-styles that appeal to me within a larger search. These tracks would need to be maintained separately, as they are usually immiscible, if only subtly distinct. Results would need to be present alone side one another. People aren’t going to want to do a bunch of different searches, especially if each one requires a bunch of input.

Of course the inner workings of such a recommendation engine are an entirely separate issue. Some promising advances have been made in image search, however, which makes me think this is possible in the near future, if not presently.

| Popularity: 64%