Where is the de facto Swahili search engine? (Part 1)

The three most popular search engines in
sub-Saharan Africa, according to Alexa

Search engines help you navigate the world wide web (the Web). They help you find things, they help you know what others think is important, and, most importantly, they help the builder set the agenda. All of these three functions are connected. For instance, you usually cannot find information that others consider unimportant. Also, since these are the first points of entry for most browsers, the one who controls them decides what most browsers should see, etc. Hence, it is important who builds search engines and what their agenda is.

The emergence of Google reduced the barrier to entry for most content providers who were not necessarily interested in selling a product or promoting their material. Before then, search engines were just portals that listed websites that marketers fancied, so the entrance of one that considered the structure of networks of web pages to rank them meant the discounting of the opinions of deep-pocketed players--a welcome intervention in the opinions of most browsers. However, that was just the beginning of what a search engine could become.

A study the structure of the Web requires an analysis of the connections and what they mean to both content creators and consumers. This enterprise requires data and algorithms to handle the scale of information that is generated. But it also requires the discretion of the developer. Firstly, there is the discretion to discriminate against which pages to include in the map of the web that one uses for analysis. Then, there is the discretion to discriminate between different meanings of a query--a task that can sometimes be accomplished with the help of natural language processing algorithms. These two liberties that the developer is free to exercise lead to differences between search engines that would otherwise be, for the most part, algorithmically identical. For a long time, some did not appreciate the significance of this point, arguing the impartiality of search algorithms and the blandness of data, but just a cursory examination of some search queries in a language that one understands, or a search for a terminology whose context one could reasonably assume is unfamiliar to the developer of a search engine would reveal how the developer chose to exercise these liberties.

I will continue in another post bearing with our continually diminishing collective attention span.