Why do we always need to integrate search with structure?

Question

Jakob Nielsen posted in May 2009 an article called “Top 10 Information Architecture Mistakes” where he states that mistake no 2 is not having an integrated search with the structure of the site. I’ve seen many sites that don’t have this integration at all – such as Wikipedia where the articles don’t have relations (through hierarchical navigation). I think it is perfectly OK not to integrate every article on Wikipedia in a monstrous hierarchy which leads to navigational elements. There is a hierarchy on Wikipedia, but not prominent enough to use as navigation.

But there are other sites where this make sense, such as the documentation of Microsoft .NET platform. The navigation is hierarchy dependent in my mind and if there is no hierarchy there can be no integration of search into a structure. In these cases I support Jakob Nielsen’s statement:

Sadly, search and navigation fail to support each other on many sites. This problem is exacerbated by another common mistake: navigation designs that don't indicate the user's current location. That is, after users click a search result, they can't determine where they are in the site — as when you're searching for pants and click on a pair, but then have no way to see more pants.

Still, that doesn’t mean that Wikipedia made it wrong. I must be missing something here, please let me know what?!

Answer 1

Search and navigation covers two different goals, but should sometimes (often) be combined for interlaced and varying goals.

As with the examples mentioned, Wikipedia and a clothes store. If a Wikipedia visitor searches for Eg "Finland", then that user is interested in something regarding Finland. The user is probably not very interested in the hierarchical path that took them there, even if one could be formed, Eg: Human notions -> World -> Countries -> Europe -> Scandinavia -> Finland.

The user is requesting something specific and unique, in this case a country. From there it's very hard to predict what the user could be interested in next in addition to the problematic in forming a taxonomy amongst totally unique items.

Other types of sites however that has content that can be easily categorized, such as a clothing site, should provide a breadcrumb control adjacent to the product a user is viewing. Simply because the user is quite obviously interested in that type of clothing and is probably keen on seeing what alternatives are in stock.

I believe what Nielsen was talking about were sites like Home Depot. Navigating in that site can be a handful, plus if you search for an item (Eg. power outlet) and enters a product page the breadcrumb will say: Home -> Search Results for "power outlet", with no indication of how to find more products in the same category.

This is a problem for a user trying to find alternative similar items, because searching for "power outlet" on Home Depot there's bound to be some, searching for "Finland" on Wikipedia there's not.

Answer 2

The structure you are referring to is a tree.

When the content can be structured more or less [1] like a tree, it makes sense showing where you are in the tree and thus enabling the users the browse through nearby locations, or travel up the tree.

When the content is relatively flat, showing the location makes less sense, however, showing related and similar topics does. (Which in its self, acts as a sort of an ad hoc structure.)

Youtube for instance, shows similar videos (at the right) and lets you navigate into the list of the videos uploaded by the same person (below video).

Wikipedia for instance, show the "structure" to the top right (in LTR format), related topics throughout the article and all references at the end.

[1] Content is often structured as a graph, yet it can be viewed as a tree, where the same item appears more than once. In these cases, navigation should take into account the multiple locations of the current item.

Answer 3

I think Wikipedia is lacking in structure (structure in the IA sense). There is a categorical hierarchical structure but it's separate from the main body of content and somewhat clumsy.

Take the target subject of "Yosemite Park". A search for "Yosemite" takes me right there. I can also get there by navigating (clicking links) by following this path: Wikipedia -> Contents -> Categories -> Geography and places -> Protected areas -> National parks -> National parks of the United States -> United States National Parks by state -> National parks in California -> Yosemite National Park.

But the Yosemite page itself has no indication of the context suggested by that navigation trek, no indication of its place in the categorical structure. I think including that context in the page would be a definite improvement. But including that context throughout Wikipedia would require tremendous resources and is just not feasible, especially given the fluid nature of that content. Crowdsourcing has organizational limitations. It would take a structured people organization to organize a large information structure - my guess is that the category section, which is organized but separate from main content, is made by a more stable (permanent) organized group.

So in a sense I would say Wikipedia has it "wrong", but it's not feasible to get it "right" (to maintain structure in a massive fluid body of info), and as it is it's still very useful. It is what it is.