Problem discussion
X-raying the watermelon
Jan 18, 2024
Mohammed Kheezar Hayat
Product
Why 'search' may not be the best way to do research
The problem with search
Sometimes when I am using Google for research, I can never shake off a mild, annoying anxiety that I might be ‘missing out on something’. The web’s obvious biases notwithstanding, repeatedly typing search queries into a box and opening links seems to me like a terrible way to ‘research’ anything. Firstly there is the arbitrary limit placed on how many links you can open without the tab bar becoming impossible to use. This by itself makes the endeavour sub-optimal.
Besides that, I (like you I am sure) often have to do multiple searches, trying to ‘cover as many bases’ as possible. Part of that is down to the search-engine paradigm being fairly old. The Google search team is always engineering more intelligence behind the search interface, eking out life from a form of user-interaction which is now as fundamental to computing as any, and quite old. This though, in my opinion (and of others), is only part of the problem.
Intuitively, we know that interacting with a knowledgable human being, like the travel agent in this LinkedIn post by adman Rory Sutherland; is more fruitful than searching online. While Rory does not explicitly lay it out, his example illustrates a problem quite familiar to most of us:
Google search (like any search) is naturally a 1-dimensional affair. We have been conditioned to treat the top of the page as more important and the further down you go, the less relevant the results are. From more to less, 1-dimensional. But we intuitively know that expertise (as in the example of the travel agent that RS above talks about) is multi-dimensional. To use a visual analogy, it is more like a 3-d mind palace than a list of items arranged in order. The things we do know well, we know them as a big ‘blob’, not a neatly ordered list. When unfamiliar information is presented to us in this way, we appreciate the neatness but also recognise the limitation. And it is recognition of this conflict that causes the annoyance.
The internet, and computing in general is full of these one-dimensional interfaces. The 1-d ness is down to commercial concerns, or technical limitations, maybe even aesthetic preference.
When we interact with these 1-d interfaces for information that is multidimensional, it is like trying to get a sense of the anatomy and structure of a large watermelon by repeatedly x-raying it
When we interact with these 1-d interfaces for information that is multidimensional, it is like trying to get a sense of the anatomy and structure of a large watermelon by repeatedly x-raying it (or slicing it). See it this way, and then slice differently now, and a bit that way now. This is analogous to what we do when we are ‘researching’ on Google. You would have to do it multiple times as it is hard to see how things are in relation to each other. You may get there in the end, it’s just going to take some trying.
There is a lot of chatter about how the next generation of ‘user-interfaces over AI’ is going to look like, but I suspect they would need to address this problem of 1-d ness. We want to see the whole watermelon now. Slices might not cut it anymore.