Rethinking search engines to restore a fragmented world
Pointing towards a new way to navigate the internet
Our planet and society suffer as a result of inner and outer fragmentation. We live on the same planet - yet destroy it because our awareness and efforts are too fragmented. Our outer fragmentation mirrors our inner collective fragmentation. The internet is a technological mirror of this inner collective consciousness, into which we tap almost primarily through the use of: the search engine.
We are so used to querying the internet by typing into a text field. The very concept of a search engine is completely taken for granted. We think we have the knowledge of the world at our fingertips not seeing that today’s “lookup”-engines are actually one big crutch - a magic teleportation tool without which we would utterly fail to get around. This is because the landscape of the internet is very fragmented. Hyperlinks do provide paths from place to place - yet are not suited as a means of introducing any kind of higher coherence and order. The resulting landscape is not well suited for meaningful traversal - which is the reason we turn to the search engine.
I think future search engines should be concerned with the landscape at large. What I am hinting at here is a new feature / operating mode that I would call “journey-mode”. It’s supposed to be a new way to browse the internet - a new way search engines could expose the user to the vast internet in an unprecedented way.
Search with extra steps
In terms of search engine interaction such a journey-mode would mean moving from a one-step lookup to a multi-step search-process where appropriate.
Why would one add “extra steps” though?
There are about 2 billion websites. A single search-phrase yielding a handful of immediate results sounds very reductionistic to me. It implies a lot of choices being made for the user. I’d rather have more agency for a few “extra steps” when desired.
Such a way to navigate would be desirable for queries and questions of open ended nature, such that are based on opinion, taste, mood or other “context“. The current internet is not the best place to ask such questions. Their exploration is not encouraged and the spectrum of answers is not revealed by current search engine design. You won’t get far from where your search engine dropped you off.
I imagine the hypothetical search engine to not immediately yield “search results“ but instead present the user with “paths“ to follow along. The journey-mode would be treating the user as a kind of traveler where the 2 billion websites are the territory and the search engine is facilitating a journey into that territory. Imagine navigating a network of associations, relations - zooming in and out - traversing a kind of virtual landscape.
A new landscape
The landscape can be thought of as a graph structure even though simply thinking about vertices/nodes and edges/links would be an oversimplification and limit the imagination.
Currently there is no such nicely traversable landscape! The only landscape that we have available are the billions of website-islands. The website-islands of today are like cells floating in a primordial soup waiting to be transcended and included into an encompassing structure.
The creation of such an encompassing structure could start with a very small landscape that is coherent and into which more and more fragments can be weaved in.
In order to build “the landscape“ I will describe 2 kinds of structures that I think are required to make a journey-mode actually attractive: One structure for “horizontal” navigation, the other one for vertical navigation.
A fabric of traits (horizontal navigation)
How to interconnect webpages if not merely by the use of hyperlinks? As a user I would want to travel along meaningful relations. Any kind of similarity could serve as a relational mesh. I would approach such a relational fabric generally in terms of traits. Common trains would pull webpage-islands closer together. Cluster analysis would be one way to approach this technically. A practical example for this would be the google bird sounds experiment where various bird sounds are clustered on a plane according to their similarity. As an example for the “journey-mode-landscape” think of the following: Two website-islands both blogs, one politically left the other one politically right, could be in the same “city“ as they both talk about politics. “about politics“ would be one shared trait yet “left-oriented“ and “right-oriented“ would be 2 other traits putting the 2 sites into different ”neighborhoods”.
The traits could be considered the DNA of a webpage from the perspective of the journey-mode. Each bit ( 0 or 1 ) of the “DNA-sequence” would be standing for the presence or absence of a trait. Webpages defining their traits by the means of the same DNA-sequence could be considered to be in the same gene-pool. Within a gene-pool all website-islands are automatically interconnected. Through their traits they share a horizontal relatedness which is the requirement for horizontal navigability.
The data-structures described here as traits and sequences of those traits really define the nature of the horizontal landscape. Choosing individual traits, and sequences and fingerprinting webpages with these sequences calls for a collaborative community effort. Different groups of users would desire and design different traits and sequences.
One kind of trait that I find especially interesting and that current search engines particularly struggle with are the subjective ones. The TED talk recommendations for example are based on a few subjective traits like for example: hope, inspiration, “issues that matter”, “courageous“. Subjective landscapes done right could dampen political polarization by allowing sides to more easily reach each other. The goal for such a landscape should be to allow and contain contradiction in a constructive way.
Horizontal structures are essential yet they are not the whole picture. The landscape laid out so far has a bazaar-like vibe to it. It is like the entire landscape is one suburban landscape. Getting around in journey-mode is technically possible but would take a lot of steps and the traveler might still be confused due to a lack of big picture orientation.
Holarchical structures (vertical navigation)
If you use google maps and you are in New York but want to go to Berlin you will not click your way through street view. You will zoom out to space - then zoom in to Europa, to Germany, to Berlin. You would use vertical navigation. The landscape we are talking about can of course not be displayed in 2d or 3d without dimensionality reduction. The described horizontal navigation implies the existence of a plane and even though accurate visualization of such a plane is not possible we can use it as a mental model. Reusing the previous example of the politically left and the politically right blog vertical movement would be an abstracting and generalizing one - zooming from the left blog “up“ to “Politics” and then “in” to “american politics” to “to outspoken politically right“ to “individual voices“ to “blogs” to “critical of subject X“ where we might find the other blog. This example illustrates that the vertical movement in our landscape will have a subjective flavor to it. Up can mean going into politics - but why not move into philosophy or into categories of popularity instead?This is not a contradiction - all three examples are valid vertical movements. “Above” any plane we would therefore add an arbitrary amount of vertical places. Wikipedia has such vertically nested places - they call them portals. Subreddits can serve as examples of subjective vertical planes too.
Vertical planes can take many concrete shapes as the given examples illustrate. For a journey-mode to take off verticality is essential. I believe the guide to build such a structure has been written down already by integral philosopher Ken Wilber in his 20 tenants of the holarcy. The holacy that is our landscape would look like this: Webpages-holons are embedded into “horizontal-planes“ which are transcended by higher vertical planes/places which are transcended by even higher vertical planes/places. When approaching an actual implementation of such a holarical landscape the implementation of the 2. tenant of self-preservation, self-adaptation, self-transcendence and self-dissolution could be a starting point.
Scalability
The described landscape can also be understood as a kind of curatory effort. Everything that is webbed into the landscape becomes more available. I am not sure how far such a landscape can, in practice, be scaled up while still being highly coherent.
If the answer would be - “not much further that maybe a few tenth of thousands of pages” or “not spanning more that a certain amount of active users or topics” then it would in my eyes still be a valid and meaningful venture as a curatory effort with a limited scope, for example on a organisational scale or the scope of a specific topic.
If the described landscape however could be scaled up without a loss of coherence then this would have much larger implications. Then it could change the way we search and navigate the internet.
Let’s do this
The ideas presented here arose in part through my own software-prototyping. I believe to have gathered a few pieces but also realized that this is too big a fish to bring ashore alone. I have written this post to reach out. I want to find collaborators, discourse, my people. Please reach out and recommend this.
What’s next
How can internet technology serve us in our evolution and not distract us from what is essential? I want to explore this question further in future posts.
source of the post preview image: @sloppyperfectionist