Design/Archive/Wikimedia User Interface/Analysis/Navbox

Observations[edit]

Rows in navboxes are a series of 'relation data'. Much as wikidata has "parent of" "followed by", "owned by", "produced by", "part of" etc relations between subjects/topics.
The navbox template is a 'view' on top of this relation data that allows a select set of readers and a large set of curators/editors to easily traverse this relation data (and often that of it's siblings/parents/children at the same time).
Many of such relations are also grouped as categories on Wiki's.
A navbox groups these sets of relations into a higher level topic.
A stack of navboxes represents multiple higher level 'first degree' topics of the article in question
navboxes are bi-directional. Every article linked from inside a navbox links back to the article on which you are using the navbox
navboxes are a structured interface for related articles
navboxes are both content (summary of relations) and interface elements
they encourage exploration
they prevent noise (too many links to only tangibly related articles) in the article body
a series can only be part of a navbox if all articles within the series (and all it's cousin links?) are 'notable'. Branches, not leaves makes navboxes ?
It's an interface to navigate a graph

On English wikipedia, when you have more than 3 collapsible elements, then all those elements are automatically collapsed. This means that many navboxes are almost always collapsed.
This avoids too much 'visual clutter'.
The auto-collapse code is the side effect of 'lazy authoring'. Basically we initially had collapsed and uncollapsed, then we made navboxes, and then we became too lazy to specify 'collapsed' or 'uncollapsed' per page, so we invented auto-collapse (if more than '...' it's more likely we want something initially collapsed, then initially uncollapsed). But there are no semantics involved here. If you add three collapsible elements to the infobox, then one single navbox at the bottom will also be collapsed.