A streaming service with tens of thousands of titles presents a few dozen on its home screen. This is deliberate, and it follows from what happens when people are shown everything.
Choice has a cost that rises with volume
Deciding between a handful of options is quick. Deciding between thousands is work, and the effort is spent before any enjoyment begins.
Beyond a certain point people stop choosing altogether and leave, or default to something already familiar. Both outcomes are worse for the service than a mediocre recommendation.
Interfaces are therefore built to close a decision fast rather than to represent the catalogue accurately. A viewer who starts something within a minute is a success by the only measure the interface has.
The home screen is a ranking, not an index
Rows are assembled per viewer from watch history, similarity between titles, and what the service is currently promoting. Two people see substantially different screens.
Order within a row matters enormously, because attention falls away sharply from left to right. The first few positions carry most of the plays.
This makes placement a scarce commodity that the service allocates according to its own priorities, which are not identical to a viewer's.
Owned titles are worth more than licensed ones
A licensed show costs money each period and may leave. An owned show costs nothing further and stays permanently.
Promoting owned material therefore builds habit around content the service controls, which is why originals occupy so much of the visible surface.
The licensed catalogue still exists and is still watched, largely by people who go looking for it through search. Those titles earn their keep without ever appearing on the front of the service, which suits an accountant and frustrates a viewer.
Search is a separate and neglected system
Because most viewing comes through the home screen, search receives less investment. It is common for a service's search to handle titles well and everything else poorly.
Genre browsing is similarly shallow, often a short menu rather than a real taxonomy. Deep catalogue navigation is not a priority because it is not where the plays are.
Viewers experience this as a service that seems to contain very little, while the actual library is enormous and mostly invisible.
The catalogue's size is a marketing figure
Title counts appear in announcements and comparisons, but they describe an inventory rather than an offer.
What a viewer can realistically encounter is bounded by the interface, and the interface is designed to show a few hundred things repeatedly.
Judging services by catalogue size therefore measures something that neither the viewer nor the service actually uses.