Online rankings have traditionally been grounded in the approaches pioneered a century ago with the totalling of sheet music sales and the home collection of viewer diaries. In this era of connected society we can go beyond these techniques and truly learn the standing of the content we interact with.
The ratings that power our decisions, whether it is deciding which film to rent or track to listen to, are collected without context. A context that establishes what the stars or score actually mean in relation to other items in the same domain.
In this void of context our decisions, rather than use actual content value, increasingly become powered by network effect (social interactions, content views, sales and advertising). With a proven dip in organic reach on social networks the rating problem needs to be tackled head on.Establishing the context of ratings then becomes a concept that is able to:
With Aeolipyle when a choice is made to commit a rating, instead of a prompt ranging from “love this” to “hate this” a comparison is made with a similar previously experienced item. And using a viewing or purchase history the question becomes a simple “which did you prefer?”
Aeolipyle uses a naïve meta-Condorcet election mechanism to assess the pairwise comparisons. And after computation, items are categorised into a simplified banding, ranking or percentile rank. Constantly re-evaluating the standings and individual ratings as additional comparisons are made.
Suddenly ratings have meaning and we can start to ask interesting questions of content:
“Was this Oscar year better than last year’s Oscars?”
“Was this year’s Christmas number 1 preferred to last year’s Christmas number 1?”
“What is the greatest sci-fi of all time?”