There is no single price of oil. There are hundreds of crude streams, each with its own quality, its own location and its own buyers, and each worth a different amount. What holds the system together is a small number of benchmarks: grades liquid enough and transparent enough that everything else can be priced against them.
A benchmark needs three properties, and they pull against each other. It needs enough physical production to be hard to corner. It needs a transparent price discovery mechanism, whether an exchange settlement or a published assessment. And it needs to be somewhere the market actually wants barrels. Grades lose benchmark status when production declines far enough that the first property fails, which is the slow structural problem underneath the North Sea complex.
Three benchmarks carry most of the world. WTI prices the American mid-continent and, through it, much of the Atlantic basin. Brent prices waterborne crude for Europe, Africa and the Atlantic. Dubai and Oman price Middle Eastern crude flowing east, which is the largest single trade route in the market. Everything else is a differential to one of the three.
The mechanism that converts a benchmark into a cargo price is simple to state and does a great deal of work. A seller and a buyer agree the differential, not the price. The price is then whatever the benchmark averages over an agreed window around loading. Both sides have taken a view on relative quality and location, and neither has had to forecast the level of oil.
Two of the three benchmarks are not exchange settlements at all. Dated Brent and Dubai are assessments produced by price reporting agencies from the deals and bids they observe. That is a different kind of number from a settlement, with a different error profile, and knowing which kind you are holding matters when you compose them.