I think this excerpt sums it up nicely (the emphasis is my own), but I urge you to read the full article when you have time.
Storysensing is not storytelling. In a dramatic arc, the structure, pace and timing matter a great deal for delivering impact, but storysensing is better when focused on enhancing the portrait of the game world. Unlike storytelling, storysensing does not need to be dramatic. It can afford to be loose around the edges as long as those edges are not too apparent to the point that the player is seeing the frame.
Left 4 Dead creates a zombie-infested world and places guns in the player’s hand. Through a combination of co-operative dependent loops, and a game dynamic that deliberately paces out the encounters and objects, the game conveys a world in motion. Add a layer of snappy character dialogue and numina such as posters on walls and other touches, and Left 4 Dead draws the player into its world so completely that he wants to play it again and again.
Storysensing is best when deft rather than deep. Roleplaying and adventure games have tried for decades to use mechanisms like branching dialogue to storytell to the player, but in practise these mechanisms are heavy handed. Mass Effect in particular is an example of a game whose rush to storytell is so replete with redundant detail and branched dialogue that it just becomes tedious. It actively works against the sense of story because it reminds players too often that they are watching a mechanical game system simply go through motions.Source: Video Game Writing and the Sense of Story
Incidentally, the idea of 'painting a portrait' rather than 'telling a story' is a technique I have been experimenting with for Luminesca.