I think that the question itself, and none of the answers provided, is wrong.

Today, it is more than the eye-candy that makes a game immersive, interesting and attractive. There is more and more focus placed on other areas to deliver the target for any good story, game, or movie.

I consider the suspension of disbelief to be the most imporant factor.

Graphics are a portion of the disbelief equation, and often the first item that gets noticed, but they are not the complete answer. Physics is being adopted in many shapes and forms, from collision detection and reactive materials, to ragdoll and blended animations. As techniques in these areas are refined and improved, so will their ability to immerse the player into the plot and aid in suspending belief. Storylines themsleves are important, and the most often poorly planned and executed portion of a game. Sure, it's fun to blast away at blocky Doom models, and to WoW at the latest ultar-uber-shaded water droplet animation, but that is a distraction from why we paid to come to the circus. Innovative UI's are another area that impacts the suspension of disbelief. If it's in the way, feels un-natural, or just detracts from the feeling of "being there", it can change an otherwise great experience into disappointment.

Character interaction is also pivotal. How many times are you going to visit an NPC player to chat if all he can say is "Wha?" "Job's Done." "Okay." Now if that same NPC could actually conjugate a verb, pick out keywords, reply with useful information, now that is interesting. Wonder what he will say next time when you mention the cloak of hiding that you noticed over in house #2...

Is photo-realism the right direction for games? Yes, for some. Is planning, testing and implementation towards suspension of disbelief the right direction? Almost always.

Just my 2¢.
You are now returned to your regular chaos and mayhem. Proceed.
Mark