This thread seems to mix two different comparisons. One comparison is between realism and immersion and the other between realism and abstraction (or immersion and abstraction).
Realism and immersion have been compared in this thread enough already, and I have to agree that realism is not an important thing as such for immersion, but what is important is consistency, especially with regard to being consistent with the player's expectations. Realism won't necessarily hurt immersion either. On consistency, I will simply include a quote:
Quote:Oscar Wilde
Man can believe the impossible, but can never believe the improbable.
Magic, supermen, and invaders from Mars are impossible; that won't mean I will not be immersed in a world which has them. The lack of consistency creates an atmosphere of improbability, and that will harm my immersion. For example, in most RTS games the game itself is rather abstract (usually one unit really depicts a whole number of individuals), but the representation is rather concrete (the unit is displayed as a "realistically" animated human figure). For me, this makes the game less immersive, because what I see is not consistent of what I would expect from the system by my observation of the rather concrete representation.
Anyway, the original poster juxtaposed a single health measure to several locational health measures. Neither are particularly realistic, but they have different levels of abstraction, so the comparsion is really about abstraction and realism/immersion. Immersion and abstraction don't need to interfere with each other. Tetris can be immersive and it's abstract. Unreal is much less abstract, but can be immersive too.
As for abstraction and realism, I am more than mildly annoyed by the proposed idea that abstractness and realism are somehow opposite concepts. I claim that
abstraction will not necessarily lead into non-realism.
Imagine that I was to make a game about running about in forests picking berries (MMOBPG, [grin]). Because the number of berries in the bushes the player encounters has a profound effect on the gameplay, I want to make it realistic. What should I do?
Of course, I could start observing forests and plant life in general and spend all my free time studying about plant biology and the growth of berries. Then I would create a simulation of a forest (using genetic algorithms or what have you for the evolution of the plants etc.) and really make that simulation as good as possible. The end result would be realistic, as it is based on the real world and to an outside observer it would behave like the real world. It wouldn't be perfect, however, because any such model would still be a simplification.
Then imagine I wouldn't do the simulation, but would rather use all that time for running about in forests, counting berries in any bushes I encounter and build up a huge database of berry growth in the local forests. I could then use this data to build a statistical model that would be used in the game to populate the bushes with appropriate number of berries. Again, the end result would be realistic, as it is based on the real world, and it would seem to be like the real world. Again, it would not be perfect, as it is a simplification.
The latter model is easier to control, as there will probably be less parameters to set, and it would be considerably faster. It is also a lot more abstract. Still I wouldn't say it was any less realistic than the former model; realistic in the sense that it is both based on the system it tries to model and also seems like that system to an outside observer.
I'm not saying that checkers or any other similar game would be realistic, necessarily. If you think checkers is realistic, feel free to tell me what real world system checkers models and convince me that it models it well. Still, just because not all abstract systems are not realistic it does not follow that no abstract system is realistic.
Even the most concrete games are abstractions to begin with, anyway. You don't (usually) start by modeling atoms and just letting the game world emerge from this.