It is often said (especially in Narrativist circles) that death and failure are not exciting, and should not be regarded as meaningful options or outcomes. In my opinion this is an understandable position, but lacking in nuance.
As I've often written, categories are not real, and there are very few judgments which can rightfully be regarded as categorical. My tendency is to dissolve classifications, and look within them to make local determinations based on specific circumstances and the differences between them. So it is with death and failure.
In some games death should be a real risk; in a Simulationist model it is part of what you're simulating (i.e., a "real world" with all its unpredictable and uncontrollable aspects). But in games with a more Narrativist bent it is not death but rather failure that presents the most interesting challenges. Like a season-finale cliffhanger, failure to achieve an important goal represents not an end, but an escalation of problems to carry into future sessions.
In other words, while a Narrativist game might choose to eschew death, failure is different. Failure is interesting; often so interesting that it opens up whole new levels of conflict, tension, understanding and gameplay.