Moral questions can be tricky if the concept of morality is not adequately explicated.
There are many schools of thought on this question: deontology, which emphasizes the importance of moral duties at the expense of the good or bad results; utilitarianism, which requires a prudent measuring of the harm or lack of harm created by certain actions or certain rules of action. Divine command theory - the content of moral rules come from religion and we are duty bound to follow them (hence divine command theory is akin to deontology rather than consequentialism). Virtue theory, stemming from Aristotle. And so on. To be curt, the meta-ethics must be complete else the ethics is vague (
Meta-ethics - Wikipedia, the free encyclopedia).
I tend to think that in our practical life our actual actions of applied ethics are informed by a mixture of these different theories, though the components are often not analyzed out.
For example, most would claim that we have a (deontological) duty to assist those in need; but as many reported in previous posts, any rescue must not produce more victims (this is a consequentialist condition which over-rides the duty to assist).
The deontological duty to assist those in need - I doubt many people would deny that this exists; and we see many examples of this in areas of life outside of diving where there is not a formal coupling of the duty of mutual assistance as is the case with the dive buddy. Of course, people are often inconsistent applying this principle. If you are interested about your own consistency you can try the 'drowning child problem':
The Drowning Child
Measuring up the utility of action is difficult for it is often hard to predict risk or to measure harm; so, making a prediction of what one would do from a utilitarian orientation is difficult for under this paradigm the unique facts matter in so far as the unique facts will influence the calculation of utility.
But, if it is true that our applied ethics is a mixture of theories having mutually exclusive conditions (and deontology and utilitarianism are mutually exclusive in so far as deontology ignores utility calculations - this difference pops up in policy considerations regarding drug and prostitution when 'harm reduction' strategies are suggested instead of the absolute prohibitions from the 'law and order' crowd) then decisions in the abstract will have little meaning.
That said, I would make it clear to any buddy that I myself use a mixture of moral reasoning and that my duty to assist is limited by risky consequences; risky consequences created by wilful neglect of the dive plan on the part of the buddy are not going to be insured, so to speak, by me. I would make this clear at the beginning or pre-dive. Indeed, the buddy, in so far as he willingly places himself in harm's way willingly pollutes or negates his obligation to render assistance to me should I have an accident; maintaining that reciprocity of duty to assist is, I think, a necessary condition for maintaining a duty to assist on my part, provided the consequences of assisting are not unduly harmful.
MT