Running tally — rB − C across rounds (positive means helping spreads)
What you just did has a name
rB > C is the whole of Hamilton's rule. You applied it
five times across wildly different levels of biological
organization — alarm-calling mammals, cooperating bacteria,
cooperating gametes, and a negative control where relatedness was
zero. The unit of selection changed every round; the rule did not.
Inclusive fitness counts gene copies in the next generation, including copies carried by relatives. A prairie dog that gives an alarm call and dies has zero offspring — but if it saves four full siblings, those siblings carry r = 0.5 × 4 = 2 genome-equivalents, and helping is net positive. The same accounting works for a bacterium that shares iron with clonal neighbors and for a sperm cell that hooks onto its brother-sperm to swim faster. Kinship is kinship.
Round 2 (cousins, rB < C) and round 5 (strangers, r = 0) did not favor helping. These are not exceptions — they are the test. If the rule "worked everywhere" it would not be a rule, it would be an assumption. The point of Hamilton's rule is the boundary: when does the inequality flip?