Since its heyday in the 1960-80s, the relevance logic program has lost much of its momentum. The underlying idea has been transposed into various formal systems all of which, however, lack the intuitive transparency that was initially hoped for. Today, relevance logic is not part of the mainstream of the broader discipline of logic, nor has it been widely taken up by computer scientists, mathematicians, or philosophers. Has the program failed, or has it merely stalled? We suspect the latter, proposing a natural way of generating a relevance logic by means of suitably controlled semantic decomposition trees.