Actually, historical evidence suggests the opposite of your argument: multiple implementations of a language typically mean it is succeeding, not failing.