Relays?
Lighting the bike, required in all states at all times, draws 5 amps. My original loom used AWG 18 to provide lighting power, which represented over 100 milli-ohms of series resistance. Furthermore, there were three series switches (ignition, headlight/pilot, and headlight dip) in the circuit that totaled another 100 milli-ohms of series resistance. The net result, often cited in this thread, was a bulb that never saw more than 11.5 volts, and thus operated at only 85% of its designed output power. As the switches age (mine are new at the moment), the lighting power decreases even more as the switch contact resistances increase. So the observation that adding relays increases headlight intensity (and horn intensity) is very real.
Jean's point that relays present a potential failure point is true. We are obliged to operate relays in a higher humidity and vibration environment than, say, my Dodge truck. Still, no bikes produced today are built without relays in the lighting and horn circuits. Moreover, the circuits under discussion here are ancillairy to running the bike; whatever the potential failure probability is, both headlight relays would have to fail simultaneously to lose lighting, and even then the bike could still be operated while the rider finds a safe place to stop. Relays fail open, in the excitor circuit, and will not bring the ignition circuit down with them.
Adding a relay in the ignition circuit seems to me to be a particularly bad idea, as this circuit must be isolated and protected from failure. But relays in the lighting and horn circuits, IMHO, bring benefit with very little downside.
