S2 exists because many cameras/flashes emit one or more preflashes before the actual exposure flash. A simple optical slave in S1 would fire too early—on that first burst—and may contribute little or nothing to the exposure.

S2 simply tells the slave to ignore the first flash it sees and fire on the next one, so the slave fires at the correct time.

You are correct that this does not create true TTL coordination. The camera usually does not know about the optically triggered slave, so that slave’s output is not directly included in TTL metering. In practice, S2 is mainly useful when:

• the on-camera flash always uses a preflash,
• red-eye reduction or flash-based AF assist causes extra flashes,
• the slave is being used in a way that doesn’t strongly affect subject exposure, such as lighting a background.

If you want predictable results, use manual flash power/exposure and adjust for the added light yourself. S1 is for single-flash/manual situations with no preflash; S2 is for setups where a preflash would otherwise trigger the slave too soon.