Event-driven programming can make our lives a lot easier. But no sooner than we accept events into the design do we come across a string of new problems that require a thoroughgoing resolution. Specifically, we saw in the code sample 4-2 how C# properties for health and ammo are used to validate and detect for relevant changes and then to raise events (such as OnDead) where appropriate. This works fine in principle, at least when the enemy must be notified about events that happen to itself. However, what if an enemy needed to know about the death of another enemy or needed to know when a specified number of other enemies had been killed? Now, of course, thinking about this specific case, we could go back to the enemy class in the code sample 4-2 and amend it to call an OnDead event not just for the current instance but for all other enemies using functions such as SendMessage, as we've seen in the previous chapters. But this doesn't really solve our problem in the general...