I thought that the Chernobyl reaction stemmed from the graphite moderators breaking, hence a cessation of moderation.
Losing the moderator would have shut down the Chernobyl nuclear reaction. Unfortunately, since it was graphite it wasn't going anywhere until the core was breached.
Chernobyl stemmed from a large positive reactivity addition that was further amplified by a core having a positive void coefficient. Reactivity is the fractional change in neutron population per effective neutron generation. If reactivity is zero, there is no change per generation and since power is proportional to the neutron flux, power is constant. If reactivity is less than zero, power will decrease. If reactivity is greater than zero, power will increase. All of these follow an exponential law, so unless there is a mitigating factor, power will increase or decrease each generation by the same fraction. Additionally, if the reactivity is very high, it tends to decrease the effective neutron generation time. If the reactivity is phenomenally high, a condition called prompt criticality occurs; the neutron generation time is essentially only the time it takes to moderate the neutrons. A prompt critical reactor, like Chernobyl, will destroy itself.
What caused the prompt critical reaction is due to the the positive void coefficient in the reactor and the design of the control rods. A positive void coefficient means that with a separately moderated reactor (like Chernobyl), water acts as a neutron poison instead of a moderator. When neutrons hit the cooling water in this type of core, they effectively disappear from the reactor (a simplified description, obviously). If you create a void (a place where no water exists or water exists as steam), the probability that a neutron will go from one piece of fuel to another is increased. This increases reactivity, and is known as a positive void coefficient. So if your positive void coefficient designed reactor runs out of control and heats up rapidly boiling off water, it is only going to add more reactivity. This is what occurred in Chernobyl.
Second, the control rods in the Chernobyl reactor were designed with riders on the end of the control rods to make sure fueled burned more evenly. These tips have basically no effect on poisoning the reactor (in fact, they were made substantially out of graphite). While the rest of the control rod is a powerful neutron poison, these riders would add reactivity to the core for the first instant if the control rods were withdrawn all the way to the top and then inserted. This occurred.
I'm not going to go into the exact timeline of events or how the operators were stupid enough to place their reactor is such a dangerous configuration, though this page seems as good as any to describe those (note: it fails to describe the significance of the control rod insertion). I will say that at one point in time the increase in reactor power due to a mild level of voids in the reactor made the operators 'scram' the reactor. A reactor scram is where you insert all of the control rods fully to the bottom of the core in order to shut down the reaction. But as I described, the first instant the control rods were inserted they added positive reactivity. This further heated the reactor which began a runaway cycle (due to the positive void coefficient) that blew up the core.
One thing to note about most accounts on the web of the Chernobyl accident: only about 5% of the people have any clue on how to operate a nuclear reactor or the engineering and physics behind it. Many accounts are flat out wrong. Almost every account misses the importance of the reactor scram. Many miss the importance of the positive void coefficient.