The jury deliberation process is effectively represented as a Markov Chain process. Deliberation starts with 0 to n jurors in favor of a guilty verdict, where n is the number of jurors. Jurors then deliberate until they reach an absorbing state, either 0 (not guilty verdict) or n (guilty verdict).
You can simulate the deliberation process for yourself. To start a simulated deliberation, select the starting node. Notice the arrows going to and from the nodes. From any nodes, there is a probability of gaining a guilty vote, losing a guilty vote, or staying on the same node.
The outcome of a simulated deliberation is random. If you start deliberation on node 7, for example, the deliberation could end with a not guilty verdict, or a guilty verdict. The outcomes are random, but over the long run, we can determine the probability that deliberation will yield a guilty verdict based on where the deliberation starts.