As architects we usually work with deterministic business cases that follow that you will get the same output with identical input. All such deterministic business cases are the low hanging or simple projects for the first 70 years in computing. However we now slowly move into a stochastic phase as we all know that business is not deterministic in many cases. Additionally we also use computing in areas that were seen too complex or non deterministic in the past.
Stochastic or hybrids are currently the majority of the non deterministic business cases and designs, as there are potentially also other business cases in development. Stochastic business designs are currently main found in finance and science, but are also already starting in other industries, as just controlling deterministic events is often not enough to stay on top of the market without needing manipulative techniques.
Most architects however are completely blown out of their depth when confronted with stochastic solutions. An example is when designing a load balancer for a Value at Risk and risk distribution that uses a eGARCH also with any copula to compute the risk stress (I deliberately described it more simple) where most architects use a traditional round robin deterministic principle instead of looking on a Markov clustering (just one of many solutions) so that nodes concentrate on similar behaving groups of securities. We as architects also need to look at all our tools to see if they are usable for stochastic based designs. As such while I have worked in this area I only want to point out that you seek advice before entering in this field as I do not want to loose all my readers that are unlikely to ever enter this field.