Like this article? We recommend
Like this article? We recommend
Like this article? We recommend
What
What is mostly an ontological Question because it asks, "what is there?". The answers tend to be concrete things. However, an answer might also be more abstract - a role that a person or thing plays in relation to the system or even rules and constraints about the system. Like Who Questions, these Questions tend to generate classes, actors, subsystems, objects, and components - the static elements of the UML model - but they may also generate dynamic models that describe the behaviour of those things.
What Questions can also refer to processes through the mechanism of nominalization. In Generative Analysis, we consider all communication to be subject to the forces of distortion, deletion, and generalization, and have a meta-language called M++ that allows us to identify and neutralize these forces to recover the distorted, deleted, and generalized information easily and effectively. Nominalization is a distortion whereby processes are spoken about as though they are things. For example, we might talk about "my life" as though "life" is a thing we possess - but it isn't, it is a process that we participate in.
It is important to identify nominalizations in answers to What Questions so that they can be tracked back to the processes they may be obscuring. In Generative Analysis, you identify nominalizations by representational tracking, as follows: Imagine the nominalization in sensory terms (try to form a mental movie of it) and observe if it is a thing or a process. On tracking back to processes, nominalizations resolve into a How/What cascade - How the process proceeds step by step, and What happens at each step. It is interesting that to elucidate How a process proceeds step by step, we need to fall back on a nominalization by asking "What happens next?" at each step to find the individual activities that comprise it. What and How play together to generate the answers we need.
7. How
How is a process Question. To answer a How question, we need to understand What happens. Because "What happens" generally happens to things, this generates ontological Who and What questions, that aim to determine the things that execute the process. For example, when interviewing a stakeholder about their job, the sequence of questions is usually "What do you do" to find the name of the job then "How do you do that". If the stakeholder can't answer that question in a clear way, then take a step right back to the beginning of the job:
- "How do you know when to start your job?" - Make sure the answer tracks back to reality (representational tracking), so look for a physical, sensory, based answer such as "I sit down at my desk and log into X" or "I go to the warehouse". Always look for the very first thing that indicates that the job is starting. Oddly enough, some people find it difficult to know this, because they just don't think about it. However, on prompting they can usually give you an answer. If you can't get anywhere, you can always ask to observe them in action.
- "What is the next thing you do?"
- "How, specifically, do you do that?"
- "What is the next thing you do?"
- "How, specifically, do you do that?"
- ...
As we pointed out above, a How question often resolves into a cascade of What/How questions. The answers to these questions may be atomic actions, events, or complex processes. The aim is to come up with a sequence of steps that can be transformed into a use case diagram, activity diagram, sequence diagram communication diagram - part of the dynamic model. We often give the sequence of steps to a Generative AI such as Microsoft Copilot and ask it to generate the Plant UML code for the diagram we want. This can be a great time saver.
As an aside, the use of the word "specifically" in the numbered list above is a characteristic piece of Generative Analysis style. Asking for a specific answer will almost always get you a much more detailed response and save you a lot of time with follow-up Questions. This also works with Generative AIs! The main difference is that whilst you can only use the word "specifically" a few times in a conversation with a human without appearing odd, you can use it as many times as you like with an AI.