On Data Validation

[holland]

As we design something, we usually want to have some restriction on the options available to use for each property we're trying to assign a value to. For example, in designing an adult human character, we might restrict the "height" property between the known tallest and shortest adults. These restrictions might be for sanity, realism, practical limitations, to encourage creativity through limitation, mere stylistic choices, etc. But for whatever reason, there is some palette of options we're working with.

When we choose to apply these restrictions -- i.e. to validate that data -- varies from designer to designer. If you work with a form, program, or profile sheet, for example, you might have restricted options available to you to begin with. On the other hand, if you work from pen & paper brainstorms and braindumps, you might only assess the validity of your design choices when you start organizing afterwards.

Data that fails to validate might be shifted to the nearest valid option, or you might be taken back to the drawing board to rework the idea.

The data validation itself can take many forms depending on the specific property. Are we trying to keep it in a certain range? Make sure it has at least some kind of value chosen before we move on? Ensure it's a type compatible with the tools we're using? (For example, you might try to give a character multiple hair colors in different places during a brainstorm, but the lightweight modeling program you plan on using turns out to only support one hex value for the color.)

Data validation can also spread across and account for multiple fields. For example, checking that there is unity (in principle) across the project, or that newly selected properties aren't conflicting with the established unity, is itself a form of data validation. A lot of the tools we put in our toolbelt as designers are just additional layers of checks like this.

So data validation is an important part of the decision making process, especially when we are designing with intention. When we don't realize that validation is A Thing and don't have enough tools to manage it, we can get caught up in indecision or analysis paralysis. Every decision is left feeling like some subjective, abstract, wishy-washy matter.

But by acknowledging it, we can recognize that being stuck between two options is just a validation issue. And if it does turn out that both are equally valid on all the fronts we're accounting for in our process, one can be selected at random, as they're both "valid options." A final decision must be made, though, to ensure we can validate future data against whichever was settled on.