Is manufacturing a science? Is marketing a science? Is engineering a science?
None of these are sciences. But they all can be informed by science or ignore science. To be “scientific” is to rely on observation and measurement instead of intuition and superstition. Controlled experiments, surveys, observational studies, testing and validation can be used in all stages of these industries to make better decisions. These do not replace creativity and innovative thinking.
Usability is not guideline compliance
There are a lot of generic guidelines, heuristics and rules-of-thumb floating around the Internet about how to make an interface more usable. If you think of usability as just generic advice such as “users don’t scroll” or “put the most important content in the upper-left” then it’s easy to see why usability is dismissed as some half-baked pseudo-science.
Many, but not all, guidelines and heuristics have data from observing users in specific contexts (websites and applications). Over-time some general guidelines can be extrapolated to interfaces. These guidelines can be helpful. But for every guideline there are at least 10 exceptions. There’s no need to blindly follow guidelines nor blindly ignore them when designing. Generic guidelines can stifle better design—if we’re all designing from the same template the pace of innovation slows.
With some innovation comes more usable software.
Software is not usable when it adheres to guidelines, heuristics and checklists. Software is not usable when a guru says it is. Software is only usable when there is evidence users complete more tasks, take less time completing tasks, commit fewer errors and have some favorable attitude about the usability of the interface. If you want to know which airline website is more usable, don’t just rely on some vague checklist that’s open to interpretation: measure users and see which ones actually have a better experience–even if they violate some guidelines.
Is usability really that important in software? Maybe users want software that evokes emotion more than usability. Perhaps, but if they do there should be some evidence for it–evidence that can be measured.