Interesting Cognitive Research On Road Signs

I'm increasingly motivated to study Cognitive Psychology, and this is an excellent example of why: engineers often take for granted what they're designing, oftentimes with unintended consequences or side-effects. A quick summary of the article here:

There's a problem with most of the highway signs currently being used in the U.S.: Overglow. The signs are fairly legible in the daytime, but at night, when they're illuminated solely by the reflected light from car headlights, reading becomes trickier.

My interest in this subject (Cognitive Psychology) extends to computers and information security. What design decisions are being made by software engineers that have the net effect of undermining security? And I'm not just limiting that question to poor coding practices that introduce buffer overflows or XSS or SQL injection attacks. Instead, I'm talking about the much higher level question of how the UI is designed, how prompts are written, and what expectations technical people have of non-technical people to understand what their computers are doing and/or telling them. I think the average software engineer grossly overestimates the ability of the average, reasonable person to understand what these demented engineers were thinking when they created something in a certain way.