Everybody has those moments where you come across your old work and groan. Sometimes you don't even need to see it. Out of the blue, you just remember something, and boy, does it ruin your day. A couple jobs ago I was employed as a developer, but in reality I was the UI guy. The previous UI guy left before I got there. The interface he left me was...shall we say...less than ideal. The user interaction model included a limited set of interface paradigms (every problem was solvable with a hammer or a screw driver). The interface was IE specific TBL madness. The color scheme was atrocious. The Information Architecture was fragmented and unclear. The CSS for all pages was in one huge unorganized file with useless nomenclature. Excessive amounts of procedural javascript were required to make the interface function at all. And beyond all that, it did a rotten job of addressing the customers' operational requirements. It was quite the house of cards. Well, after some ramp up time, I decided to start introducing some interface changes. Nothing too big or sudden. I had grand ideas and plenty of notes and sketches, but those were a major rewrite away. So in the mean time, I took some basic principles and began introducing them into the UI. And introducing them. And introducing them. MAN, did I introduce them! Iconography, for example. I went completely overboard with the icons. On various pages, a medley of different component objects would be presented in a table with discriminating information in a tiny font amidst misaligned columns. The problem I attacked was that the users couldn't easily identify what they were seeing. That, however, wasn't the problem, but the symptom. The real problem was that the users didn't care about what we were showing them. The tasks they needed to accomplish and the tools we'd given them to do it just didn't jive. We'd solved their problems our way and expected them to get on board. The puzzle pieces we used didn't correlate at all to their MO. Therefore, went my thinking at the time, every bit of extra information was helpful, right? I went on a rampage. I wanted those icons everywhere. Sometimes one item would have a little zoo of icons—one type indicative icon, and a collection of others representing various meta data. In lieu of an intuitive UI, I wanted to give the user as much data as their little color receptors could take in. The use of color to communicate simple concepts is more efficient than relying on text, but what I should have done was spend time removing information. The graphic elements drastically reduced the active real estate and served no good purpose. Within 1024 x 768, only about 500 x 300px were gainfully used. Removing some of this cruft would have offered more breathing room to the bad IA. The dense TBL was largely preventative when it came to reorganizing layout. Breaking up, annotating, and categorizing the CSS would have made the surgical removal of table after table more manageable and the introduction of new interface paradigms easier. In retrospect, I wish I would have attacked the fundamental issues rather than the symptoms. I also wish I didn't have all that other work to manage while doing it :) It was a more educational experience than I realized, and I'm happy to say there were things I produced that I am still proud of. I still think about how to improve that interface, though. It's amazing what people pay for these days.