Applying the high level design concepts to the low level construction of icons was a nice thread to the article. It made me think of our own recent “breakthrough” where we realized the visual choices we’ve made in our individual icons could be generalized to affect our higher level design as well.

I’ve posted a short article on my own web site that describes what we’ve learned.

http://www.jarrettinteractiondesign.com/2003/10/14#a354

As an aside, I found it oddly sad that your article didn’t include any graphics of the icons you’ve developed. In writing my own article, I realized I was also reluctant to share my work, for two reasons: some of the graphics may be meaningfully proprietary (for now anyway, since they relate to unreleased parts of the product), and, more importantly, I never feel that I’m very good at it. I hope you don’t have the latter concern!

http://www.amazon.com/exec/obidos/tg/detail/-/0471599018/qid=1066159997/sr=1-4/ref=sr_1_4/104-3738131-3660741?v=glance&s=books

The author emphasize that you have to test, test an test the icons before they are finished. It has many ilustrations, and few text. It´s certainly a visual and pratical book.

But the greatest thing is that all guidelines from icon design values to the whole interface design. It was the most useful book on this subject I read.

During Modernism, as Western society began to question rigid adherence to traditional methods, techniques, and forms, artists began to experiment with materials and processes that were formerly the domain of crafts people. In the 20th century, the art-craft dichotomy, rigidly in place since the western Renaissance, began to soften. Craft is currently more the provenance of artists than of auto mechanics and bricklayers.

I’ve spent many hours doing the kind of icon designing Uday describes, and concur that the Photoshop portion of the program requires fine art and design skills, as well as an awareness of tradition and convention. There are some contemporary visual symbols the designer can now call upon (documents, folders, trash cans, etc.). But I’d guess more often that not, the designer has to creatively construct new ways of symbolizing concepts that may also, themselves, be new. So this part of the process could be called craft in the broadest sense — skilled labor applied towards a practical goal; or it could be called craft in a narrower, more modern sense — handwork with an overlay of innovation and creativity.

None of these definitions of craft really get at what’s going on, though, when one looks at the end-to-end icon development process. The interaction designer must work with user expectation, aesthetics, and pragmatics to create contextual fit both within the application and the user’s workflow. At it’s core, icon design is information design; every bit of visual information contained in the image must serve a stated goal.

Good icons are little kernels of information, functionality, and meaning that work on the user at a kind of pre-verbal or non-verbal level. The process of achieving this is 1 part design, 1 part intuition, and 4 parts analysis. Designing icons is somewhat like designing a screwdriver — it has to look good, it has to feel good, and it has to *work.*

I don’t think you can over-emphasize the value of solid analysis done up front in advance of any design. The time I spent grouping functionality into related concepts allowed us to begin the project by creating the basic images upon which we could build a consistent icon library.

I’ve also reeped the rewards of this prework each time a maintenance icon is required, since oftentimes the image is an obvious almagamation of existing images.

An old, but still basic book about the topic: Brenda Laurel “The Art of Computer User Interface Design”