Babar,

You very accurately describe what I've come to think of as the "normal process" of beginning a design:

"Its whenever I have to start an altogether new program from scratch. I start it very lazily. After making an algoritm I find it difficult to quickly finalize the initial details of the actual software"

It takes time to take in all the information and to "cook" it around in one's mind. Unless you have a short deadline, its probably just your mind searching continually for a better solution.

This process is like what computer science calls "Simulated Annealing". Its a progressively restricting random search for a near-optimum solution patterned after the way molecules form crystals (hence the name). We used the technique in Telecom for getting "good" solutions to optimal-incalculable problems.

In the beginning of a project you can jump around and examine various approaches and components but in time you take less and less radical re-directions as you zero-in around a good design. The random searches into other solutions prevent you from being trapped in a solution niche initially expected (forcing the design); most optimal solutions to complex problems are not where you expect them - if they were they'd be considered common sense.

I don't think someone has to be in the office to be thinking about better solutions. I do some of my best innovation riding my motorcycle or even sleeping at night. Of course, I don't get paid for either. :-)

-Jay C. Box