From a different perspective, it is my experience that most programmers/developers do not plan for success and the "prototype" is almost always becomes Release 1.0. I have first hand experience with 4 separate original products in which the classy, sexy, and highly functional front-end (basically the UI) resulted in wide-spread user adoption and enthusiasm. In each of these products, performance problems began to creep in within relatively short times (1 to 2 years) particularly as larger, more demanding customers, began to adopt the product. Very soon performance dominated the issues list, although new feature development dominated management's priority list. Customers became more and more frustrated as each release added new features which sounded great but were almost inaccessible due to performance issues.

So, very fundamental design and implementation flaws that were of little or no concern in the "proto-type" became major stumbling blocks to long-term success of the products (and the companies).

Your customer demo may look and perform great on your laptop with XML DOMs, SQL Express, and lots of client-side cached data. The production system will probably crash a burn if you are successful.

In 1976 we were still debating the optimal ways of calculating a square root or sorting a large array and Don Knuth's adage was directed at the mistake of focusing on optimizing that sort of low level routine early in the design process rather than focusing on solving the problem and then optimizing localized regions of code.

When one repeats the adage as an excuse for not writing efficient code (C++, VB, T-SQL or otherwise), or for not properly designing the data store, or for not considering the net work architecture, then IMO they are just demonstrating a very shallow understanding of the real nature of our work. Ray