This post originated from an RSS feed registered with Java Buzz by Florian and Rolf. Original Post: A little knowledge Feed Title: Heulen wie ein Schlosshund Feed URL: http://www.contegix.com/blog/index.rdf Feed Description: Florian and Rolf talk about the trials and tribulations of software development in the java world. Latest Java Buzz Posts Latest Java Buzz Posts by Florian and Rolf Latest Posts From Heulen wie ein Schlosshund

I find the people with a small amount of technical knowledge are the most dangerous. Especially when they are the individuals responsible for making important project decisions. Decisions such as devising project estimates involve a lot of experience and skill ….even in the best case they are rarely close to reality. Reality always wins.

This is the situation I find myself in at the moment. My current client has the need to develop two quite separate and public facing web sites. Content for both sites needs to be updated by external agencies and approved by a system administrator prior to being made available to the public. Mmmm.... it kind of sounds like they need a CMS. Since both applications have the same CMS requirement, the client deems that both applications are identical.

For various internal political reasons, it is easier to obtain funding for custom software than the approval of the use of a CMS (even an open source CMS) system. Did I mention that this client has no software development experience?

Okay...so Reg, who is the local weekend hacker at the clients, decides to spend a weekend whipping up a mock version of the application that involves having a static version of the entire content of one of the websites in a single XML file, a few asp pages, a little jigger pokey and hey presto we have a fully working web site. Except that it doesn’t have the content feed from the agencies, or version control or staging approval of content, or user subscription support (forgot to mention that earlier). Reg obviously knows what he is doing and is obviously qualified to give a suitable estimate for implementing the entire system. "What do you say Reg?" Reg flexes his fingers and says with thoughtful consideration “4 weeks!”

So here I am. Leading a small team to developing two application that would be at least an 80% fit for most of the of the shelf CMS systems available. Developing two J2EE applications to be deployed on WebLogic with an Oracle 10g (because of its XML support) back end. These applications are due to go live in a matter of weeks since Reg the weekend hacker said he could implement the whole lot in 4 weeks.

So far, the first week of development has been spent gathering requirements so that the development team actually know what to build. You scratch the surface with the customer and it becomes very obvious that they actually have very little idea of what they really want the system to do. What they do know is that they want it in 4 weeks...6 weeks will do...but definitely no later than 8.

What ever estimate they enforce on us, reality will win and there is nothing I can do about it.