[dalke]

Some of my consulting work integrates different packages - commercial, academic, and in-house - into a single system. It also needs to understand the local database configuration, the cluster queuing system, the shared filesystem, and other customer-specific configurations.

I have a framework for this project. It takes about a week to write. The other year of work I did for one client were the modules and components to fit their internal requirements. This included porting some of their software from IRIX to Linux. Overall, only a relatively small amount of that development could be shared with another client.

I thought about spinning that part off as a product. I haven't been able to make a viable business plan out of it. The type of people who could build their own components for my framework is the same as the one who could build the framework in the first place - and indeed, other companies have in-house frameworks for the same thing. And while I think mine is better, it's biased towards Python but this sort of software tends to be written by IT staff more interested in C# or Java.

Also, I like working close to the end-users, with frequent iterations to get what they want. Selling software (I work in chemistry, where it's hard to develop web-based apps due to the proprietary nature of the end-user's chemical structures) is one step removed from the users, so I don't have the same feeling of connection.

When consulting it's easier to know that work = income = profit. With a startup or product development, you spend a lot more effort to develop a product before knowing that you'll make profit.