The reuse of software components is the key to improving productivity and quality levels in software engineering. However although the technologies for plugging together components have evolved dramatically over the last few years (e.g. EJB, NET Web Services) the technologies for actually finding them in the first place are still relatively immature. In this paper we present a simple but effective approach for harvesting software components from the Internet. The initial discovery of components is achieved using a standard web search engine such as Google, and the evaluation of "fitness for purpose" is performed by automated testing. Since test-driven evaluation of software is the hallmark of Extreme Programming, and the approach naturally complements the extreme approach to software engineering, we refer to it as "Extreme Harvesting". The paper first explains the principles behind Extreme Harvesting and then describes a prototype implementation.
