Adaptive multiagent algorithms based upon the behaviour of social insects are powerful decentralised systems capable of solving complex problems. The intelligence of such a system lies not within a single agent but is a product of a network of simple interactions. Under the context of a mail collection environment different techniques are implemented and evaluated. The paper creates a number of strategies to tackle task allocation problems of this type based upon the principles of self-organisation and greedy search. The paper also investigates factors that may affect their performance.