Coordination systems have been gaining popularity since the early 80s with the introduction of the LINDA coordination model. Soon after its introduction researchers and practitioners alike started to realize that coordination is ubiquitous to any distributed systems but unfortunately partially responsible for the inefficiency found in these systems-coordination deals with costly issues such as process communication and synchronization. In the beginning of this decade, researchers looked for alternatives for implementing more efficient means of coordination; they turned to Swarm Intelligence; the first of these approaches was called SWARMLINDA. Performance of LINDA-based coordination systems is directly related to the issue of system entropy (tuple organization). SWARMLINDA approaches for tuple organization are investigated in this paper using a simulator. After a careful study on the performance of SWARMLINDA, we introduce modifications on the algorithm so as to achieve better entropic levels.