Alternatives ...
•Many models have been proposed, but they have had weaknesses
•LogP [UC Berkeley] is CTA+parameters
•The model measures latency and other variables related to communication -- too many free variables
•The model was effective at predicting behavior on the Thinking Machines CM-5, which motivated it
•With many parameters, it was difficult to decide which were significant, especially when their values are free
•It remains a model of theoretical interest
LogP’s parameters try being exact about l. Not possible