This paper discusses a study on system nervousness in a multi-agent manufacturing control system. Manufacturing control systems, built along this approach, are able to generate short-term forecasts that predict both resource loads and order routings. These forecasts become known throughout the multi-agent system with some time delay. If the agents make their decisions based on these forecasts, proper measures need to be taken to account for these delays, especially when disturbances (rush orders, machine breakdowns) occur. If agents react too eagerly and swiftly, the forecasts become unreliable. This paper studies this issue and the measures in the control system design that address the problem. More precisely, the agents behave in a socially acceptable manner that reconciles adaptation to changed circumstances with predictability.