This paper tries to give a new point of view that shows what should be controlled to avoid or reduce or possibly eliminate the traffic jam. First, it is shown that 1) Keeping appropriate distance between cars even when they are in low speed or stopping condition, 2) All cars should start with no delay, are 2 requirements for avoiding traffic jam and these ideas would not be provided by AI only with vast scale learning of general traffic. Next, 2 requirements for avoiding traffic jam are verified by experiments on the test course with using 10 cars in various conditions of distance between cars by 3 measurement methods complemented each other. Then, as a practice, it is reported how it can be explained and how it should be solved the problem of the place where traffic jam occurs frequently in Japan being applied the 2 simple requirements. And finally, it is suggested that all cars should be equipped the device that warns less distance and start timing to the following car that is easily implemented to safety cars, and vast reservation system should be adopted to expressway to limit numbers of cars on safe ground that is theoretically derived from the 2 simple requirement for deterrence of traffic jam discussed on this paper.
