Lowest possible power

In many CW contest QSO's and other QSO's, I use the Lowest Possible Power. To use the lowest possible power, I always answer a CQ, instead of calling CQ myself. First I adjust to a power that is probably just high enough to make the QSO. When the received signal is strong, I can use a lower the power.
When the received signal is S9, I answer with 500 mW. But when the signal is S9 + 10 dB, I use 50 mW.

When the S-meter goes up, the power goes down.

When I switch the FT-817 to 500 mW, using an attenuator of 10 dB, the output to the antenna is 50 mW.
I answer by giving my call just once or twice. Then I listen. When the station is still calling CQ after my call, my signal is not heard. Only when I notice that my signal is not heard, I increase the power and try again. If the station answers an other station, I wait patiently.

When I use the attenuator, I listen and transmit through the attenuator.

When I increase the power, I always use a step of 3 dB or 4 dB. 3 dB is a doubling of the power and with 4 dB the power is 2.5 times as large. In hundreds of QSO's "on the edge", I discovered that if my call was heard but NOT copied, I only had to double my power for my call to be copied immediately correct. So doubling the power, thus a step of 3 dB, is the optimal step to increase the power in QSO's.
The frequency must be absolutely clear for my signal to be heard. With many other callers in a contest, this can be frustrating. But many times suddenly when I am the only station calling, the station repeats my call, immediately correct and gives his report. Most of the time I receive 599. (hi) After sending my exchange, I can log the QSO.
By operating in such a way, I have great fun with very low power and at the same time I develope my operating skills. Sometimes I start with a power that is probably to low, because in a contest I can not make a second QSO with the same station, with a lower power. When I change band, while the propagation is very good, I first look for very strong stations.

Sometimes it is very difficult to be patient and NOT to increase the power, every time another station gets a report .

How is it possible
A CW signal with a signal strength of S9 is very strong. But a signal with S5 or even lower can be received easily. If a station, that is transmitting with 100 W, is received with a signal strenght of S9, then my 500 mW signal will be received with 559.
This is 4 S-point down. The signal will not be very strong, but the readable will be good. When the propagation is good to extremely good, the signal of the 100 W station will be S9 + 10 dB, so I can reduce my power to 50 mW, using an attenuator of 10 dB and still be received with 559. hi

Distribution of power
I found out, by using the lowest possible power in contest QSO's, that in the QSO's are evenly distributed over the power categories, when the propagation is very good.
Therefor it is great fun to lower your power, when the propagation is very good.
This this phenomenon is clearly visible on 20 m in the OK-OM DX contest of 2011.
Notice that more than half of the number of QSO's are made with 80 mW or less.

(Nog doen: Dupes en TU)