The core algorithm of the controller hasn't changed in a very long time, a year since any changes have been made, and several years since any major change. If your output is 90% within 1 degree of setpoint, that's because that's how it is configured to respond to the conditions it is experiencing.
If the P constant is too high (like 50), 1 degree could equal 50-100% output.
If the temperature is dropping, the D constant might be too high.
If the temperature has been hanging below the setpoint, HeaterMeter keeps pushing its foot down on the accelerator more and more until it gets to the setpoint. How fast this happens is compounded by the I constant.
If your temperature is bouncing around every second, there is too much noise to make a good reading so there's no telling what output to expect.