I have looked ar soil moisture sensing and it truely appears that the only reliable method is via capacitance techniques.
The probes can be gold plated or better still can be insulated and inert.
The resistance method is incredibly unreliable with very little calibration of errors possible.
The capacitive technique also allows depth testing as the problem area can be lumped at the tip of the probe and the capacitance read as the probe is driven in the soil. You can now profile the soil. Also you can rotate the problem to get closer to the root zone. Remember that you want the roots deep to get moisture and nutrients. You then can keep the surface evaporation rate lower. The cost of water is now a significant cost of agriculture.
Also you can incorporate a small temperature sensor to aid in getting better calibration and again soil temperature profile.
The capacitance sensors available commercially are good but expensive. An integrated sensor with the Arduino doing drive signals and ADC keeps the cost down. Tricky part is designing the analogue circuits. But there are still a few of us geeks out there that can do this.
Oh yes the capacitance measurement can be fast. The temperature sensor will be the slower response,