Iran once had 70,000 qanats — gravity-fed tunnels that carried groundwater to the surface through gently sloping channels dug by hand into hillsides. The oldest have operated for over 2,500 years. They were called “everlasting springs” because they never ran dry.
The reason they never ran dry is the reason they were replaced.
A qanat cannot pump. It taps into the water table at depth and lets gravity carry water downhill through a tunnel with a slope of less than 0.5%. The flow rate is controlled entirely by the height of the water table above the tunnel floor. When the aquifer is full, more water flows. When the aquifer drops, flow decreases. When the water table falls below the tunnel, the qanat goes silent. There is no way to extract water faster than the aquifer recharges.
This was seen as a limitation. Starting in the 1950s, Iran began installing motorized wells with powerful pumps — over one million in forty years. The pumps could extract water regardless of aquifer level, regardless of recharge rate. They were faster, deeper, and more productive than any tunnel.
Within decades, Iran lost 210 cubic kilometers of stored groundwater. Water successfully extracted from those million-plus wells fell 18% even as well numbers doubled, because there was less left to extract. Half of the original 70,000 qanats went dry — not from their own extraction, but because the pumped wells around them pulled the water table below their tunnels.
The qanat's design encoded a physical constraint that prevented catastrophic behavior. The limitation wasn't a bug in the engineering — it was the engineering. The tunnel's position relative to the water table created an automatic feedback loop: extraction rate was coupled to supply. Remove the coupling, and you remove the regulation.
The pump didn't add capability to the system. It subtracted a constraint. What it subtracted was the thing keeping the system alive.