Short circuit, usually.
Not on its own. Short-circuit drag isn't far off open-circuit drag. AIUI on some hubs it's actually lower, because *handwaves* eddy currents.
(Hence the cunning strategy of limiting voltage by crowbarring the hub with a thyristor, rather than attempting to shunt the power in a zener diode.)
A sufficiently manky short might happen to hit the maximum power point for some speed, I suppose.
Try riding with the hub wiring disconnected, this will indicate if its an intrinsic problem within the hub or a system issue.
The torque loading of a shorted out dynamo hub is massive - they normally turn the axle in the frame. If ridden with a very low resistance short the windings can overheat and kill the insulation (hub will be scrap and the rider will also be knackered). The turning of an axle in the frame can also cause the short due to wiring damage (chicken and egg debate).
The windings of SP PV8 hubs measure at 2.5ohms - you need a decent multimeter and leads to measure low resistances. The test load used for standard drag plots on the manufacturers sites is 12ohms (giving 3W load at 6V).
The other common cause of high drag on these hubs is bearing failure, there is usually a visible sign like brown water or grease around the end caps. I have accessed and rescued the bearings on PV8 hubs by timely flushing and re greasing, its a real pain to do. All of my bearing issues have been traceable back to riding through deep flood water, the rapid cooling of the air in the hub causes the unit to draw water in past the seals. I expect using a jet wash would not do these bearings (or others on bikes) any favours.
A usefull measure of drag is a run down test method - spin the wheel by hand and see how long it takes to stop. New PV8 builds (700c with tyre and tube) take about 12 seconds (wiring disconnected).