You can't generalise ETTR in that way, whether and by how much it overexposes or underexposes depends on what the autoexposure does (since that's the frame of reference in this case) and on the concrete scene (the maximum difference between the "average" and maximum brightness).The point of ETTR is to allow more light to hit the sensor.
No, the point of ETTR is to avoid blown highlights while not losing detail in shadows. It might mean less light hitting the sensor, if average metering was going to blow important highlights.
In digital photography, exposing to the right (ETTR) is the technique of increasing the exposure of an image in order to collect the maximum amount of light and thus get the optimum performance out of the digital image sensor.
+1. It's all about getting as much light as you can without clipping the highlights while reducing the shadows. It's equivalent to an overexposure of around 1/3 EV.
A dark scene with one slightly brighter spot will need a few stops of overexposure; a normally lit scene with one very very bright object will need some underexposure.
So I'd say that Wikipedia is wrong; ETTR is a "technique" to ensure the best exposure without clipping the highlights. In some cases it increases the exposure, in others it decreases it. If ETTR were a simple overexposure or underexposure we wouldn't need a new term for it.