Iron losses are one of the losses caused associated with an electric machine:
http://en.wikipedia.org/wiki/Iron_loss
Copper losses are another source of inefficiency that we worry about. In general, iron losses are considered a bad thing because they make an electric machine less efficient! In fact, here's a quote from a reputable motor designer on the negative impact of iron loss:
http://www.lynxmotiontechnology.com/applications.htm
In many cases, efficiencies in excess of 98% are achievable. This high efficiency is achieved through the elimination of iron losses inherent in most electric motor designs.
I have to caution that this 98% number does not include bearing and windage losses. I should also mention that I was involved in the test of a SEMA motor (I wrote the statement of work, test plan and procedure for the motor as well as managed the $150,000 contrace our company gave to Lynx) so I have a certain amount of knowlege when it comes to electric machines.
That said, I was impressed by the video because I saw almost everything I would have expected from such a test. I didn't see a dynamometer, though, and that is somewhat disturbing. But I did like the big thermal box around the machine. We did something similar with the SEMA motor so we could estimate how many BTU's of heat were being generated. Along with a motor, we were designing a cooling system so that information was very necessary for our project. But even with the insulating box and thermocouples all over the motor, we never really got good numbers for the heat output. It is a difficult thing to measure with 'no error', which is something not mentioned in the video.
Anyway, I have to say it is B.S. to include heat output as part of the efficiency number. There's always going to be 'extra' heat because math models are based upon the electromagnetic equations and they generally ignore friction losses (bearing, windage). This means you'll get more 'heat' than you expected based on the math model. It seems these guys are claiming they're getting extra energy but what they really have is more heat than predicted by their model.
Finally, the idea that they are getting extra heat can be attributed to measurement error. For example, during our project we had a 270V power distribution box that had voltage sensors on the input and output. What was interesting was that we never were able to measure a voltage drop across the box, even when we had 300-400 amps running through it. If I had wanted to, I could have claimed that our box was actually a superconductor! Of course, there was a small voltage drop but since our sensors were calibrated to read up to 400V, the small drop was beneath our measurement error (my guess is that the heat output measurements from EBM are similarly flawed and they are taking advantage of their measurement error to get an 'over-unity' energy output). Most of the others on the project were mechanical engineers so they didn't appreciate our 'lossless' power distribution box. Fortunately, our managers never heard about it, otherwise, they would have made us patent the thing!