You can conserve energy if you turn energy just into an accounting process, so that it plays the same roll as currency in economics and use cosmology science
Ethan shows that approach nicely in an earlier article
http://scienceblogs.com/startswithabang/2011/12/02/dark-energy-accelerated-expans/It's a nice story unfortunately it's only true in cosmology to a target audience, Bill G would love it
Let me show you the problem using the positive side of that cosmological argument and it goes like this
Photons are redshifted, losing energy as space expands that is our observation we see. If we observe a fixed number of photons, the number stays constant while the energy per photon decreases, so the total energy in under observation decreases. Apply that to the whole universe and the universe photons are bleeding energy. So now not even the visible positive side of the universe is not conserving energy using Ethan's approach.
The problem was picked up here and they give the correct answer which is ticked
http://physics.stackexchange.com/questions/13577/photons-in-expanding-space-how-is-energy-conservedThe problem is you can't easily build a single reference frame and Ethan's answer although it seems to comfort him it fails for the same reason. There is no reference frame for what he does and the argument is complete fail but it gives the answer he feels it needs to be. He means well but it is a very glossy media answer to "settle the science" for layman which is unfortunately entirely wrong at it's heart.
The specific problem is Energy conservation only holds on a static background with time-translation invariance. You can't create a reference frame for the universe as a whole so the question is not answerable under normal classic physics and it can not be simplified to a layman level. I don't care how many glossy media articles they put that garbage in and what idiots they drag out to say it, the question can not be answered in that way.
The correct answer is beautifully simplified by Lubos Motl
The time-translational invariance is broken, so via Noether's theorem, one doesn't expect a conserved quantity. Also, if one defines the "total" stress energy tensor as a variation of the action with respect to the metric tensor, it vanishes in GR because the metric tensor is dynamical and the variation has to vanish because it's an equation of motion (Einstein's equations).
If the space is asymptotically flat or AdS or similarly simple, a conservation law - for the ADM energy - may be revived.