I propose we make a reactor large enough to use gravitational confinement and get the energy out via electromagnetic radiation. We could make it safe by having it operate in vacuum at a safe distance from the earth.
I suppose it could work, we would just need to be careful about atmospheric composition to manage radiative cooling, and block some harmful em frequencies. But I don't foresee any problems with that.
The density of the plasma determines the rate of fusion, and is itself determined partly by temperature. If we were to reflect a significant fraction of the emitted radiation back, then the plasma and its surrounding gas should expand, reducing pressure on the fusing core. That should in turn reduce the rate of fusion, lowering fuel consumption.
In theory, we could even diffuse the entire hydrogen supply into a warm cloud that is not dense enough to fuse at all, and store this until we need more heat. As the system cools, it will contract, eventually reigniting itself.
Even if we fail and die, it will still reignite itself, which clearly demonstrates the inherent safety of my proposal.
He didn't really say it won't work, he mostly just said it wouldn't be as clean as people think it will be. But you are right, it does put news like this one into perspective.
Link to the HN discussion from two days ago: https://news.ycombinator.com/item?id=14202488
And if the thing happens to be rotated even by a milli-arc, for example due to asteroid strike or a simple software error, the receiving area will be fried by microwave. Or hackers might want to introduce a deliberate rotation to use it as an orbital based frying weapon. No thanks, that's way too risky.
The actual failure case to be worried about is that the reactor blows up towards the end of its life cycle.
It would act as a switch, where, as soon as the incoming beam stops, the feedback stops and the orbital transmitter would stop too, e.g. if the energy beam went out of position.