The eventual solution is to not care how big blocks get.  

When the demand arrives (which will be decades from now), 10 GB and even 100 GB blocks will be feasible.  This is fairly straightforward engineering.

But for now, our research shows that the throughput capacity of a global network of bitcoin nodes is limited to about 100 transactions per second (30x times the current throughput), due to the single-threaded mempool code path of Satoshi-derived clients.  After Andrew Stone's parallelization of mempool, the capacity was increased to about 500 transactions per second (150x the current throughput).  At this point, block propagation becomes the bottleneck.  There is still lots of work to be done.

.

I realize that you're being sarcastic, but you couldn't actually scale bitcoin to a galactic payment system due to speed-of-light constraints.  Happily, since all our world is reachable by an electromagnetic signal in less than 100 ms, the actual bitcoin network doesn't have this physical scaling limitation.