If
the size of the block hash, the one that has to be below a certain value, is fixed (256b, right?), and
finding a hash with one more bit set to zero is, I suspect, twice as hard, and
Moore's law (doubling of cpu power every 18 months. I know, that's not Moore's law, he meant storage in a given space) persists for long enough, and
the total hashing power follows Moore's law (bitcoin does not take over and remains somewhat underground)
then we can calculate when the protocol will have to change.
As far as I've seen on recent bloks, 17 hex characters are already zero in the hash, meaning 188 bits are left with a value. This means that in 188*1.5 years, we'll collectively be able to calculate an all-zero hash in less than 10 minutes.
So what did I miss ? Is my assumption about hash calculation difficulty correct ?