Not sure about some of the calculations in that thread:
I spotted this in several fork. For the one occurring between 32900 and 33000 the attacker had an average time between blocks of 6 minutes.
so according to the 6 minutes between block at 188 difficulty it would take in my estimation 2-2.5Gh/s (thep33t on
http://forum.feathercoin.com/index.php?topic=786.0 suggest 2.25Gh/s)
Block 32900 was at timestamp 1369144421
Block 33000 was at timestamp 1369329066
Subtract one from the other and I get 184,645 for 100 blocks or 30.77min per block not 6min. which is a huge time for an attacker between confirmations, also means the hash rate calculation was wrong.
If someone could please double check the timestamps and my figures.
At any rate this should be a warning to devs what can happen if you let your block rate get too long between confirmations.