Have you found any more addresses like that recently?
I am attempting to generate a second 80-bit address pair, but no luck yet. Pairgen is a bit different than vanitygen, in that the more work you do, the probably of finding a solution increases. I was actually lucky to find the first pair so quickly (after one week, was expected about two weeks).
You can however easily generate smaller pairs, e.g. 51-bit collision (for 10chars) generated just now:
message = "This is a real Bitcoin address."
sig[1] = HxupEyHga60CrSmm298Zp1thMYIsAoGaaGCusFKSSYkkf40nQx0dZsxd8jdQlW1wkJIQ8GPPsvOyZXd 2rjOpI8o=
sig[2] = IGUrV4deRXq7DfSOzHgl6DgbwGr5dSp+SCi4ApehLf4AADNVhb1BgfHxZuwpHgicd2ioHY342WbZFsAiTXMx2Vc=
shared = 10chars
addr[1] = 17qvygfdtmtnqAkGpStab7rfXQTvirMhZT
addr[2] = 17qvygfdtmharpKrTahbL1oQvSb9vTqVYt
This took about 46 seconds to generate.
As a very very rough guide (based on my CPU), ~40bits = 3seconds, ~50bits = 1minute, ~60bits = 1hour, ~70bits = 1day, ~80bits = 10 days, etc., etc.
If you want to really go nuts, you can use the "
--job" to distribute address generation across multiple machines (see the docs). This is similar to vanitygen's split key mining. With enough cores, probably 100+ bits are do-able. Alternatively, one could port pairgen to use vanitygen's
calc_addrs.cl for GPU mining.