hacker.org

OK, now there is no doubt how to decode it.
Excell is accurate enough to generate familiar first 4 letters ...

Hmm, even python precission does not go past the first 4 letters ... except when I use 2361 when it fills both first words correctly and continues with giberish . ... (I sum up to 99999 resp 100000).

Using 0,075 instead of 0,074 helps ...

BTW: Excell did well on first three words, but failed for the rest.

Yeah...
Damn. It seems that the NTL library does not calculate accurate.

eulerscheZahl wrote:The value was changed back to 2.360 just two days ago from a German IP (teebee, is that you ?).

Well ...

eulerscheZahl wrote:Maybe that's the reason, that noone solved it for a long time.

Know the past to act in present and rule in the future.

Yes, the list of solvers is growing rapidly. Whenever go.to.hell solves a challenge Samora do also, and vice versa. A very extraordinary coincidence.

I agree that the method of encoding which was used here is probably worthy of being included in a challenge. But I strongly disagree that it should depend on luck to find the "right" frequencies for the symbols. (as in: the same ones that were used when encoding.)