The calculator is useful, but some people misunderstand how to apply the figures.
Suppose you're running 1000 kilo hash per second. The calculator says: 50% chance in 6 days, average is 9 days, 95% chance in 27 days.
Now if you run for 9 days and haven't generated anything, you might think you only have 18 more days to go until you are up to 95% chance (based on 27 days minus the 9 days you've already spent).
Uh uh, it doesn't work that way.
After you have spent some time generating hashes without gaining any coins, that time is lost and doesn't affect the future. So if you haven't generated any coins after 9 days, on average you need to wait another 9 days to generate coins, and you need to wait another 27 days from that point until you have 95% chance of having generated coins.
For the record I'm on 2000 kHash/s and have generated 50 coins in 9 days.
I wrote that calculator.Suppose you're running 1000 kilo hash per second. The calculator says: 50% chance in 6 days, average is 9 days, 95% chance in 27 days.
Now if you run for 9 days and haven't generated anything, you might think you only have 18 more days to go until you are up to 95% chance (based on 27 days minus the 9 days you've already spent).
Uh uh, it doesn't work that way.
After you have spent some time generating hashes without gaining any coins, that time is lost and doesn't affect the future. So if you haven't generated any coins after 9 days, on average you need to wait another 9 days to generate coins, and you need to wait another 27 days from that point until you have 95% chance of having generated coins.
For the record I'm on 2000 kHash/s and have generated 50 coins in 9 days.
My understanding was always that (based on those numbers) in any 27 day period, you have a 95% chance of generating one or more blocks. In any 6 day period, you have a 50% chance of generating one or more blocks. I might have screwed up the math somewhere - I don't know stats that well. The formulae came from a friend of mine who read the design paper. Apparently it has some code for calculating block generation.
P(p,t)=1-e^(-pt)
Where p is the probability that you will win in one second: p=khps*1000*target/2^256, and t is the number of seconds.
So I solved for t:
t = -ln(1-P)/p
If anyone has a better understanding of the math and sees a mistake, let me know!