View Single Post
  #17  
Old 03-03-2009, 02:37 AM
admin's Avatar
admin admin is offline
Administrator
 
Join Date: Jul 2008
Posts: 148
Default

Hi again -

Well, I've been crunching some numbers and have made some conclusions about this issue. There are definitely "bogus" times being counted as valid which should not be, and these will be removed from the system.

The tricky part is trying to determine what is a "valid" score and what isn't. I've run a few models based on half-a-million solved puzzles, plotting the total solve time against how frequently they occur in the system, and I came up with the following lop-sided bell-curve:

graph.jpg

You'll notice there's a rather misshapen "bump" that proceeds an otherwise normal-looking bell-curve, starting at 11 seconds, peaking at around 32 seconds, and then dropping back to near-zero at 50-60 seconds. From 60-65 seconds onwards the graph looks like a normal bell-curve.

My conclusion from this is that all scores below 50-60 seconds are bogus, and I will be treating these as such when we refactor our statistics. If there is anyone here who can legitimately solve a puzzle in under 50-60 seconds, please contact me. Otherwise, I will stick with my assumption and discard these scores.

As far as how the bogus times are being achieved... well, I won't put that information here, but suffice it to say that after crunching some data over the past few hours an obvious trend emerged and I am 99.9% sure I know how its been done. There are measures that can help reduce the effectiveness of this tactic, and I will be employing them over the next several weeks.

Anyway, thanks to everyone for bringing this to our attention. While it is next-to-impossible to absolutely prevent cheating on an online game of any sort, I do want to do whatever I can to minimize its impact, so that folks can compete here and know that they're getting a fair shake in the standings.

Cheers -

Stephen
Reply With Quote