I don't like the "after a couple of rounds, 7s are allowed". It undermines part of the randomness. Our 'house game' simply had the problem that the robber arrives too quickly. So we made the rule that the robber can't advance until Player 1 is rolling for the third time. From that point forward, the robber advances. This gives the advantage of allowing more development, but without obviating the 'randomness' of robber landing.<p>Also: Free settlers-like multiplayer: <a href="http://games.asobrain.com/" rel="nofollow">http://games.asobrain.com/</a>
With three d10s and one d6 you could do this:
d10 a: 6,6,6,6,6,8,8,8,8,8
d10 b: 4,4,4,10,10,10,3,3,11,11
d10 c: 5,5,5,5,9,9,9,9,1,12
d6: d10a,d10a,d10b,d10b,d10c,d10c<p>One roll, no relabeling. d6 tells which d10 to pull your number from.
This is interesting, but this is a bit more hackish than simply re-rolling the dice. Most people would just reroll.<p>My personal opinion is that dice-hacks should be done to reduce the amount of thinking necessary, not increase it while producing a slightly better result.<p>I designed a role-playing system once based on the median of 5 d10s. That worked pretty well. It generates a similar bell-curve distribution to the one you get by adding dice, but doesn't require mental arithmetic with large numbers (e.g. 5d10 involves a range of 5-50). I think m5d10 is better than 5d10 because it's easy to develop a feel for "difficulty 7" vs. "difficulty 8"... whereas if the numbers are 37 vs. 38, it gets to the point where 1 point doesn't mean much, and the numbers seem completely made up (which, of course, they are, but they shouldn't <i>seem</i> so).