I think this one went wrong right at the concept stage. The idea of "risk without luck" just makes absolutely no sense whatsoever. Not even the concept assessment, but rather the concept itself (incidentally, it is hard to tell how much difference the concept assessment even makes, bearing in mind that a lot of people voting might not have even read it). That one paragraph by itself is not workable. As much as I like BMB, I'm afraid that letting such an ill-defined concept slip through the net was a grave error. I can't believe I actually voted for it in the final stage - obviously I didn't read the concept carefully enough! This is also quite true, although I think some of the blame lies with the process, as for some reason only a very small number of options are made available for voting each round, when there is really little reason for that number to be limited much at all. Oh and of course no guard was a ridiculous decision, but I think we knew that already. The other two abilities were both good choices, but not together because illusion is so obviously superior. When deciding which abilities fitted the concept, people seemed overly concerned by the weakness of weak armor, when this easily could have been made up for at a later stage. People kept talking about a lack of "reward" that it offered, which was totally irrelevant - not to incur a negative effect could be considered a reward. And then they seemed to completely forget about its weakness when picking a second ability. All of the options in the second poll were WAY too strong - BMB effectively invalidated the first poll before the second poll began (not deliberately, I'm sure). The solution, IMO, is largely just to have more options in all of the polls. The added comlpexity would be more than made up for by removing needless additional polls (you only need one round of voting each time - the slightly exploitable part of IRV is in early eliminations, and having extra polls afterward does nothing to help this). We likely would have ended up with the same results, yes, but at least we could then see that the community as a whole was to blame, voting for the wrong options. As it is, the TL can easily screw up, and leave noone else to blame but himself.