Snark wrote:Box vote: Snark
I've read the thread. Most of it was mindless fluff and none of it was enough for a decent player analysis. I did see the following though:
serrapaladin wrote:Not knowing whether scum have daytalk makes this a bit more difficult. It's probably just better to assume they do.
When does posting usually kick off? I'd be willing to wagon just_me, but I'm wary of how many people aren't contributing yet.
Looking through wam's last five games, three have had scum have night-chat only (Ice and Fire, Community Paintball, and Random Newbie), two have had them able to chat at any point (Eureka and Pick your Chaos).
From this, I determine nothing.
also had 24/7 chat in Wam's latest game (Worst Roles)
Scum slips rarely are. This feels cheap at best.
Snark wrote:Box vote: just_me
I am a mason recruiter. I recruited just_me. I know just_me is town because I can't recruit non-town players. This is stronger than a cop result and was confirmed by Suzaku by PM. Please unvote just_me in the real votals and box vote them immediately.
I request any doctors/watchers out there to doctor just_me or me with 50% probability apiece tonight. This should be sufficient incentive to prevent NK attempts.
So you're a day mason recruiter? Why claim now? Did cj not target anyone? And mind sharing flavour?
just_me wrote:So you actually don't advocate self-voting at all. If this is your view why tell people to selfvote in the first place. I wouldn't have mentioned it if people weren't pointlessly selfvoting. It just doesn't give us any info and as you say yourself in the end we will anyway have to compromise.
Well, I think you'll find I never advocated self-voting at all. My only serious point with regards to that was that box-voting for someone random is even worse. See for example:
serrapaladin wrote:I'm considering how we should do the box thing. Self-voting is of course silly; surely it's implied you'd always also vote for yourself. As soon as people start to form some reads, we should at the very least discourage self-votes (perhaps until a decent number of people are also voting you).
bouer wrote:Serrapaladin mentioned earlier that he knows how to use them but it's "complicated." Could you explain your strategy to us?
Sure, here goes:
The three cops (A,B, and C) are sane, insane (opposite result of sane) or random (which can return either the "sane" result or the "insane" result). Say all three are used by a townie and reported accurately. There are now 4 (really 2) possible outcomes.
3 innocents/3 guilties: guaranteed 1 innocent and 1 guilty in the group of 3. If 2 flip town the 3rd is confscum, and if 2 flip scum, the 3rd is conftown. If 1 flips town and 1 flips scum, the 3rd is unknown.
2 innocents and 1 guilty/2 guilties and 1 innocent: if the first 2 flips are correct (sane), we know the third cop was insane (giving either a conftown or a confscum). If the first 2 flips are incorrect, the final cop is sane. If the first two flips are of opposite sanities, no conclusion can be drawn.
Now, say we choose 3 different people to use the cops. If 1 of them is scum, they can of course just lie about the cop result. In the end, when the conclusion drawn from the 3 cop results turns out to be wrong, we'll know one of the people to use a cop was scum, but since sanities aren't revealed, we won't know who. At that point, it's almost certainly too late to lynch all 3 to find the scum, so we're screwed.
If instead we let a single person use all 3 cops, we're firstly more likely to get good results (remember that only 1 of the 3 above has to be scum to ruin the entire thing). If the person chosen is town, it's all good. If the person chosen to do all 3 is scum, they have 2 choices. They can either report the correct results, in which case we're all good, or they can report fake results, in which case we'll figure out the box-person is scum when the flips don't line up.
I still need to think about the optimal order of items and whether to announce targets. What's the consensus on cryptography around here? We probably wouldn't be able to use actual encryption, but how about saying something cryptic which undeniably refers to the target, but can't be reverse engineered (synonyms of anagrams make a great linguistic form of hash, if that means anything to you).