QuickTopic free message boards logo
Skip to Messages


Eco and KJ/Mentoring Thread in Osie's/ Haiku Mafia

^     All messages            167-182 of 182  151-166 >>
EcophagyPerson was signed in when posted
02:00 PM PT (US)
/m181 OK. I'm not going to go back and answer questions I couldn't because they were about specific reads and events, and I don't really want to have to read haikus to formulate answers. But the general points and process is more important. I did have a quick skim through your ISO though.

Overall, you played fine. Even though your early reads were very innaccurate (not an uncommon occurrance, it would be very foolish to bank everything on D1/2 reads), you did correctly come around to Rum and Rod, and I think you deserve some kudos for being open enough to change your role-related town clear of Rod (even if it took the slightly lucky track to achieve). More important than just the reuslts is the process.

I think that a lot of your reads posted in thread were very shallowly supported, and you posted a lack of actual, meaty analysis. Obviously the posting restriction would have contributed to this, but there wasn't really much in here either. I could have done more to foster it (that's what /m152 and /m153 were trying to do), and analysis in here is slightly hollow since I couldn't have comment on it. But my point is that there wasn't much indication that you were really sitting down and analysing someone and coming to a supported conclusion either privately to concretely record your thoughts, or publically in the thread to push cases and explain your positions. This indicates a lack of depth to your approach to this game, and I think you can kind of see that in your read progression where they only really changed as your scum reads died, and you ended up on Rum by default since he'd always been a null read, and Rodemy took a track result to nail down.

The one real analysis post you did, #740, was good, but still lacking. For 1.15 I really liked the "LnGrRRR and Chris/Rum probably not buddies" observation, and there probably being scum on Dawning is not a bad observation even though it turned out to be wrong (but you could consider: Dawning was a really easy lynch for being Dawning, but also refusing to claim at the death of the Day. Scum didn't really need to try and push it). With the magic of hindsight here, I think Rodemy looks really bad here on a vanity wagon, but you kind of just totally glossed over that. I acutally think you polluted your wagon analysis by going off and thinking about other things like claims. I also thinkn that colouring Shadow in blue is risky: even though he was cop cleared at the time (right?), a mod who is prepared to put a miller in a game is very capable of having a godfather too. I would have stuck to pure facts, then had a look at how things changed if Shadow got coloured in as blue. I also think you lose a lot by only looking at the final votecount of the Day: you could have learned a lot by at least tracking the growth of the Dawning wagon over a few votecounts (and if you did look at this, you need to say and explain what you saw, even if it was nothing).

Final D3 votecount suffers from pollution, but way worse: with Rodemy coloured in blue, the wagon analysis has vastly less value. Still, I can't fault you for finding Psychatrog's position suspicious there, since both scum bussing DV is probably the less likely scenario (but again, progression of wagons is important too, which you did kind of touch on - so right conclusion, but colouring in Rodemy ruins it). Analysis of 4.1 is obviously useless, so I think that including it was actually misleading/more there to "support" the push on DoTA rather than to analyse.

So overall, #740 was a solid attempt, but it was incomplete: if it was intended as wagon analysis, it should have been primarily just that: totally fact based, looking at snapshots and progressions to find patterns. Instead you took a couple of (useful) snapshots and picked out a couple of useful points (but not the whole picture), but also applied additional speculative details that polluted the analysis. There are certainly aspects of the post that were really good and you can build on. The key thing to work on beyond that is just to do it more. One meaty analysis post out of 130-odd is just not enough. Not only are they helpful for catching scum, but (more often and arguably more valuably) they clearly demonstrate your town mindset. Making yourself an obvious town read is just so incredibly valuable, and something I think you know you need to work on.

Looking over the QT again, I think I see a pattern of thoughts that hinders your ability to analyse. Many of your questions were effectively "is this thing that person X has done scummy?". Examples like "is clearing someone over nothing scummy?" or "is claiming miller usually a scum gambit?" "are mind melds town?" "is refusing to claim a scum tell?" etc. You also got really hung up on stuff like claims and dichtomonies and instances that looked weird (or buddying or whatever). I think this belies a focus on individual instances and moments of behaviour that are not actually that alignment indicative in the grand scheme of things - or at the very least, are often heavily ladened with WIFOM (which is why my answer to basically every "is X scummy?" question was "It depends"). Yes obviously stuff like claims and dichtomies can be very powerful tools (and I didn't really see any evidence of you trying to analyse the setup and claimed roles within it, which you really should have done, although the setup was pretty wonky so it might not have helped that much); but I think you really need to take a step back and look at the bigger picture more often and analyse patterns and trends, not try to divine alignment from a single data point because that is so heavily laden with WIFOM. Alongside this, it's being able to find the time to delve into an ISO and really think about a person's actions and mindset, as opposed to a surface level "he did this thing that looks kind of scummy".

So overall, decent play, glimmers of really useful analysis, but with too narrow and shallow a focus on individual moments (especially claims I think) instead of really committing to in depth analysis of patterns and trends, while obviously hampered by the posting restriction. For the sake of practise, I would strongly suggest not worrying so much about the accuracy of your reads, but ensuring that you come to them for really solid reasons, and making extra sure to explain those reasons to the thread.

Thanks for asking so many interesting questions and letting me wax lyrical about mafia theory, this was fun. Hopefully you learned some stuff!

KilljoyPerson was signed in when posted
11:41 AM PT (US)
I think final observations might cover it?

Unless you want to go back through the chat and answer those questions you couldn;t before.
EcophagyPerson was signed in when posted
10:17 AM PT (US)
Game is over town won hooray! Got any more questions/comments/thoughts before I post a bit of a summary/final obsverations?
KilljoyPerson was signed in when posted
12:06 PM PT (US)
So it turns out I would recieve a successful track result of my target is role blocked.

So I probably just caught the guy trying to kill me. So that's coo.
EcophagyPerson was signed in when posted
04:46 AM PT (US)
Did they claim Doc before or after they knew they had been tracked? If before, then it's more likely to be real (because it isn't as convenient). If you think there's more than one scum left, you can always take the Doc claim for now and find the other scum.
KilljoyPerson was signed in when posted
01:53 AM PT (US)
Well... I tracked the claimed Doc to me, and there was no kill.

So either the claimed Doc has another ability that he can use on me and his partner chose to nokill, or he's the doc and protected me. Or tried to.

Spicy news: The claimed RBer who is deepwolfing and was cleared by the cop quickvoted Rodemy seeming to imply he blocked him.
EcophagyPerson was signed in when posted
12:52 AM PT (US)
/m175 it's certainly not impossible, and gets more likely as the game approached LyLo and/or the scum get more desperate. Like with all claims, you have to:
-Consider how likely the coincidence of the situation is: is their claimed Doc protect both a reasonable Doc protect and a reasonable scum kill target?
-What did they do other Nights? Why?
-How does their play match up with their claim? For example, Bulletproofs or ICs can play way more aggressively/loosely, Vigs shoot their own scum reads (usually) and often don't care as much about getting them lynched, Cops keep their head down and may completely change reads for no obvious reason (because they investigated someone).
-Can they prove anything? Can trackers or watchers back anything up? Do they have privileged information?
-How convenient is the claim? The more someone has to say "it's not what it looks like", the more it probably just is what it looks like (e.g. Phoenix Fire in Zero Escape). Similarly, a strong but unprovable claim on a widely suspected player right at LyLo is very suspicious (but not always!)
-How prone is the player (and the rest of the scumteam) to gambits? Some people play very conservatively, but some scum are wacky enough to try off the wall things like skipping kills on purpose.
-How does it fit into the setup? Heavily dependent on the mod, but something like a Doc fills a very specific role of protection. Compare power levels of each team and role overlaps. For more complex games/claims you can look at the role itself too: if it has an inherent contradiction or perverse incentive (that isn't an interesting player decision), then it might have just been slapped together on the fly.
KilljoyPerson was signed in when posted
11:27 PM PT (US)
So. What are the odds of scum no-killing to fake a doc claim? I feel like its not very high.

On a side note, I've done it.
EcophagyPerson was signed in when posted
11:14 PM PT (US)
Depends on the dichotomy. A genuine one like "two people have claimed cop in a Basic" does trump behaviour. But a lot of people like to claim "dichotomy" when what they actually mean is "we're two townies reading each other as scum and we're too worked up about it to back down or consider other worlds", which is when you get eyerollingly bad things like people saying "lynch me and when I flip town lynch this other player I have a strong read on".

So mechanical dichotomies are slam dunks and trump behaviour, but you have to be very sure they are real - and behaviour can be used to work out which side you lynch first.
KilljoyPerson was signed in when posted
03:27 PM PT (US)
Which is better: Dichotomy or behavior?
KilljoyPerson was signed in when posted
05:38 PM PT (US)
Ok so if we lynch the guy who claimed Miller, and he's not miller... we can look at the guy the cop cleared as possible godfather.
EcophagyPerson was signed in when posted
12:38 PM PT (US)
/m169 CONSTANT VIGILANCE. You catch deep wolves like you do normal ones: look for consistency and behaviour over time and work out what win condition they're workiong towards. They're just wolves who are playing well, which means you have to be prepared to reevaluate your town reads and look hard.

/m170 It's tough, but if you can at least be aware of your bias and the reason behind it that helps, because when bias kicks in you can say "I'm not so sure, maybe that's just his posting style that I hate" or whatever. Also try and look at things from both alignments: something might be obviously scummy but maybe there's also a town motivation behind it.
KilljoyPerson was signed in when posted
09:32 AM PT (US)
Also is there a trick to rereading without bias?
KilljoyPerson was signed in when posted
09:27 AM PT (US)
How do you uncover a deep wolf?

I'm beginning to suspect someone of being one. The way they handled the dead scum was... Probably not a buddy?
EcophagyPerson was signed in when posted
02:38 AM PT (US)
/m167 Pretty likely, that's a nice place for scum to be. The justification for the park is important: town wouldn't usually want to be sat on a wagon that's doing nothing, so were they exploring other wagons or presenting cases that supported their vote? See if they had opportunities to switch that they didn't take - especially if those opportunities were joining a wagon on scum.
Edited 01-16-2018 02:39 AM
KilljoyPerson was signed in when posted
02:36 AM PT (US)
How likely are scum to just park their vote on town all day?
^     All messages            167-182 of 182  151-166 >>

Print | RSS Views: 432 (Unique: 74 ) / Subscribers: 2 | What's this?