Tislen wrote:Lets take a look at the average amount of Jailor-visitors in an optimal group in a game where we assume there is a Lookout (as if there is none it doesn't matter)
TIs -- This nerf only matters if there's a LO in the game, so we'll assume one of the TIs is LO. Second TI has a 50% chance of being a LO or Spy. Average Jailor visits produced by TI: 1/2.
TP -- This will always visit Jailor so that's +1 Jailor visitor.
TS -- This will visit Jailor if it's an Escort (out of the five options). Getting Mayor or Retri also increases chance of RTs being jailor-visitors. So +1/5.
TK -- 50% chance of being Vet which will increase the chance of RTs being jailor-visitors.
RTs -- Of the 13 town role options, five visit the Jailor. Ignoring the unique-role complexity (so getting a lower bound for how many jailor-visitors there are on average), it's a 5/13 chance of getting a jailor-visitor per-RT. Thus +15/13.
So for town visits alone the total amount of Jailor-visitors on average in a Lookout game (not counting Lookout) is 2.85384615, since this game would only be in whole numbers that seems to suggest more often than not you'll reach your limit from town alone.
but that was just the town
Now we'll consider RMs, since with the Lookout nerf it is now optimal for RMs like Framer, Disguiser, and even Forger to always visit Jailor n1 so that their Mafioso and GF can safely claim TP and even if Spy sees a visit to Jailor n1 the LO can't be certain they were shown the RMs visiting (in fact this entire theory depends on which visits the LO sees when they get too many: is it random? Based on lobby order? Role order? Role order would be the worst option by far).
RMs -- of the seven RM options, 3 will visit jailor (four if you want to include Consort), so that's +6/7.
NKs -- Arsonists will also visit Jailor n1, so that's +1/3.
Leading to a total average amount of Jailor visitors in a game with a Lookout (remember this is an underestimate since it didn't consider the increased chance of RTs being visitors due to unique roles)
4.044322344319011* Jailor visitors in a Lookout game.
*albiet to be fair this number doesn't consider Jailor jailing one of the Jailor-visitors, but even if it does that still hits the important three on average.
The amount being four isn't really relevant, if you get three visits as a Lookout that's all mafia, NE, NK, etc. needs to be able to safely claim TP, so it's a big nerf.
The problem with this is the strictness of your assumptions and the focus on n1.
1) Spy may not visit jailor (n1, or even at all). This will be especially true in sub-masters elo(*).
2) Escort may not visit jailor n1, and likely won't on later nights. I believe this holds even more true than the above claim for spy.
3) If we assume visits will change in accordance with this patch, there's no reason to assume it will only move in one direction. So if we assume RM+arso visits increase n1 to take advantage of the patch, I see no reason to neglect the assumption that esc+spy visits wouldn't be outright excluded as a counter.
4) Even assuming evil visits will increase with this patch, it's too strong a claim that RM's would overload visits. They serve their own purposes, and wouldn't all forgo the use of their abilities for the sake of a lookout-block.
5) If the focus is on n1, then accounting for a jailed otherwise-visitor
is significant.
Now, all that said, I'm grateful that someone's taken the time to address my post, as it's made me realize I made an error in my math. And with all this in mind now, let's put it together:
First I re-ran the numbers with your assumptions just to check for myself. 4.04 is a correct average, but it's still important to consider the probability of the effect itself. In your case it turns up to about 64.58%, meaning just more than 3 in 5 games will have 4+ visitors (aside from the watching lookout) on the jailor n1.
Now, factoring in various fixed assumptions from above, I recalculated (visiting roles include lo, bg, doc, disg, forger, and framer).
Average night visits: 2.80 (max considering uniques is 3.01)
Probability of >3 night visits: 25.14% (max considering uniques is 32.04%)
If you want to be ballsy and include an arso, that raises n1's avg to 3.13 (max 3.34) and prob to 36.22% (max 43.16%).
So alright, not as bad as I'd first thought. But the point still remains that in a majority of games this will not allow the level of freedom it purports to (or at least seems to imply to, given the public response). As long as lookouts are consistent in their watches, this change would not allow evils any more claimspace as tplo than they already have, neglecting extraordinary circumstances.
The
only way an evil would have tplo claimspace, where they otherwise wouldn't without this change, is where they know the visits a lookout has seen, and the number of visits is 3.
If an evil visits, faking as tplo, and the number of visits was less than 3, there is 0 difference. This will be the case in at minimum 68% of games.
If an evil visits, faking as tplo, and the number of visits was equal to or greater than 3, that evil cannot safely hold claimspace unless they know
who was seen visiting. It would not be an optimal strategy to blind-claim.
(*)I understand we should not build a meta focused on low-elo play, but similarly we cannot build a meta focused only on max-elo play. We must take into consideration possible metas within a range of elos, imo between mid to mid-high, in order to meet a fair medium between skillful play and fun play.