Always use Lab for honey if able

From Fallen London Wiki (Staging)

Action purge is not allowed on comment pages.

Disagree with: "Upconverting Drop of Prisoner's Honey; with Honey obtained from Doing the decent thing... Using Send him out for a little more honey becomes more efficient around 120 Persuasive".

because it fails to take into account actions spent mitigating scandal gain.

If Scandal is removed by Attend a Church Service (the only solo, costless, on-demand scandal reduction I know), at -1.1 Scandal/Action (-2*0.7 + 1*0.3), then gaining 1 Scandal costs 0.909 actions (1/1.1)

Failure at "Send him out for a little more honey" increases Scandal by 1, so costs an additional 0.91 actions.

Failure at "Doing the right thing" increases Scandal by 2, so costs an additional 1.82 actions

At 120 expected honey from "Send" is 0.41*125 honey costing 1 + 0.59*0.909 actions to roll and then clear gained scandal, or ~33 honey/action

At 120 expected honey from "Doing" is 0.72*50 honey, costing 1 + 2*0.28*0.909 actions to roll and then clear gained scandal, or ~24 honey/action

The actual breakpoint (HPA_send == HPA_doin; scandal action cost abbrev Scan) here is when P_send*H_send / (1 + (1 - P_send) * Scan ) = P_doin*H_doin / (1 + 2*(1 - P_doin) * Scan ) where

P_send = scalar * Persuasive / D_send

P_doin = scalar * Persuasive / D_doin

Plugging in (scalar, a; and abbrev) and solving for Persuasive (Pers) we get


(H_s*a*Pers/D_s) / [ 1 + Scan*(1-a*Pers/D_s) ] = (H_d*a*Pers/D_d) / [ 1 + 2*Scan*(1 - a*Pers/D_d) ] <=>

(H_s/D_s)(1 + 2*Scan - 2*Scan*a*Pers/D_d) = (H_d/D_d)*(1 + Scan - Scan*a*Pers/D_s) <=>

(H_s/D_s) + 2*Scan*(H_s/D_s) - 2*Scan*a*Pers*(H_s/D_s)/D_d = (H_d/D_d) + Scan*(H_d/D_d) - Scan*a*Pers*(H_d/D_d)/D_s <=>

Scan*a*(H_d/D_d)*Pers/D_s - 2*Scan*a*Pers*(H_s/D_s)/D_d = H_d/D_d + Scan*(H_d/D_d) - H_s/D_s - 2*Scan*H_s/D_s <=>

Pers*Scan*a*(H_d/D_d - 2*H_s/D_s) = [ (H_d/D_d)*(1 + Scan) - (H_s/D_s)*(1 + 2*Scan) ] <=>

Pers = [ (H_d/D_d)*(1 + Scan) - (H_s/D_s)*(1 + 2*Scan) ] / [ Scan*a*(H_d/D_d - 2*H_s/D_s) ] <=>

Plugging in our values: D_d = 100, D_s = 175, H_d = 50, H_s = 125, a = 0.6, and Scan = 0.909, the Persuasive skill breakpoint between these two is

((50/100)*(1+0.909) - (125/175)*(1+2*0.909)) / (0.909*0.6*(50/100 - 2*125/175) = 2.09 persuasive fuckin what

Use the lab as soon as it's open to you?

Let's check Persuasive = 100 (lowest POSI)

P_s = a*Pers/D_s = 0.6*100/175 = 0.343

Expected honey 0.343*125 = 42.5, costing 1 action to roll and gaining 1*0.657 scandal costing 1*0.657*0.909=0.597 to remove, so 42.5/1.597=26.6 HPA

P_d = a*Pers/D_d = 0.6*100/100 = 0.6

Expected honey 0.6*50 = 30, costing 1 action to roll and gaining 0.4*2 scandal costing 2*0.4*0.909=0.727 actions to remove, so 30/1.727=17 EPA

Unless my initial formulation of Scandal-reduction-EPA is wrong, yeah, use the lab as soon as it's open for you. Significant changes of the the value of Scan in above (action-cheap Scandal removal) make little difference, Scan=0.1 (-10 Scandal/Action) yields a crosspoint of only 5 Persuasive.