OK, here’s my attempt at math for chess. This is all rather suspect though, because determining from the wiki what constitutes an action is very unclear for social stuff.
I think a full game requires 5 actions on both sides: 1 to send/accept the invitation, 1 for the first move (which gives no rewards), 2 for the regular moves, and 1 to end the game.
Assuming everyone does standard intellect moves, and that the same person loses the last two moves (and thus the whole match), the winner gains +6cp Watchful, +4cp Making Waves, and 1 Sudden Insight, while the loser gains +55cp Watchful and 4 Sudden Insights. Assuming the 1-70 range on trading Sudden Insights is correct, 1 Sudden Insight and .2 Actions -> +.2cp Persuasive, +.2cp Dangerous and +7.1cp Watchful.
So, after conversion, the winner spent 5.2 Actions for +13.1cp Watchful, +4cp Making Waves, +.2cp Persuasive and +.2cp Dangerous, with ~2.52 cp/A of Watchful gain. The loser spent 5.8 Actions for +83.4cp Watchful, +.8cp Persuasive and +.8cp Dangerous, with ~14.38 cp/A of Watchful gain. If you assume both players are around the same level (or rapidly catch up to the same level), they would start winning at the same rates, both averaging to 8.45 cp/A.
This seems to be significantly better than any other with no pre-reqs.
Edit: Here’s my math for the Porters/Chicanery loop:
Porters cost (1/x) Actions and (25/x) Cryptic Clue -> +4cp Investigating, 1 Sudden Insight, 108 Whispered Hint, +f(x) cp Watchful and +(1/x - 1) cp Nightmares, where "x" is the success rate in [0,1] and f(x) is a complicated function, but will generally be either 1.9 or 1 for the interesting ranges of x.
Raid a Message Drop should be tried ASAP for farming secrets, but when going for Watchful it’s better to reduce actions and take the sure thing. It also makes the math easier. A sure success requires level 8, which is 36cp or 9 runs of Porters. The other costs are 1 action -> +2cp Shadowy, +270 Cryptic Clue, or in terms of Porters: 1/9 action -> +2/9 cp Shadowy, +30 Cryptic Clue.
Putting it together, plus the conversion: (1/x + 1/9 + 1/5) action -> (30 - 25/x) Cryptic Clue, 108 Whispered Hint, +(f(x) + 71/10) cp Watchful, +(1/x - 1) cp Nightmares, +1/5cp Persuasive, +1/5cp Dangerous.
The cp/A for watchful is (f(x) + 71/10)/(1/x + 14/45) which is… an ugly function. It has a local maximum of ~6.18 at x=1 (Watchful 180), dropping to 5.75 at x=.91 (Watchful 164) and then jumping to ~6.36 at x=9.05 (Watchful 163) and continuing to decrease from there. However, as the success chance decreases the overall cp/A is hurt by having to deal with Nightmares, which this doesn’t account for.
There’s also the Echo/A gain of exactly .9 when x=1, which isn’t too shabby. And that can be greatly improved by changing the Raid strategy. So although this isn’t as good as chess for pure Watchful gain, and it requires being almost maxed on Watchful already to be most practical, I can see it being a better hybrid strategy in many circumstances.
edited by d0sboots on 2/12/2020