วีดีโอ

8:39 Complementary slackness 9, 10, 11 makes sure that the multipliers correspond only to active constraints (those constraints when they are binding, or =0). In this case, multiplier>0 reflects that this constraint is influencing the optimal solution.

21:46 here it should be since x is strictly preferred than y implies x is weakly preferred than y. Then transitivity gives x is weakly preferred to z.

Hi team, thank you very much for the tutorial session. It is very helpful. I just have one comment about the way you explain the calculation that started at 37:00-the payoff of the Prosecutor. I think that calculation should be called the probability that the designed investigation generates signal "g" which is: P(g)=pi(q|G)*mu_0(G) + pi(g|I)*mu_0(I) = 0.6 (or simply the marginal probability of signal g). From the previous results, we find that given a signal g is observed, the probability of guilty P(g|G) = 0.5. By common sense, the judge accordingly considers the case as Guilty (just like when the Probability of guilty is 0.3, as in the prior, the judge considers a case innocent. So, we achieved the result as mentioned in the paper, the judge will convict 60% of all cases. As the probability of conviction is now 0.6, the payoff of the prosecutor will be 0.6*1+0.4*0=0.6 (the same as you mentioned). I think it is a minor point but the related interpretation should be clear to avoid misunderstanding.

43:18 with hidden info, PH decreases as mentioned, XL increases (note c’ decreases in x)

17:59 bidding bi leads to payoff 0 since lost, less than a positive payoff if bid with vi.

looked forward to this, but couldn't understand through the mumbling

Intuitively, the one deviation property says that a strategy profile of a finite multi-stage extensive-form game with observed actions (perfect info) is a subgame perfect equilibrium (SPE) if and only if there exist no profitable single deviation for each subgame and every player. In simpler terms, if no player can increase their expected payoff by deviating from their original strategy via a single action (in just one stage of the game), then the strategy profile is an SPE. In other words, no player can profit by deviating from the strategy in one period and then reverting to the strategy.

Great video =)

8:45 the convex hull should be the overall convex set enclosing the function.

24:04 Correction: pi(s_1)=0.6, pi(s_2)=0.4