In contrast to ACKS, in which spying and other hijinks are usually carried out by the followers of a PC, in AD&D they are carried out by NPCs that are hired, if the PCs don't want to do it on their own in the first place. Gary provides a table for determining the success of a mission, modified by whether the mission is simple, difficult or extraordinary. There is always a chance of discovery, modified by the length of the mission and how well guarded the target is against spying. If the spy is unsuccessful, or if the spy is discovered, you roll on the Spy Failure Table, which determines the fate of the spy. Altogether, I prefer the simplicity of the ACKS system, which takes care of all of these issues with a single die roll (up to capture, when some more rolls are required).
Gary expounds on Thief abilities for the purpose of preventing abuse, which mainly means that he lists the limits and time consequences of Thief abilities - Thieves cannot backstab opponents who are aware of their presence, opening locks takes 1-10 rounds (though usually 1-4), and so on. Gary emphasizes rolling all the dice to determine success or failure in secret so that the player will not know whether his character has successfully hid in shadows or moved silently or whatever. The two abilities he singles out as the most abused are hide in shadows and climbing walls. Hide in shadows is not possible if a character is already being observed, and a hiding character is still susceptible to being detected as if he were invisible. Gary provides a matrix that lays out how quickly a Thief can climb, from non-slippery to slippery surfaces and from smooth to very rough surfaces. Gary states that most dungeon walls are on the rougher side but are also slippery from dampness and slime growth.
Thieves and Assassins Setting Traps
Thieves and Assassins get a chance to set traps that is based off of the Thief's chance to detect traps (Assassins get a chance as if they were Thieves of two levels higher than their actual level). They must have access to the components of the traps, having to hire a specialist to provide special components, and the player has to provide the DM with a drawing of the trap. Even if the PC sets the trap successfully, there is a chance (the inverse of the chance to set the trap correctly) that the trap will accidentally go off and hurt the PC while he is setting it. This all seems over-complicated to me; I imagine that this level of work for setting one trap will discourage players from setting any traps at all, which is the exact opposite of what I'd like in my games. I do like the idea of PCs setting traps, though, so I'll probably try to inform players of that possibility in my games in the future, but I'll probably just require a verbal description of the trap similar to, but with fewer requirements than, the ones I use for my Tuesday Traps series and handwave the chances of failure, at least unless the trap is being set up in a hurry.
Assassination Experience Points
On top of the normal XP gained for killing the target of an assassination, an Assassin gets XP for having assassinated the target, based on the level of the target and modified the Assassin's level and the difficulty of the job. The Assassin also receives XP for the fee he is paid, meaning that an Assassin gets XP in three different ways every time he assassinates someone! This seems to incentivize assassination to a degree I'm not sure I want in my game; I do suppose that if Assassins don't make up the majority of the party, the rest of the party probably won't go for assassination missions every session, and if the majority of the party is made up of Assassins then it's already basically an assassination campaign, so I needn't worry too much. Still, what feels to me like disproportionately rewarding assassination rubs me the wrong way.