I've been pretty busy between crunch at work and prep for my Midnight ACKS game, but there have been a number of really good posts that came up on my radar this week.
First off, Koewn's analysis of ACKS' stealth mechanics is fantastic. I was considering (am still considering, actually) an ACKS assassin that trades plate for something else, and this has reasonably convinced me that they don't need Naturally Stealthy (-1 to enemy surprise rolls) to be effective at sneaking at low levels. Honestly, reading his analysis, I'm pretty convinced that ACKS' stealth system is fine and we've just been Doing It Wrong; I actually kind of like that a 1st level thief is ~18% more likely to surprise a target than a common man. As Koewn points out, this is about on par with the incremental improvement in combat ability that fighters get, with a 1st-level fighter being better but not massively better than a common man. The question remains, though: what to give assassins? I'm thinking Poison Use (some combination of bonus to saves against poison, bonus to proficiency throws to produce poisons, no chance to poison self when using poison), but I haven't set on anything yet.
Second, Trilemma has a pair of solid posts on knowlege and preparation and DM-player bandwidth. The first is a good statement of the general philosophy behind his prior (excellent) posts on useful dungeon description and monstrous effects on terrain, as well as explaining the importance of intel in old-school gaming, the prevalence of monsters with crazy immunities, and (indirectly) why vancian casting was awesome. In a way, it explains a lot of the mechanical oddities of the Old School that fell by the wayside as cultural, playstyle elements of the Old School (which made those mechanical elements sensible) faded due to dilution, and in identifying those cultural elements provides a path forwards/backwards, depending on how you look at it.
Trilemma's second post, on bandwidth, is particularly relevant to me at the moment, as I am engaged in world creation and trying to get data to my players. It's also somewhat interesting because "the bandwidth problem" was something parts of my group have been talking about for years, but never thought to write down. I have the good fortune of having a player this campaign who really likes lore (at one point he explicitly asked me for more lore, and I was taken aback; not a problem I've ever had before!), but my approach is still mostly a combination of a (written lore), c (no lore), and d (players propose reasonable lore). Some but not many things are fixed (written lore), many things that aren't relevant to the game just don't exist (no lore), and if my players propose something that makes sense, it's on me to either find a good a reason it isn't true, or to adopt it. I imagine this is probably true of most campaigns, that they follow hybrid approaches. Still, interesting stuff to think about.
Finally, today from the Hill Cantons, two solid posts on building and running dynamic sandboxes [1, 2]! Since I'm running a dynamic sandbox at the moment, these are extraordinarily relevant to my interests. The first article discusses having a campaign news cycle and dynamic encounter tables, while the second focuses on his Chaos Index world engine and campaign-scale event charts. The campaign news cycle is probably not reasonable for a setting like Midnight, where news is sparse and literacy is sparser, but dynamic encounter tables are something I've been playing with lately. ckutalik proposes a "New Development" slot on one's random encounter tables, where something related to a world event or past PC actions shows up when rolled. I've been thinking about having an encounter queue instead - when the PCs take an action which generates Consequences, those consequences go in a queue. When you roll "queued encounter" on the table, you pick something sensible from the first couple of things in the queue and that's what they find. This means that consequences can linger in the queue for quite a while, but will probably find the PCs eventually. I've also considered having multiple queues on a per-area basis; failing to assassinate the duke might queue a squad of royal guards in his domains, and bounty hunters in each of several nearby domains. Things like that.
From the second post, I could see setting up a Shadow Index, where certain actions increase the grip of the Shadow on the world and that alters things cosmologically (again, ideally I think I'd want per-area or -domain indices, so that some places can be deeply shrouded while others are 'points of light'). I've been running some loosely-defined world engines for various plots that are afoot; when the players decide to engage the ghouls hunting the farmers of Ostergot, the mushroommen of the Monastery Caverns grow and multiply, and when they fight the mushroommen, the ghouls spread. Event charts are something I should consider; my plan currently is to use politics as the primary event-driving factor for the campaign, but the occasional natural disaster or similar would not be amiss. My thought is to set up a network of high-power NPCs who have various relations with each other, and in any given unit of time some may undertake actions (declaration of war, assassination, charm / domination, slander, ...) against others, the consequences of which trickle down to the PCs. Event tables would be a simpler way to handle this, though, with "assassination attempt on X by Y", "raiding of X's domains by Y", and "warfare of X against Y" as table items with fillable variables as sensible. Maybe I'll do that instead.
Anyway, good posts, well worth reading.