ebering wrote:If you want to read more about why the devteam is in agreement that shafts are good dpeg wrote a great post years ago that I agree with. The short of it is: shafts increase the probability of a tense memorable random encounter in which skill makes a real difference; such encounters are a goal of crawl development.
I had a look through dpeg's posts back in 2016 on shafts, I don't know if those were the ones you were referencing, but I understand the basic idea behind his philosophy: Shafting adds another level of randomness and difficulty to a player's exploring. I also get where you're coming from with your interpretation of that, insofar as it adds tension to the game by increasing the stakes. I kind of see your point about skill, even though I don't necessarily agree with it completely. Regardless, that I do think a shaft trap is a cool concept, it just needs very fine tuning to be successful.
I think that the current version of shafting is needlessly punishing early on. I think it limits the incentive to fully explore a floor and has an exponentially higher impact for players on early floors.
So the big difference between being shafted on floors 1-5 and being shafted at any other point in the dungeon is resources. Getting shafted late means I've already IDed the majority of the scrolls/potions, I have reasonable enough items to have made it to that point in the dungeon, and higher XL translates to a higher likelihood of better rolls. Getting shafted early means I've got unIDed scrolls/potions or even few at all, barebones equipment, and I'm rolling garbage.
So, in that early game situation, what skill is really being reasonably tested? What can a new player really learn from an early shaft experience? About the only answers I can come up with is the ability to get lucky on a random potion chug/scroll read and pressing the "o," button is bad.
My proposal is that the likelihood of being shafted should follow a flattened sigmoidal curve, or something akin to one. It should be zero for the first floor or two, increase and then eventually plateau. I'd add the additional factor that the likelihood of a deeper shaft should follow that curve. For example, using arbitrary numbers:
A trap on floor 4: 66% chance of 1 floor shaft, 33% 2 floor, 1% 3 floor
A trap on floor 8: 30% chance of 1 floor shaft, 30% 2 floor, 30% 3 floor, 10% 4 floor
This means that shafts have an effective window where they have the most impact on a player and the highest chance of actually testing a player's skill (mid-game). This also means that earlier players are punished less severely, which I think fosters the ability to learn from easier shafting experiences so they can be managed later in the game when they are more prevalent. This is pretty in line with your goals of shafting based on statistical analysis. I think the only point we really differ on is how often shafting should occur (and how severe it is) in the early game.