[Sil] Do you clear levels?

[BlueFish]

I was unaware of this one. Do you know the exact no. by any chance? This seems to me a bigger issue then the artifact settings in that if you know how many turns to the next guaranteed forge you know to clear that level where as if you don't there's a chance you dive past it & reset the counter.

Yes if you're playing a Smithing character this is something you should keep in mind. Smithing characters are much more motivated, in many situations, to fully explore levels than other builds are. Especially for that second forge, I suspect based on my own games that it is nearly always of the "forced" variety, and I'm always aware of the chance I'll get a forced one when I enter a new level.

Edit: Is it a sliding probability? If so it would be nice to know the chances to know when to slow diving & look for a forge & when to grab song of aule on a singer smith.

The chance slides from 0 to 100%, from 2000 to 5000 turns since the last forge was generated. So if I'm within that initial 2000 turns since a forge was generated, I feel much more free to leave levels unexplored, knowing I'm unlikely to be missing a forge.

[Scatha]

The chance slides from 0 to 100%, from 2000 to 5000 turns since the last forge was generated. So if I'm within that initial 2000 turns since a forge was generated, I feel much more free to leave levels unexplored, knowing I'm unlikely to be missing a forge.

Yes, that's the mechanic. The idea of the sliding probability was to ensure a certain regularity of forges without making it useful for the player to count turns or do anything to manipulate the turn-count levels are generated at.

From your description it sounds like we may not have gone far enough to stop that being useful. What would it take to stop you turn-counting? If, say, it went from 0% to 100% over the interval (0, 8000) turns instead of (2000, 5000), would that be enough?

[BlueFish]

I would always be aware of the number of turns since the last forge was generated. Or at least since the last forge I saw was generated. I would think it would be a more elegant mechanic to begin the chance for a forced forge at zero turns since the previous one was generated. In that case, as a smith, I would be very likely to explore levels fully in virtually all cases. But I would still be counting turns, just for reference if I have a difficult decision to make whether to leave a level.

[taptap]

I would always be aware of the number of turns since the last forge was generated. Or at least since the last forge I saw was generated. I would think it would be a more elegant mechanic to begin the chance for a forced forge at zero turns since the previous one was generated. In that case, as a smith, I would be very likely to explore levels fully in virtually all cases. But I would still be counting turns, just for reference if I have a difficult decision to make whether to leave a level.

Imho, it would still be fine to dive after a forge.

Am I correct to assume that the forges in greater vaults (Chambers of Thu, Cat Fortress) are not included in this count?

[evilmike]

Yes, that's the mechanic. The idea of the sliding probability was to ensure a certain regularity of forges without making it useful for the player to count turns or do anything to manipulate the turn-count levels are generated at.

From your description it sounds like we may not have gone far enough to stop that being useful. What would it take to stop you turn-counting? If, say, it went from 0% to 100% over the interval (0, 8000) turns instead of (2000, 5000), would that be enough?

I don't play smithing characters much, but I tend to keep a mental note of how many turns it's been since I found one, because of the way they generate. I'd do that for any system like this... going from 0 to 8000 would probably be better but it still means you want to try to avoid abandoning levels after a certain amount of time elapses (or to put it another way, once you find a forge you can dive/abandon levels more freely for a while). There's something weird and game-y about it to me, but forges are really valuable and it sucks to miss one when it means you probably won't find another for a couple thousand turns. If the probability started increasing from 0 turns, you'd still have a good amount of time where the odds are very low of getting one on a level.

My own preference is for a system where each level (or room I guess?) just has a flat probability of getting a forge. Maybe adjust the quality though so deeper forges are more likely to have extra uses or are enchanted, and early ones tend to have fewer, or something like that anyway. I understand a system like this has its own problems, but as a player it would feel a lot cleaner. I believe an old version of sil had something like this?

[taptap]

I believe an old version of sil had something like this?

And lots of complaints about it, afair.

[half]

And lots of complaints about it, afair.

Yes, that's right.

We used to have the clean 'memoryless' system, but this produces a binomial distribution, and if you want a mean of, say, 5 discovered forges in a complete game, then the relative variance seemed to be too high. i.e. it was too easy to have a game with only one forge after the first one (happened to me with a Nogrod 'smith' character), and just adding more forges made them too common. The gap sizes in time between forges were also an issue in the memoryless system.

[Psi]

The old way made forges too common for my playstyle - or am I being paranoid

I liked the old way

[Scatha]

There were a couple of extra disadvantages of the old system:

(1) Due to the way dungeon generation worked, forges were much more likely in the depths than further up.

(2) Some characters would explore several times as much of the dungeon as others, and find correspondingly as many forges (or more, if their extra exploration time was very deep).

Both of these exacerbated the balance issue where we found some characters found too many forges while others found almost none.

Now, we could fix (1) in a memoryless system (for instance have an independent 25% chance of any level containing a forge, independent of depth).

It's almost impossible to do anything about (2) with memoryless level generation. We do a little bit by making forging use turns, so those using many forges get less time to find even more.

I think it's possible that just fixing (1) would give you a memoryless system where this wasn't too much of a problem. It's not obvious to me that it would be enough, though, and playtesting this is a somewhat painful process, since to say that it's not fixed will require a few frustrated smiths played most of the way through the game.

If we had a good idea of exactly how much floor space different characters explored we could model forge-finding as a Poisson process and consider directly how likely very high or low numbers were ...

[BlueFish]

There were a couple of extra disadvantages of the old system:

(1) Due to the way dungeon generation worked, forges were much more likely in the depths than further up.

I'm quite sure this is still the case... I thought it was intentional. Seems reasonable to me.

[taptap]

I'm quite sure this is still the case... I thought it was intentional. Seems reasonable to me.

Feels the same to me, but I usually spend about three times the time below 500 ft. than I do above.

[Psi]

Feels the same to me, but I usually spend about three times the time below 500 ft. than I do above.

There are still more forges deeper than shallower - however they are much rarer per level than before.

[Scatha]

I'm quite sure this is still the case... I thought it was intentional. Seems reasonable to me.

Sorry, yes, there's still a factor which works like this. However the automatic forge generation is a component which is independent of depth.

[BlueFish]

I'm becoming more convinced that a memoryless system which give a flat percent chance for any level to contain a forge would be for the best.

The most important thing it alleviates is this artificial, gamey motivation smiths have to fully explore levels and to count turns since their last forge.

I wouldn't want forges to be too common, but even if they were, I don't see how that unbalances a lucky smith who gets a forge every level. You'd just end up skipping them anyway. A forge in itself does not add power to a character. You need something you want to smith, and you need to want to trade the turns for it.

The only reason I wouldn't want them to be too common is because I'd miss the excitement of finding a semi-rare thing. Also I suppose you wouldn't want enchanted forges to be too common, and their frequency would scale with overall forge frequency.

With the addition of chasms in 1.2, it makes that desire/need to fully explore levels even more potentially frustrating than it was before. No other character type has that need. I think it would be good to bring all character types in line for something that has an important effect on turn count and the frustration of being chased off levels or not being able to explore them due to obstacles.

I wish I could compile the code; I'd love to try this out for myself. Maybe starting with a flat 20% chance of a forced forge, each level. Combined of course with whatever normal random chance a level has for a forge. It would probably be for the best to make all forced forges normal rather than enchanted/unique. Maybe even force them to be 3 uses. Let only "random" forges be possibly enchanted and with random charges.

[locus]

If forges are infinitely common, you can smith artifacts to go in all your slots after building a perfect set of forging gear. If they're rare, you have to decide how far to press your luck and prioritize your slots.