Mr. decoy goes to Nogrod (Sil AAR/strategy talk)

[half]

Which brings up a question: What is the defense against entrancement?

Entrancement is an opposed Will check, thus it depends on the Will of the monster, which is typically higher at deeper levels. A Spider of Gorgoroth has Will 11 and you have Will 12. Since it is acting, it needs to beat your score on 1d10 + 11 vs your 1d10 + 12. It thus has a pretty good chance of entrancing you if it hits and does net damage. A higher Will would help (it is not impossible to get Will 20 or so, but rarely worth it in my view). Free action, like all such abilities in Sil gives a +10 to the skill check, guaranteeing you safety in this case.

My advice is that if you don't have Will that high, you should immediately run from foes like that spider. Luckily for you, they are territorial and don't chase you. The same is true for dragons and some other creatures. We are prepared to make monsters pretty deadly if they are territorial, as you can generally choose not to fight them if you wish.

Sorry about the one turn death though!

[decoy]

Yes, blocking only works when you pass your turn (z or 5). Without combination with other abilities it is almost useless (though not always useless!).

Hey, I'm not above standing behind a shieldwall regenning health.

It is best used in combination with a kite shield and some other abilities. An additional 1d6 protection is quite amazing, and possibly worth the same as one or two points of Con. The thing is that you need to be doing something useful with the turn, such as focusing for an attack, singing, riposting, dominating your zone of control etc. It is, of course, also useful as a pre-req for the fantastic Heavy Armour Use.

Righto. So if you want to hunker down and play "defensive offense" it's a good way to go. I must admit, I think I prefer flanking. Thanks for the suggestion!

A higher Will would help (it is not impossible to get Will 20 or so, but rarely worth it in my view). Free action, like all such abilities in Sil gives a +10 to the skill check, guaranteeing you safety in this case.

Yeah, Will that high seems to be a luxury with points best spent elsewhere. But Free Action was a stupid omission on my part.

My advice is that if you don't have Will that high, you should immediately run from foes like that spider. Luckily for you, they are territorial and don't chase you. The same is true for dragons and some other creatures. We are prepared to make monsters pretty deadly if they are territorial, as you can generally choose not to fight them if you wish.

Seems like a fair enough approach to me. Cuz like, dragons and other assorted Servants of Morgoth are deadly and stuff. I think you have ruined Vanilla Angband for me forever, by the way. Every time I try to go back to it I get bored after 100 turns or something.

Sorry about the one turn death though!

No worries! It happens.

I don't know if you keep stats on your testing deaths, but with my stealth builds I found that 750' was a "critical depth" where I lost a lot of characters. I don't think I kept track of how many, but by contrast, if I could live through the 700'-750' range fairly comfortably I don't think I lost many after that. This is the deepest I've ever been with this build, so my sample size is 1 in this case. If I graphed "number of deaths per floor" normalized by the number of characters I got to that floor, I think it would show some particularly dangerous zones, including 700'-750'. (Can you tell I'm a scientist by trade?)

Danger in this case probably made the problem worse, but I could not leave such an awesome sword behind. It definitely helped me lay the beatdown on some beasties.

I can also see now why you dual-wielded the meteoric blades. You should have taken Parry...+9 defense (rather than +6) and +6 offense at the expense of 1d6 protection and I think a -1 to offense? I'd probably take that action.

[Scatha]

I don't know if you keep stats on your testing deaths, but with my stealth builds I found that 750' was a "critical depth" where I lost a lot of characters. I don't think I kept track of how many, but by contrast, if I could live through the 700'-750' range fairly comfortably I don't think I lost many after that. This is the deepest I've ever been with this build, so my sample size is 1 in this case. If I graphed "number of deaths per floor" normalized by the number of characters I got to that floor, I think it would show some particularly dangerous zones, including 700'-750'. (Can you tell I'm a scientist by trade?)

I have been keeping a vague eye on deaths by depth on the ladder. There are a couple of issues here: more characters see the earlier depths, but on the other hand they are more likely to be posted if they get deeper. Below 300', there's a surprising consistency of reported deaths at each depth. For a while there weren't any deaths reported at 850 or 900', which made me suspect something was up (perhaps a particularly dangerous area at 750-800ft such that characters who survived that tended to be strong ones), but this effect has mostly disappeared so I think our sample size was just too small. Interestingly 750' is currently under-represented. The throne room is more dangerous than normal, but I think that's good.

All that said, I think it very likely that for many particular character builds and playstyles there are markedly dangerous zones. But as they don't seem to be clumping too much across total characters, I'm not sure there's an issue here. Still, I have almost no more data than you (as the number of games since release is much larger than the playtesting ones before hand), so if you think you can produce some interesting analysis then please go for it -- and if there are demonstrable trends we'd be delighted to take them into account when thinking about balancing.

[half]

There is an interesting question for permadeath games of what the probability of death at each depth should look like as the depth increases. Should it be constant? Increasing? Decreasing? I'm not sure, but it probably should be roughly monotonic and continuous (with a possible exception at the deepest level, and in Sil, at the Gates).

I'm also unsure whether probability of death at a given depth is the right variable, as opposed to probability of death per turn at different depths. Sil allows more time at each level as you descend if you go at the minimum pace and I think many people thus spend more turns on the deeper levels.

[decoy]

I have been keeping a vague eye on deaths by depth on the ladder. There are a couple of issues here: more characters see the earlier depths, but on the other hand they are more likely to be posted if they get deeper. Below 300', there's a surprising consistency of reported deaths at each depth. For a while there weren't any deaths reported at 850 or 900', which made me suspect something was up (perhaps a particularly dangerous area at 750-800ft such that characters who survived that tended to be strong ones), but this effect has mostly disappeared so I think our sample size was just too small. Interestingly 750' is currently under-represented. The throne room is more dangerous than normal, but I think that's good.

Yeah, I think the ladder has some pretty serious sampling issues, though. Reporting bias, as you say.

Also, is it necessarily bad if there are (possibly) fewer deaths at the lower levels? You guys worked hard creating the game, and people spent a lot of time playing it if they're running around at 850'-950'. Given that the game is on a timer, it seems reasonable to give players a bit of time to enjoy the powerful character they've built.

All that said, I think it very likely that for many particular character builds and playstyles there are markedly dangerous zones. But as they don't seem to be clumping too much across total characters, I'm not sure there's an issue here. Still, I have almost no more data than you (as the number of games since release is much larger than the playtesting ones before hand), so if you think you can produce some interesting analysis then please go for it -- and if there are demonstrable trends we'd be delighted to take them into account when thinking about balancing.

I suspect you are right here. For example, for me, the stealth game is easier once I have some key abilities, but down at 700'-750' depth one is still vulnerable to getting surrounded in hallways by fast-movers such as Greater Werewolves and Raukar (even if one has Exchange Places, as they hit hard and travel in packs). But if I can live past that, I can basically "complete" the build by upgrading Song to Throne Room levels, enabling me to make them snooze. Ergo, for that build and my style, 700'-750' is probably just a natural confluence of amount of experience available (which is just pretty straightforward arithmetic) and necessary attributes.

Similarly, it's possible that my death spots for this build are 350', 450'-550', and 750', but I don't know if 750' is a fluke or not. I had several tight squeezes at shallower depths. Certainly without Free Action, See Invisible, and resistance to Fear/Blindness (well, okay, the Lantern of True Sight would help with two of those) life might have gotten very interesting lower down. Still, the build was almost complete. Time will tell, I suppose.

[Derakon]

Thinking about likelihood of death as a function of depth reminds me of Super Meat Boy. If you haven't played it, it's a platforming game with very short levels; typically you could think of a level having at most 10 obstacles that would need to be dodged/navigated past/ridden/etc. It's a very long level that takes even 30 seconds to beat. It's also a very hard game, though, so your actual time spent trying to beat each level is significantly longer. Moreover, when you beat a level, you get a replay that shows all of your attempts playing out simultaneously, which gives you a good idea of where most of your corpses ended up.

So I found myself thinking "Okay, my odds of passing obstacle 1 are 90% now, but my odds of passing obstacle 2 are only 50% (it's really tricky). Having passed #2, I'd say my odds of making it to the obstacle that consistently kills me are probably 66%. So how hard should that obstacle be to be fair?" Of course the numbers here are made up, but they aren't that unrealistic, and the upshot is that I'd have only a .15% chance of making it to the part that's actually still "giving me trouble" and thus presents the most interesting challenge. Most of my time on any given level is thus spent replaying sections that I'd already "solved". Fortunately the replaying was brief because the levels as a whole were brief.

Sil may be much shorter and harder than Angband, but it nonetheless has a much bigger time commitment and lower instantaneous challenge level than one level of Super Meat Boy does (of course, the type of challenge is also completely different). But I think a similar thought process applies here.

1) The difficulty of beating the game can be roughly described as the product of the difficulty of passing each section.
2) Any given experienced player will have far more practice passing the early sections than the late ones, because the early ones must be passed before the late ones can be attempted. They won't have a 100% success rate but their rate will be much higher than that of a newbie.
3) Interest in a game depends roughly on challenge and novelty. You need a minimum level of challenge to keep the player from zoning out; a minimum level of novelty to keep them from getting bored.
4) The early game will eventually become low on challenge (due to practice); the late game inherently has strong novelty (because the player sees it less often).
5) Ramping up the early difficulty to keep veterans interested early on risks alienating newbies (though you may not care). Ramping up the late game difficulty serves relatively little point so long as a base level of challenge is maintained to keep the player interested.
6) Adding novelty requires significant dev effort.

I don't have any answers for you (I'm avoiding playing Sil for now because I don't want to get distracted from other things), but I thought perhaps this perspective would be interesting. Vanilla of course "solves" the early game issue by making it flyover territory -- an experienced player can basically skip it while freefalling through the dungeon. This does kind of work, but it's not without flaws. Meanwhile, the lategame isn't a walk in the park, but it's also typically not as prone to wild swings as the early game is -- the player has multiple contingencies and far more control of the situation in the late game than they do in the early game. Again, mostly works, not without flaws.

[Fendell Orcbane]

I have been keeping a vague eye on deaths by depth on the ladder. There are a couple of issues here: more characters see the earlier depths, but on the other hand they are more likely to be posted if they get deeper. Below 300', there's a surprising consistency of reported deaths at each depth. For a while there weren't any deaths reported at 850 or 900', which made me suspect something was up (perhaps a particularly dangerous area at 750-800ft such that characters who survived that tended to be strong ones), but this effect has mostly disappeared so I think our sample size was just too small. Interestingly 750' is currently under-represented. The throne room is more dangerous than normal, but I think that's good.

All that said, I think it very likely that for many particular character builds and playstyles there are markedly dangerous zones. But as they don't seem to be clumping too much across total characters, I'm not sure there's an issue here. Still, I have almost no more data than you (as the number of games since release is much larger than the playtesting ones before hand), so if you think you can produce some interesting analysis then please go for it -- and if there are demonstrable trends we'd be delighted to take them into account when thinking about balancing.

I won't post a character unless they make it past 600' or so. What I have found is that once I have Exchange and Vanish, I can pretty much make it to the Throne room. Hell I've even holed up and healed in the Throne Room with Vanish. I also usually have Sprinting so that if I can't run faster than Cats, I will at least be able to get far enough away so that they lose me.
After I escape the Throne Room life is actually easy...I can kill anything that spots me or run away from it. I would say that for myself the danger level goes down after the Throne Room. BUt then again I have only used Stealth builds.

[decoy]

I won't post a character unless they make it past 600' or so.

In truth, looking at the present ladder and given that I already have some winners, I have decided that I'm not even going to bother posting anything to the ladder unless I make it to the Throne Room. With the present ladder pushing 350, why bother? Why fill the ladder with "Guess what, guys, I died again" posts (BTW, I die a lot)? Hence the reason I posted my dumps here but not on the ladder itself. If I make it to the Throne Room, that's my personal minimum for the actual ladder at the moment. Unless, I guess, something particularly interesting or noteworthy happens. So yeah, I think there is substantial sampling bias in the ladder.

The early game will eventually become low on challenge (due to practice); the late game inherently has strong novelty (because the player sees it less often).

I really appreciate the depth of your post and I think it contains a lot of interesting material, much of which I tend to agree with. I think, though, that there are simply so many workable permutations in Sil that it would take a long, long time to exhaust the novelty aspect. Even with a particular starting setup, each game can be vastly different. A difference of a point here or a point there in stats can have MAJOR unanticipated effects.

My own experience is a perfect case in point here: I am happy to have three characters in the green at the moment, and one nearly there dead by my own stupidity. Ostensibly, all of them except my first ever are nearly the same build (pure stealth/song explorer build), and they represent three of the four available races.

I will be the first to admit that despite the similarities, each of them played vastly differently because of different stats, skills, starting abilities, etc. The general idea was the same, but the specifics were totally different.

None of them were combat builds, and only two of them (one of them the dead one) engaged in anything resembling combat. Now, with this character, I am going with a pure combat/smithing build which nonetheless plays completely differently from a Noldor character of ostensibly similar design. With my latest iteration I realized that it's like playing a totally different game.

Given all of the different possible permutations (most of which I have yet to discover), I find it hard to believe that the novelty in Sil will be exhausted by most players any time soon. A single lucky find at the initial levels can completely alter the course of a game.