3.1.0: Is it me?

[NeoWizard]

I noticed that it is far easier to dive in 3.1.0, is is me, RNG, or 310?

Anyone else have the same conclusion?

[Pete Mack]

It's not just you. The huge change in drop quality for out of depth weak monsters is part of what does it. Also, the chances of finding ego items on the floor is much higher, as floor drops havent been reduced, but their quality has gone up.


With the current drop quality, a bunch of snagas at 4500' has essentially the same odds of dropping a Ring of Power as a Great Wyrm of Law.

[hugorune]

Maybe a game option is needed to choose between basing drops off dungeon level, monster level, or a 50/50 mix?

[takkaria]

Maybe a game option is needed to choose between basing drops off dungeon level, monster level, or a 50/50 mix?

Whilst I see where you're coming from, game options really aren't the way to fix game balance issues. Moving back to the 50/50 mix looks like a fair option at this point, though I don't like it much myself.

[hugorune]

Whilst I see where you're coming from, game options really aren't the way to fix game balance issues. Moving back to the 50/50 mix looks like a fair option at this point, though I don't like it much myself.

The 50/50 mix sounds like the way I'd prefer it as well - although its a nerf to the dive strategy that I don't tend to use, so take that with a grain of salt. What was the reason it was changed?

[takkaria]

The 50/50 mix sounds like the way I'd prefer it as well - although its a nerf to the dive strategy that I don't tend to use, so take that with a grain of salt. What was the reason it was changed?

There was some discussion on r.g.r.a about it, mainly to do with removing the junk problem. The idea was that if monsters at dlev90 are dropping drops of level 50, then they are very likely to be junk.

[CunningGabe]

While it is important to prevent too much junk from being generated, it does seem a little cheesy that at any dungeon depth, your best bet is to just fight the easiest monsters that drop something. Meanwhile, as we've noticed, some of the nice "staple" items aren't dropping as much because of all the other good items crowding them out. I think it's important for the monster level to be part of the calculation, though it doesn't have to be a straight average of mlev and dlev. Other possibilities:

1. 1/4 * mlev + 3/4 * dlev. This still makes you work a little to get the best items.

2. (200-dlev)/400 * mlev + (200+dlev)/400 * dlev. In other words, at early dungeon levels, you are almost taking the straight average of mlev and dlev; near dungeon level 100, you are using the above formula. Other linear interpolations are possible as well, and help grant a healthy mix of items at earlier dungeon levels, while keeping the later ones from being clogged with junk.

I'd say #1 (or another weighted average) would be a good first try.

Another tactic would be to prevent snagas from appearing on DL 90. If I remember correctly, when generating a dungeon and populating it with monsters, the code generates a few monsters at a time and keeps the deepest. What if it wouldn't even consider monsters that were some number of levels shallower (either a fixed constant or different numbers from every monster)? Maybe we never place a monster more than 50 levels deeper than its native depth. That removes some of the cheesiness of killing easy monsters for great loot. Would this make the deepest levels too hard? (I've not been past DL50 very often).

[NeoWizard]

1. 1/4 * mlev + 3/4 * dlev. This still makes you work a little to get the best items.

I second this one. Maybe give it a try for a while!

[zaimoni]

Another tactic would be to prevent snagas from appearing on DL 90. If I remember correctly, when generating a dungeon and populating it with monsters, the code generates a few monsters at a time and keeps the deepest. What if it wouldn't even consider monsters that were some number of levels shallower (either a fixed constant or different numbers from every monster)?

I've been thinking about how to estimate monster effectiveness against the player, and using that to skew monster generation. (That is, make the weak monsters not be considered during intra-level spawning because they're empirically "harmless".) Sort of a continually rebalancing monster list.

[Nick]

[QUOTE=CunningGabe;13987]Another tactic would be to prevent snagas from appearing on DL 90. If I remember correctly, when generating a dungeon and populating it with monsters, the code generates a few monsters at a time and keeps the deepest. What if it wouldn't even consider monsters that were some number of levels shallower (either a fixed constant or different numbers from every monster)? /QUOTE]

O does a similar thing: if level 30 is being generated, the probabilities of monsters shallower than level 20 are divided by 4, and then the probabilities of monsters shallower than level 10 are divided by 4 again.

[Daniel Fishman]

Another possibility would be to make overly shallow monsters drop less items in total.

So for something with DROP_(20*N) (drop an item (20*N)% of the time) it would actually drop an item
(10*N) * (tanh(2 - (4*max(0,min(dlevel-mlevel,60)))/(60)) + 1) % of the time.

This gives, for DROP_20:
Levels OOD --- Drop percentage
negative ------ 19.6
0 -------------- 19.6
10 ------------- 18.7
20 ------------- 15.8
30 ------------- 10.0
40 ------------- 4.2
50 ------------- 1.3
60+ ------------ 0.4

The formula would probably need tweaking (and quite possibly a different base function to tanh as well, simplifying it a bit) but it would certainly be interesting to experiment with.

DROP_1 and DROP_2 would maybe use 60 for the base chance (which is doubled, so nearly 120% ie 100% indepth, and still 100% until about 18 levels OOD, but still decreasing to 2.2% at 60 levels OOD. Not sure whether uniques would be affected by this or not.

[Pete Mack]

I still think the tried and true (mlev+dlev)/2 is a pretty good bet. Something has to drop stat potions deep in the dungeon, and it might as well be 'T', 'o' and novice 'p'.

I cloned 50 death quasits and got 0 healing of any type and 1 mass banishment that was destroyed by a plasma vortex. Also, infinite ego items, Holhenneth, and Mormegil.

[buzzkill]

Haven't put much thought into this, but...

Since Angband already keeps track of how many of each monster you have killed, why not use that info during dungeon population to make monsters that you haven't been killing (avoiding/or simply not seeing) more common. That way, going after the same 'good drop/great drop' monsters all the time will lead to less of them being generated in the end.

[Donald Jonker]

Haven't put much thought into this, but...

Since Angband already keeps track of how many of each monster you have killed, why not use that info during dungeon population to make monsters that you haven't been killing (avoiding/or simply not seeing) more common. That way, going after the same 'good drop/great drop' monsters all the time will lead to less of them being generated in the end.

Frankly, I'm tickled pink. I think this has brilliant implications written all over it.

-Kill enough water hounds (or other irritating monsters) and eventually you won't be seeing them any more.

-Do you take on those orc/troll pits now for some needed xp and extra $$, or do you leave them for later when they'll give better drops?

-You could also skew things more drastically so that pacifistic diving gives you easier deeper dungeon levels... but then you'll have to face Morgoth in a relatively weakened state. Lots of stuff to play around with here.

Maybe not an ideal framework for V (or at least not as the sole framework for V), but this could be the strong foundation for an excellent new variant.. "Angband Arena"... or something.

[Jungle_Boy]

Haven't put much thought into this, but...

Since Angband already keeps track of how many of each monster you have killed, why not use that info during dungeon population to make monsters that you haven't been killing (avoiding/or simply not seeing) more common. That way, going after the same 'good drop/great drop' monsters all the time will lead to less of them being generated in the end.

This sounds like the beginning of an awesome idea actually. I like it!