Constructing a set of randarts

[Nick]

I'm currently working on a better way of generating the randart set as a whole, and could do with a sanity check on my thinking.

The current method is approximately:

1. Measure the power of all the standard artifacts (for these purposes I'm ignoring how this is done)
2. Go through the standard artifacts one by one doing the following:
   1. Choose an item to base the randomised artifact on, based roughly on the standart distribution (although the large number of swords and polearms is reduced to make a viable set more likely)
   2. If the power of this artifact is too great for an item of this tval, try (a) again
   3. If the power of the base item is too close to the artifact power (so the result would have few properties beyond those of the base item) or too low (so it would need too many added properties), try (a) again
   4. Make an artifact on the base item
3. If the resulting list fails to reach minimum quantities of tvals:
   

   "bow",		4)
   "hafted",	5)
   "polearm",	5)
   "sword",		5)
   "boots",		4)
   "gloves",	4)
   "helm",		3)
   "crown",		1)
   "shield",	4)
   "cloak",		4)
   "soft armor",	2)
   "hard armor",	2)
   "dragon armor",	1)
   "light",		3)
   "amulet",	3)
   "ring",		3)

   return to (2) and do the whole set again.

That last one is the killer, because if you have a relatively unlikely tval (boots, say), it's really easy to end up with less than 4 and you start again.

My proposed new system is approximately this:

1. Measure all the standarts
2. Set a minimum number for each tval (currently they're hard-coded as above - I'd prefer saying something like at least 80% of the number in the standart set)
3. Set appropriate power bounds for each tval (something like no more than 10% less than the minimum for that tval in the standarts, and no more than 10% more than the maximum)
4. Pick upfront, for each tval, a collection of artifacts to scramble which fit within the power bounds
5. Allocate the remaining artifacts to tvals randomly according to the standart distribution (enforcing the bounds)
6. For each tval, pick appropriate base items for each artifact (these should exist because of the way the bounds were set) and scramble them

I think this will be efficient. One consequence I can see is that all randart DSMs will be really good, being no worse than 10% less than the power of the worst of the standart ones (probably Razorback) - I don't think that's a big problem.

Anyone see any bad consequences, or flaws in my thinking?

[Pete Mack]

This is a much better idea than the current system, so it sounds like an excellent start. Of course it may need balancing, but not as much as the existing system did.

[Derakon]

Seems worth trying. In general I would prefer to implement filtering as constraining the set to be selected from, rather than starting over if the random process doesn't pass the filter.

[Nick]

Just for info - probably mainly for me:

Power for tval bow:          min 144, max 271, avg 197
Power for tval digger:       min 256, max 261, avg 258
Power for tval hafted:       min 127, max 551, avg 250
Power for tval polearm:      min  56, max 372, avg 235
Power for tval sword:        min  85, max 489, avg 206
Power for tval boots:        min 101, max 365, avg 205
Power for tval gloves:       min  44, max 228, avg 115
Power for tval helm:         min  60, max 293, avg 150
Power for tval crown:        min  18, max 286, avg 184
Power for tval shield:       min 137, max 221, avg 176
Power for tval cloak:        min  70, max 241, avg 137
Power for tval soft armor:   min 162, max 237, avg 189
Power for tval hard armor:   min 189, max 383, avg 281
Power for tval dragon armor: min 353, max 783, avg 523
Power for tval light:        min  39, max 177, avg 110
Power for tval amulet:       min  69, max 207, avg 147
Power for tval ring:         min 111, max 672, avg 358

Those are the current standarts.

[Pete Mack]

@Nick--
Make sure not to include fixed arts in your list. I suspect 672 is the power of The One Ring.

[PowerWyrm]

@Nick--
Make sure not to include fixed arts in your list. I suspect 672 is the power of The One Ring.

672 is Vilya -- the One Ring has INHIBIT power.

[Derakon]

Out of curiosity, how does a Ring of Speed (+10) rank on this scale?

[Pete Mack]

@Nick-
Recommend not including traditional 'bad' items in the averages and min values. Beruthiel, Mormegil, Gorlim all should get their heavy curse back and should be generated as genuinely dangerous items, not as weak items.

Calris is a different issue: it's pretty much worthless for Warrior and Mage, as they can't possibly enchant it enough to be useful. It used to be a good weapon for high level mages: the CON is invaluable, and the aggravation doesn't matter much with banishment and destruction spells.

[Pete Mack]

@Derakon-- less than Feanor, which come in at 365.

[PowerWyrm]

Out of curiosity, how does a Ring of Speed (+10) rank on this scale?

Best shooter is 271, which is probably Cuby with +10 speed.

[Pete Mack]

I'm a little surprised that Deathwreaker is coming in so low. It's a ridiculously powerful artifact--just utterly murderous. For most characters, I'd prefer it to Vilya.

[Nick]

Recommend not including traditional 'bad' items in the averages and min values.

That's probably a good idea.

I'm actually not sure what to do about the bad artifacts. They haven't really made much sense for a while, and probably make even less with the new curses; but on the other hand, I'm reluctant to abandon them completely.

Calris has a different issue; it was always a genuine mixed blessing item, but now there's the problem of the big minus to-hit and too few scrolls to enchant it up. Possibly it should just get a couple of different, suitable curses, and the player has the choice of taking it as it is, or trying to uncurse with the risk of losing it.

[Derakon]

That's probably a good idea.

I'm actually not sure what to do about the bad artifacts. They haven't really made much sense for a while, and probably make even less with the new curses; but on the other hand, I'm reluctant to abandon them completely.

Calris has a different issue; it was always a genuine mixed blessing item, but now there's the problem of the big minus to-hit and too few scrolls to enchant it up. Possibly it should just get a couple of different, suitable curses, and the player has the choice of taking it as it is, or trying to uncurse with the risk of losing it.

Maybe one possible curse would be "large penalty to to-hit", so Scrolls of Repair Curse could fix that malus. I don't know how well that would interact with combat stat tracking (i.e. the fact that there's now a mysterious modifier to the to-hit for the weapon, which would need to be tracked whenever to-hit is used).

[Nick]

Maybe one possible curse would be "large penalty to to-hit", so Scrolls of Repair Curse could fix that malus. I don't know how well that would interact with combat stat tracking (i.e. the fact that there's now a mysterious modifier to the to-hit for the weapon, which would need to be tracked whenever to-hit is used).

You're right, and that is actually really easy. The new curses are implemented via a fake object whose properties are added to the existing object properties; for example the vulnerability curse "curse object" has -50 to AC and the aggravate flag.

[Gwarl]

That's probably a good idea.

I'm actually not sure what to do about the bad artifacts. They haven't really made much sense for a while, and probably make even less with the new curses; but on the other hand, I'm reluctant to abandon them completely.

Calris has a different issue; it was always a genuine mixed blessing item, but now there's the problem of the big minus to-hit and too few scrolls to enchant it up. Possibly it should just get a couple of different, suitable curses, and the player has the choice of taking it as it is, or trying to uncurse with the risk of losing it.

You could rescue the bad artifacts with the new curse system and show off the new curses at the same time (I think they're really neat but there's not much reason to ever use cursed equipment currently). Switch the stat penalties for interesting curses, like the crown which offers telepathy also inducing hallucinations would be a good one, or gorlim having the stoneskin curse, wormtongue boots getting siren, etc.