Question regarding good/great probability

[ChodTheWacko]

I noticed a recent change greatly increased the chance of getting a good/great item.

Okay, that's cool. However, are you planning on reverting obj-make back so
that if a monster does a *good* or *great* drop, it will actually drop an item which could possibly be great?

Right now, it doesn't. It will randomly generate an item 3 (or 5, if it's *great*) times, and check if obj_is_good, and if not, it uses it anyway.

This reduces the value of killing monsters with "good" drops, and vaults.
I personally would rather see this modified, than have the overall base chance
for all monsters dropping good objects increased.

But I thought I'd throw the question out there, for the masses to debate.

- Frank

[Pete Mack]

yes, it's a problem-killing umiques, wyrms,etc no longer is so useful because the size of their drops is much smaller but the chance of getting something useful is way down. It's a known problem, and it's a bad one. I'm playing NPP and/or Eddieband til it gets fixed.

[takkaria]

I noticed a recent change greatly increased the chance of getting a good/great item.

Actually, just the chance of getting a great item, but 3.1.0 overall greatly increases the chance of both:

good items:
3.0.9: the chance is (level+10)%, capped at 75%
3.1.0: the chance is (2*level + 10)%, uncapped

great items:
3.0.9: the chance is (level/2 + 5)%, capped at 20%
3.1.0: the chance is (level + 5)%, uncapped

It was always a bit ridiculous that the chances of getting an {excellent} item with normal generation wouldn't get any higher after dlev30. Now you're guaranteed to get {excellent} items from dlev95 onwards.

Okay, that's cool. However, are you planning on reverting obj-make back so
that if a monster does a *good* or *great* drop, it will actually drop an item which could possibly be great?

Not reverting, but yes, when I have a spare day this weekend or next I'll be recoding object generation again so that DROP_GREAT monsters always drop great items.

Also, with the new chances, all monsters will do the equivalent of DROP_GOOD after dlev40 anyway. I will probably make it so all uniques are guaranteed DROP_GOOD and remove the flag, since there are no non-unique monsters before dlev40 with DROP_GOOD at the moment.

This reduces the value of killing monsters with "good" drops, and vaults. I personally would rather see this modified, than have the overall base chance for all monsters dropping good objects increased.

This isn't an either/or thing; both upping base chances and sorting out DROP_GREAT drops are necessary, just because monster drops have been reduced so much in 3.1. e.g. the number of drops at dlev100 has been reduced by a factor of three, and so some countermeasure to make {excellent} items at least three times more likely is necessary. Drops at dlev30 have also been reduced by a factor of around three; so perhaps the chances of finding excellent items should be upped by a factor of three too (from 20% to 60%, rather than just to 35%).

We'll have to see how it plays.

[ChodTheWacko]

Also, with the new chances, all monsters will do the equivalent of DROP_GOOD after dlev40 anyway.

It doesn't matter. DROP_GOOD/DROP_GREAT is broken.
I don't think you understand the real problem here.
You can hard code good to 100% and great to 100% and that won't fix the real problem.

There are two basic steps to object creation:
1) Pick an object
2) use the good/great percentages to apply relevant powers to it.

You are modifying step 2.
What got broken is step 1.
Step 2 doesn't even happen unless step 1 picks an object where step 2 even applies.

This is why you can Kill a Great Wyrm and just get 3 scrolls, a flask of oil, and a food ration.
apply_magic() is called with good=100% and great=100%, but it just ignores those objects.

You need to insure that kind_is_good() is true if you are creating a good/great object.
Otherwise your good/great percentage is irrelevant.

Right now the probability of kind_is_good() is capped in make_object()

- Frank

[takkaria]

It doesn't matter. DROP_GOOD/DROP_GREAT is broken.
I don't think you understand the real problem here.

I do, don't worry. The bit you excerpted was a side-comment. As I said, I'll be restoring DROP_GREAT from being a no-op when I have a spare day in which to get knee-deep in object generation code again.