reduce artifact drop rate

[Pete Mack]

For basic artifacts, I think randarts and standarts have about the same frequency.

With a few exceptions (like boots, bows, steel helms, shields and longswords), the really uber standarts are less common than their corresponding randarts. If the base object is PDSM, BoC, MoD, etc, the chance of seeing the artifact is extremely low, even before applying the artifact frequency. The last time I checked (admittedly long ago), there was no correction for this in the randart code.

Edit:
Also, i don't think people really remember how common artifacts really were in 3.0.6
Yes, early in the game they were fairly rare. But it only took clearing a single dragon pit at depth to pick up 3 to 5 at a time, never mind a significant stock of consumables. (Ego ammo, !Exp, !*Life*, ?Mass Banishment, etc.)
At dl 98, you could generally find several high-end potions and scrolls with each activation of the Palantir. That is just not true anymore.

I used to make the high-end scrolls hot-pink ? so I wouldn't miss them in the noise. Now, however, scrolls, etc, are sufficiently rare on the ground that I am usually willing to examine item to determine its type now.

[Magnate]

For basic artifacts, I think randarts and standarts have about the same frequency.

With a few exceptions (like boots, bows, steel helms, shields and longswords), the really uber standarts are less common than their corresponding randarts. If the base object is PDSM, BoC, MoD, etc, the chance of seeing the artifact is extremely low, even before applying the artifact frequency. The last time I checked (admittedly long ago), there was no correction for this in the randart code.

Well, the correction has been there since r1248 so it really must have been a long time ago - you're talking about the old GW code.

In fact the uber-artifacts with common base items, both standard and random, are more common because max rarity is now 100 (gen chance = 1). Artifacts like Zarcuthra and Feanor are significantly more common, but Bladeturner and Deathwreaker shouldn't be.

Your reminders about 3.0.6 were interesting though. I'd forgotten that.

[EpicMan]

For pseudo-by-power I would prefer absolute ratings instead of relative, because a {good} SoS is superior to most {excellent} warhammers. I think using the power rating system would work well for this. We may also want to use a k_info.txt and/or e_info.txt modifier to the formula-derived power to deal with base items/egos that get lower or higher scores than they should have.

In fact, if we go that far we could just base the pseudo on the base+mods+ego brand+random power score. For instance an average dagger or short sword would get a fairly low rating (but higher than a broken dagger or cursed halberd) while an average power dragon scale mail would get a fairly high rating (because we would have a high power modifier in PDSM's entry in k_info.txt) even though it is just an 'average' PDSM - an average PDSM is better than most excellent items and some artifacts.

Pseudo on pickup would be fine with me, then pseudo on walkover at a certain level (should be tied to current pseudo-strengths such that warriors get it before anyone else and magic users get it later).

I suppose I could live with pseudo on LOS, but detection magic should *NOT* pseudo out of LOS, otherwise those spells would be even more overpowered than they are now.

[PowerDiver]

Very ironic that it should be you who suggests this, because ... doesn't this lead us towards pseudo-by-object-power? What you're effectively saying is that you're more interested in the base damage of the weapon than in its other properties (for melee chars at least), so {good}, {excellent} and {splendid} should essentially represent tiers of damage (either absolute, or relative to the plain unenchanted base item - I'd favour the latter). A 5d5 katana with +15 to_d should pseudo higher than a 3d5 weapon of *slay orc* with only +7 dam, right?

Nope.

Pseudo is about recognizing whether an item is an ego. It has nothing to do with the power of the item. Heck, it is what you can tell about an item while it is in your pack and you are still too scared to touch it. I couldn't even get Tak to agree to differentiate excellent from splendid on pseudo. That takes a wield to differentiate. The best way to fix pseudo is to eliminate it.

I still think your power ratings are broken. Sorry, but they don't work IMO. I would hate to see anything used for squelch to have anything to do with them.

You should get absolute information at some point or another. The question is what you get when. I've suggested what I called the rune-based approach in the past. Then the question becomes when can you see a rune. Is that wield or pickup or walkover or LOS? Any of those can make sense, so the answer should be determined by gameplay considerations.

One possible answer to the damage dice is to have a particular flag that means extra dice. That doesn't solve the +30 to plusses problem. However there is a simple answer. Give the plusses when you can see a rune. Problem solved. So what if it makes no sense. Gameplay is the most important thing.

Personally, I think extra dice makes no sense. All this is just my singular opinion of course. A longsword should be inherently 2d5. If you want a 4d5 weapon, it should have a different base class. If Ringil must be a longsword, make it 2d5 and give it x4 slay evil and x6 cold brand etc.

[Derakon]

I don't think pseudo on LOS is necessary, personally. If you can't even be arsed to walk over the pile of loot then there's no help for you.

I do like the idea of basing pseudo on power ratings, or adding a secondary power pseudo. Something like {good, mighty} or {excellent, unimpressive}. The only problem then is choosing where the divisions are between different power pseudo levels.

[Timo Pietilä]

Personally, I think extra dice makes no sense. All this is just my singular opinion of course. A longsword should be inherently 2d5. If you want a 4d5 weapon, it should have a different base class.

I see extra dice weapon as very good quality weapon. IE. non-magical quality is excellent for 4d5 longsword. A very very sharp and hard blade instead of Xena-like longish lump of metal with a handle. OTOH it doesn't make much sense in blunt weapons. A mace is a mace no matter what you do to it.

[Timo Pietilä]

I don't think pseudo on LOS is necessary, personally. If you can't even be arsed to walk over the pile of loot then there's no help for you.

I do like the idea of basing pseudo on power ratings, or adding a secondary power pseudo. Something like {good, mighty} or {excellent, unimpressive}. The only problem then is choosing where the divisions are between different power pseudo levels.

In order to make power levels work they should be based on what character has at that instant and what he might want at the future. What he has in home, inventory and wearing. All possibilities covered. There are times in game when you value Gauntlets of FA higher than Gauntlets of Power. Item with rPoison totally useless or valuable treasure over anything else. When speed is no more valuable. When ESP doesn't have value.

Power ratings don't work. Pseudo-ID squelch based on power rating doesn't work.

[Laie]

I don't think pseudo on LOS is necessary, personally. If you can't even be arsed to walk over the pile of loot then there's no help for you.

I fully agree.
Although... I kinda like the idea that it might work similar to perception. Walk up to a pile and notice some items. Low-level characters might need to trudge through the loot a few times.

(this assumes that this new pseudo-ID doesn't immediately assess everything, but improves with character level. Although I complained about it myself, I sincerely believe that picking up many items that turn out to be "worthless" is a necessary part of Angband. Though it would be convenient if it weren't quite as many as right now.)

[Magnate]

I still think your power ratings are broken. Sorry, but they don't work IMO. I would hate to see anything used for squelch to have anything to do with them.

Yeah, I remember. You're in the camp that says it isn't feasible to estimate a numerical value for any item given the number of different circumstances altering the subjectivity of the valuer. Well, at least you are crystal clear on that (Timo likewise). Yet you did say

In these days of acid rings, or more particularly randart armors with brands or slays, a good katana 5d5 may well be more valuable than a katana 3d5 of *slay undead.

I thought that meant you wanted a way to automate (partially) the differentiation of "valuable" but clearly, like Timo, you want to do it game by game, item by item, without any assistance from the game engine (except a good squelch UI, of course). Which is fine - but a lot of people are asking for some assistance, and that's what this thread has become (with apologies to fizzix, for that's not how it began).

Personally I think squelch-by-power makes vastly more sense than squelch-by-ego-type, but you seem to have a religious attachment to the latter. We can agree to disagree, as ever. You can rest assured that changing squelch is a lot of work and isn't on anybody's radar at the moment. I'm quite encouraged by the modicum of support that's been expressed for the idea of using power for pseudo/squelch levels, but not enough to do anything radical yet. (The codebase is currently undergoing major internal restructuring, so nothing particularly noticeable is happening to gameplay. We're also in the process of moving from svn to git, which is why nothing is happening on trac.rephial.org at the moment.)

[Timo Pietilä]

Personally I think squelch-by-power makes vastly more sense than squelch-by-ego-type,

It would if you could estimate item power level in any sensible way. Problem is that you can't. As I said you need to do it based on what player already has including items in home and inventory, and even guess what he might want in future and that just makes it impossible.

If you could do squelch based on item abilities (damage output, AC, stats, pval, flags) like in debug-mode reroll excellent and concentrate keeping only items that have flags you have not marked as useless, then it could work. Even that damage output is difficult one, because Pain at start with weak hobbit mage would be much less powerful than Sting, but could be very powerful in future. You would need to determine damage output as potential damage output assuming STR and DEX 18/220. And then there are combinations of abilities. One could dismiss all single-basic4 armors, but value high item with all of basic4.

[Magnate]

It would if you could estimate item power level in any sensible way. Problem is that you can't. As I said you need to do it based on what player already has including items in home and inventory, and even guess what he might want in future and that just makes it impossible.

I know, that's precisely what I meant by "it isn't feasible to estimate a numerical value for any item given the number of different circumstances altering the subjectivity of the valuer". I know you and Eddie (and doubtless others) think that - I respectfully disagree. I think the number of circumstances in which the value of an item changes from its modal value is much much smaller than the two of you seem to think, and it is therefore feasible to reach a mathematical approximation of value which, while it can never be perfect, is certainly good enough to be useful.

I'm not saying that power-based squelching is the answer to the age-old problem of junk generated at depth. I don't think there can be an answer to that problem, I think it's inherent in Angband and I think the game would be a lot worse if that "problem" was "solved". I am just curious to see if it would help at all. One day there will be some serious effort put into improving squelch, and it wouldn't hurt to have an optional power-based dimension (as long as that didn't prevent people who wanted to squelch by type from continuing to do so).

[ewert]

In my experience, most games with randarts end up with some equipment slot being taken by only a mediocre or even relatively weak item just because it has that single or two needed things. For example my current priest with quad immunities and everything covered except SI (thank god for the spell =)) wears a lowly Aman cloak with rDis. He could swap a weaponmastery, sure, and lose speed. None of the artifact cloaks are worthwhile, as he already has max Wis, Con, enough str/dex for 4 hits, etc., but lacks rDis and I prefer carefree play and thus like to max the whole page ...

I guess a powerbased rating is semi-useful too, but ... it is hell of a lot of work to tune even somewhere "good", and even then I reckon most players will keep it a bit low if used for squelching compared to their gear, just to not miss a semiuseful swap weapon for example.

Then again, I just squelch all but artifact weapons after a point, and have autoinscribe !k on the top 3 weapons. =P I have much much more problems with armor ego's, due to DSM, not remembering always if one ego is splendid or not, etc, so tend to waste a lot of time id'ing armor. And with armors ego-type squelching is killer good.

[fizzix]

I think people are too caught up in getting an accurate numerical power based rating. In reality you just need a rough estimate. You're not trying to distinguish between a 3d4 broad-sword with +20 to dam and a 4d4 great axe with +10 to dam. You're trying to distinguish between a 1d4 dagger of slay troll to a 4d4 HA great axe.

If you're going for an automated squelch, the bar gets set very low. You're really just trying to get rid of the chaff and leave the stuff that might possibly be useful.

If you want to increase accuracy in your numerical representation you will probably need at least a 2 dimensional array for weapons (and maybe even 3 dimensional). Combat is one axis, resistances and abilities is another axis. Then you can easily distinguish between a defender weapon that you consider junk to a +20 MoD with no other effects. This is possible, and it might be something to consider if power based squelching is an option. You could squelch all weapons that have a poor combat rating except for weapons with very high abilities (meaning telep, rconf or rpois probably)

I'm also fine that the thread wandered off topic. I kind of expect that to happen with these threads, and it's my fault as well as everyone else's.

[Magnate]

a lowly Aman cloak with rDis

I think this is part of the "game is too easy" problem which sparked a big thread recently. Aman cloaks are very rare, have huge AC and a stealth pval and a high resist. The game should never reach a point where they are considered "lowly" ...

[Magnate]

I think people are too caught up in getting an accurate numerical power based rating. In reality you just need a rough estimate. You're not trying to distinguish between a 3d4 broad-sword with +20 to dam and a 4d4 great axe with +10 to dam. You're trying to distinguish between a 1d4 dagger of slay troll to a 4d4 HA great axe.

If you're going for an automated squelch, the bar gets set very low. You're really just trying to get rid of the chaff and leave the stuff that might possibly be useful.

If you want to increase accuracy in your numerical representation you will probably need at least a 2 dimensional array for weapons (and maybe even 3 dimensional). Combat is one axis, resistances and abilities is another axis. Then you can easily distinguish between a defender weapon that you consider junk to a +20 MoD with no other effects. This is possible, and it might be something to consider if power based squelching is an option. You could squelch all weapons that have a poor combat rating except for weapons with very high abilities (meaning telep, rconf or rpois probably)

I'm also fine that the thread wandered off topic. I kind of expect that to happen with these threads, and it's my fault as well as everyone else's.

Cool. I agree with you that we only need the power rating to be approximate to be useful. I also like the idea of having two different ratings - one for damage and one for defence/resists/utility. As you said, maybe even more than two. This is something to consider when overhauling the approach to power (ticket #1005).

One thing on which we all seem tacitly agreed is that a better squelch system is preferable to either wading through junk or inhibiting generation.