reduce artifact drop rate

[fizzix]

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.

I'm not sure I agree with that. What I agree is a better squelch system would be very much appreciated. I still think that a hard squelch of useless objects on creation isn't a bad idea and may be preferable. Better squelch still requires you to ID objects, and that is where most of the tedium comes in.

edit: auto-id on walkover after some clevel would probably be fine, and does deal with the tedium of ID. Specifically the tedium of ID in greater vaults and dragon pits. Those are the only times that ID is really problematic.

[Magnate]

I'm not sure I agree with that. What I agree is a better squelch system would be very much appreciated. I still think that a hard squelch of useless objects on creation isn't a bad idea and may be preferable.

Ok. I am interested to know how many others share your view. I think it is very hard to come up with multiple agreeable definitions of uselessness (for different depths, play styles, characters), so hard-coding the squelching by not generating the drop is almost bound to upset someone.

[Timo Pietilä]

Ok. I am interested to know how many others share your view. I think it is very hard to come up with multiple agreeable definitions of uselessness (for different depths, play styles, characters), so hard-coding the squelching by not generating the drop is almost bound to upset someone.

For me squelch based on pseudo is doomed at start, because you need to be precise what you want and games differ too much game to game. Auto-squelch based on level is also not very good solution because you can't know what people really want. Squelch at ID is only working solution, so we need to make ID itself more automatic.

I would think even LoS -ID for high enough clvl char would be possible. Some rune-based thing, so that if you know what "fire brand" looks like you automatically recognize it in any weapon without waiting for pseudo or anything.

For example if you have found weapons of slay orc, troll and giant before finding your first Westernesse you know that it has all those slays, but you still need to ID it as Westernesse to find SI and FA. Once ID:t you then recognize all Westernesse in LoS (and can squelch them if you don't need them).

Something like that.

For detection I don't think we should not give any info about any items using any detection method except detect enchantment, and that should only reveal location of enchantment. Clairvoyance should not show items, detection should not show items and detect treasure should not show items. Restrict detection to monsters, doors, stairs and traps.

At _minimum_ clairvoyance should not reveal them. It is too easy to just cast clairvoyance, "[" and notice that there is nothing interesting here.

[PowerDiver]

For detection I don't think we should not give any info about any items using any detection method except detect enchantment, and that should only reveal location of enchantment.

I think that is going too far. In order to use strategy, you need to be able to at least estimate gain vs risk. Otherwise you are left with mindless level clearing.

Only priests get clairvoyance. Everyone else needs the Palantir. By the time you are expected to get the Palantir, I don't mind it revealing objects. If most people want to nerf clairvoyance I won't argue, but I'm mildly opposed and I don't see that it is a sufficient problem to need fixing.

If you think that, when using clairvoyance, most levels are so dull the char will just leave, perhaps the effort would be better spent making better levels. I don't see how it improves the game to enoucrage the char to clear boring levels.

[buzzkill]

Please, no hard coded squelch. Let the player play the game, make his own decisions. If sorting the crap that drops is too troublesome, it can always be left behind. Nobody ever forced me to clear a dragon or troll pit.

[ewert]

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" ...

Well the context of the character is dlvl 98, clvl 50, and farming a bit just for the heck of it due to the quads. For example the char has artifact boots of I think +10 or +11 speed and stealth. Funky.

[Laie]

a silly idea...

Just now, I happened to "Detect Enchantment". A spell I hardly ever used. In principle, it provides you with all the knowledge you'd need in order to automatically sort out all unenchanted items... only that it doesn't work that way.

Anyway, just to toss in another thought besides pseudoID-on-walkover: how about doing it with spells instead?

[fizzix]

Yay for thread necromancy

Whatever I said about standarts being too common is wrong. Completely, absolutely, no doubt about it, wrong.

However, it does still stand that randarts are too common. For example, compare how many artifacts this randart character found, versus this standart character 2.5 times more artifacts in roughly the same amount of turns.

For more comparisons there's this standart beast who killed all the uniques and twice as many monsters, but still wound up with less artifacts than the randart guy.

It's interesting to note, that the frequency for the special artifacts seems fine. That's unsurprising because that shouldn't change too much between standarts and randarts.

[PowerDiver]

There's always the possibility of variance.

Was the randart game since the change to drop level and since ewert's changes moving drops away from weapons/armor towards flavored objects? Either of those might already be the solution to your perceived problem.

[fizzix]

There's always the possibility of variance.

Was the randart game since the change to drop level and since ewert's changes moving drops away from weapons/armor towards flavored objects? Either of those might already be the solution to your perceived problem.

Both are the same version (2041). So they are both after ewert's changes.

Yes, variance is an issue, especially with this small sample size, 4 games total (one is a dumpless YASD). However, check out Ycombinator's games for some more evidence that the randart drop rates are inflated.

[PowerDiver]

Yes, variance is an issue, especially with this small sample size, 4 games total (one is a dumpless YASD). However, check out Ycombinator's games for some more evidence that the randart drop rates are inflated.

Randarts used to be artificially rarer, then were changed.

Whether you find Dethanc at DL98 clearing a dragon pit is irrelevant. It doesn't really matter how many artifacts you find in total. A better measure is how many are useful when you find them. Then a bonus for amount of usefulness. All in all incredibly hard to quantify and depends upon playstyle, of course.

For starters, someone needs to write a utility to create DLs from 1 to 99, kill the monsters and generate all of the drops and keep statistics, reinitialize and repeat a thousand times. That's been an open ticket for a while. It's not perfect, but some quantitative experiments could only help.

[Magnate]

Randarts used to be artificially rarer, then were changed.

Whether you find Dethanc at DL98 clearing a dragon pit is irrelevant. It doesn't really matter how many artifacts you find in total. A better measure is how many are useful when you find them. Then a bonus for amount of usefulness. All in all incredibly hard to quantify and depends upon playstyle, of course.

For starters, someone needs to write a utility to create DLs from 1 to 99, kill the monsters and generate all of the drops and keep statistics, reinitialize and repeat a thousand times. That's been an open ticket for a while. It's not perfect, but some quantitative experiments could only help.

Well, I clearly made the change based on inadequate data, as things are now worse than they were before (in the unacceptable direction of making the game easier!). So as soon as that became obvious, I decided I wouldn't mess with the drop rates again until after the monte carlo generator Eddie describes above has been implemented. I'm waiting until the unit testing infrastructure is finished because I think it would fit well into that, but takk has pointed out that I could also build on wiz-stats.c

Anyway, one day I will sort this out, on a much more quantitative basis.

[Derakon]

In the meantime, could we revert randart rarity back to the artificially-scarce settings they had before? Or is that impossible to do without reverting ewert's changes to general item rarity?

[Magnate]

In the meantime, could we revert randart rarity back to the artificially-scarce settings they had before? Or is that impossible to do without reverting ewert's changes to general item rarity?

Fortunately they're unrelated (Ewert's changes affect base item drops, not the escalation to good/great/artifacts - at least, those that made it into svn were so limited - he may have gone further in his git branch), so it is trivial to revert the changes. But it wasn't right before either, so I am keen to keep this issue alive on my to-do list and give myself an incentive to write the stats generator. If takk wants to release 3.1.3 (or whatever the next release is called) before I do this, then I can revert as a least-worst option.

[Derakon]

Fairynuff.