Making randart sets more cohesive

[Derakon]

Crossposting from the "No OP Items" thread, because I thought this was an interesting idea and I'm disappointed nobody's responded/reacted to it. As background, the issue being discussed was that it's currently too easy to generate randart sets where the player can assemble "full coverage" of important abilities and also get massive stat bonuses / good damage / good speed. It was proposed that part of the issue is that each randart is generated in a vacuum, so to speak, which ends up often making them more synergistic than standarts.

I think a few heuristics should be able to handle synergy for most cases. I don't think it's necessary to have an algorithm that's guaranteed to be perfect; after all, rare outlier games (both too strong and too weak) help to keep things interesting.

Synergy in general comes from artifacts covering for each others' weaknesses, while still providing good coverage in stackable abilities like speed and stats. Therefore a lack of synergy comes when multiple important abilities are only reliably avaliable on a single slot (and usually not at the same time), or when covering binary abilities comes at the cost of losing out on stackable abilities. For a specific example, pBlind/pConf are most frequently available on helms in standart games, and so is ESP, but there's very few sources of all three in the helm slot. This frequently leads the player to have to decide which of the three powers they want (or how to juggle their gear to get the abilities "off-slot" on an otherwise-weak item), instead of just getting all of them conveniently.

So for example (just spitballing here), an anti-synergy heuristic could work like this:

* Have a predefined set of Important Binary Abilities (IBAs). This would be things like rBase, SI, FA, ESP, pConf, etc. Abilities that all players want to have.
* Break the IBAs down into sets of at least 2 abilities. Assign each set a favored slot.
* For any randart, assign higher power if it has multiple abilities from the IBA set.
* For any randart not in the favored slot, assign higher power for each ability it has from the IBA set.

For example, if rFire and rCold are in an IBA set, and the game decides to assign that set to the ring slot, then we might see randarts like:

* The Ring of Foo (rFire) power 25
* The Ring of Bar (rCold) power 25
* The Ring of Baz (rFire, rCold) power 75
* The Boots of Quux (rFire) power 50
* The Boots of Quuux (rFire, rCold) power 100

Higher power means that the randart has "less room" to fit other abilities, making it weaker on the whole in exchange for the versatility it gets. This should, if I thought it through correctly, create randart sets where you either get versatility ("off-slot" abilities) or you get useful stats, but both would be rare.

[Philip]

I figure this is a very good way to make randart sets resemble the standart ones a bit closer in terms of how they are balanced, though I don't know if it is sufficient as a balance fix on its own - I suppose if the cost of off-slot +attacks and slays goes up even further, it could be? The problem with some overpowered slots and some slots covered by egos (which in many slots are quite comparable to the best standarts) would remain, I think, but fixing that would either require some rather aggressive nerfs to high egos or making randarts boring.

[Estie]

Whats the point when ego items have enough synergy to get a perfectly good winning setup ?

[Derakon]

Whats the point when ego items have enough synergy to get a perfectly good winning setup ?

The relative balance of ego items and artifacts is a related issue, but I think it can be tackled separately. The balance of standarts and ego-items is a little precarious, but fairly clear; in principle it should be possible for randarts to (usually) achieve the same balance.

[Estie]

The relative balance of ego items and artifacts is a related issue, but I think it can be tackled separately. The balance of standarts and ego-items is a little precarious, but fairly clear; in principle it should be possible for randarts to (usually) achieve the same balance.

How so ? In your example, you can require the Ring of Baz to use up power 75 all you want, if there is an ego (rFire, rCold) that drops all over the place, it simply means that you end up with a trash artifact and use the ego instead.

Whatever artifact evaluation you want, wouldnt it make sense to "substract" the ego in some way before evaluating.

[Derakon]

How so ? In your example, you can require the Ring of Baz to use up power 75 all you want, if there is an ego (rFire, rCold) that drops all over the place, it simply means that you end up with a trash artifact and use the ego instead.

So pick your numbers with care, and maybe tweak how you generate the IBA sets.

Whatever artifact evaluation you want, wouldnt it make sense to "substract" the ego in some way before evaluating.

A hypothetical codebase that took egos into effect when designing artifacts, allowing you to add/remove egos and have that affect how randarts get generated, would be pretty cool! But the ego-items have been pretty stable for awhile, so I'm not too worried about "baking" knowledge of ego-items into the randart generator, especially if it makes the randart generator easier to write.

If you can suggest an algorithm that does take the ego items into account though, I'm listening.

[Estie]

First off, sorry for the negativity in my previous post.

As for the randart distribution: initially I thought of a way to do exactly what I proposed: compare the mods to existing ego mods in the same slot at creation and act on that - but that misses the point in the same way your proposal does.

I have said this before, but I see the problem simply in the proximity of standart and ego power. An @ with dwarven pdsm, a MoD +2 attacks, Caestus of power etc. is virtually indistinguishable from an @ with top end standarts. One reason for this to have come to pass the recent scraping off from artifact values, like Sting or Theoden - in an attempt to lower @ overall power by changes to artifacts exclusively instead of distributing changes over both item types.

I would agree that ego items shouldnt become obsolete in the endgame kit, but I would like to see them more inferior to artifacts. You said yourself that ego items have remained stable for a long time - and this is imo the root cause for all the trouble. When balancing is done, everyone looks at the easy way out and changes the artifacts yet again. And thats why my initial reply to your proposal was grumbly.

So instead of some weird randart distribution, find ways to scrape off of ego item power. Then the overall value of randarts can be lowered without ending up with mostly junk, and the outliers become less extrem.


There is the matter of "harder game". To adress that, I really would like to (have Nick ) look into buffing monsters in more or less global ways. When zephyr hounds got cut to size, I rejoiced, but they had been the cheap way to make the dungeon dangerous. It could use being more dangerous.

[Ingwe Ingweron]

When zephyr hounds got cut to size, I rejoiced, but they had been the cheap way to make the dungeon dangerous. It could use being more dangerous.

I agree, it could be more dangerous, but please don't bring back immense Zephyr hound pack sizes. Sure, they're dangerous...for awhile, but then they are just tedious.

[Estie]

I agree, it could be more dangerous, but please don't bring back immense Zephyr hound pack sizes. Sure, they're dangerous...for awhile, but then they are just tedious.

Absolutely - emphasis was on "cheap".

They have the best attack shape (unlimited range ball), the best type (elements), they come in packs, have the worst risk/reward ratio and I could probably think up some more superlatives. Oh and they dont fit into the tolkien universe.

No, but making monster archery better, giving them overall more melee damage, buffing some individual types with other stuff like additional speed, these things could be done. So, Nick, go work

[Nick]

No, but making monster archery better, giving them overall more melee damage, buffing some individual types with other stuff like additional speed, these things could be done. So, Nick, go work

That is one of the two big areas of change planned for 4.2 (the other is races/classes/magic). What I plan to do is go through every monster type from 'a' to 'Z' and rethink them. The intent is for every type to have some underlying reason for being, and for (almost) every monster to present some sort of a challenge at some stage of the game.

I've been keen to get on and work on this for some time, but have been prevented by RL becoming very busy, and by people continuing to find bugs in 4.1.x

[Derakon]

If you want a cheap way to worsen ego weapons, make their pluses lower. An MoD of Extra Attacks with only +10 to-dam is still going to be good, but it's going to have a lot more trouble competing against artifacts than that same weapon with +15 or +20 to-dam.

The same doesn't really apply to ego armor though. There I think you'll need to attack the pvals and flags.

I guess another option would be to score ego item power the same way you score artifact power, and penalize the likelihood of generating a high-power ego item.

[Grotug]

First off, sorry for the negativity in my previous post.

... [snip] ...


There is the matter of "harder game". To adress that, I really would like to (have Nick ) look into buffing monsters in more or less global ways. When zephyr hounds got cut to size, I rejoiced, but they had been the cheap way to make the dungeon dangerous. It could use being more dangerous.

From my personal (still somehow limited after all my games!) experience, I find the game is hard enough until @ reaches endgame power. Then @ is often fairly "invincible", and only falls due to hubris or carelessness. So maybe more monsters with native depth around DL90 that are even more dangerous than all the max danger monsters of the DL70s? If I'm not mistaken, the most dangerous non unique monsters in the game show up in the DL70s and the last one is like DL79 (Greater Balrog) (although maybe GWofBalance is DL84 or something). So maybe some new monsters, with native depths in the DL80s and/or 90s, or change the native depth on Reavers, Greatest Greater Demons and Greatest Dragons and make them even more terrible? Seems that the top topmost monsters should have their breath/spell rate higher than most other things (the most common seems to be 1 out of 5).

I suppose a really easy way to make the deeper dungeon more dangerous is to simply change some of the 1 out of 5 to 1 out of 3 or 2 on many of the toughest monsters. People say rShards aren't important? Make Gelugon breathe shards more often. Although I suppose it's good if the game has some elements which are, overall, less dangerous than others than trying to make them all really dangerous.

One of the problems of trying to keep the game challenging for super-vets is that invariably it will be even more unwelcoming to newcomers or casual/occasional players. Which brings up the mostly terrible idea of suggesting giving Angband difficulty settings.

Original Doom had 5 difficulty settings. In Easiest, @ took half the damage of all the other skill levels and all ammo pickups were doubled. Difficulty settings 2-4 were mostly differentiated only in the number of monsters in each. The most difficult one called Nightmare wasn't even remotely fair: cheat codes didn't work, monsters came back to life some time after being killed, and monsters were all much faster than in normal difficulty settings, and ammo pickups were doubled.

So maybe you could have 3 difficulty settings in a similar vein as to original Doom? The easiest would make @ receive half the amount of damage, and each backpack slot would hold 80 instead of 40, and max weight would be, I dunno, 210lbs instead of 180lbs. The next up would be the current, and default, difficulty setting and then a third option would increase all or some of the following: monster damage output, monster hitpoints, monster breath/spell frequency, powerful item generation/drop frequency, and dead monsters come back to life after a time . Wins on the easiest skill level wouldn't be postable to the leaderboards (or if postable, an asterisk denoting an easy road was taken).

I dunno, difficulty settings seems cheesy in a game like Angband. There should really only be one difficulty setting. Maybe, as an easter egg, pressing some annoying combination of keys when Angband loads, could display a secret difficulty settings screen with the above options, but it would be considered cheating to select a difficulty setting other than the default one, and the player would be warned of such. Wins selected on a non-main difficulty setting would not be uploadable to the ladder.

Or scratch all that, and make something tailor made to my personal interests : another birth option that halves the number of levels in the game, increases the number of dangerous levels, (ie. that contain OOD monsters and items), and reduces the number of levels with scarce monsters (how boring!), and reduce the frequency of of those small, square, maze levels.

Chance of good drops from monsters would maybe increase (I haven't decided on that point, it is probably enough to encounter more opportunities to get good drops and thus it not need be to increase drop chance). Vault generation stays the same, so if you are playing with Forced Descent, you will see half the number of vaults as a current Forced Descent game sees.

Hmm, maybe another option that halves vault generation and doubles drop chance from uniques. Did Smeagol get nerfed at some point? I haven't gotten a good item from him in ages. Lately it's a small sum of gold pieces. Or Wormtongue for that matter.

Zephyr hounds.. I'm not sure I've ever even seen these? Nevermind! It's actually nice to know there are still things in the game I don't know about! Lately I have been running into a lot of Aether hounds which are pretty nasty to actually try and fight. They are one of the few monsters a super powerful @ cannot safely take on, even with 4 immunities last game, when I was about to take on the final bosses, I had to abort taking on a pack (8 or 10?) after only successfully killing one. Time hounds at least fall relatively easily.

[PowerWyrm]

Zephyr hounds.. I'm not sure I've ever even seen these?

That's just the common term for "hounds". And it's why they are represented as "Z" (for "zephyr") and not "C" on the screen.