Magnate's new egos

[fizzix]

Ran some basic stat runs. The first post I'll give here is the rough level that you have a 50% chance of finding an object by. Only See Invisible and Telepathy have statistically significant differences. RPois is determined mainly by finding rings, and that is why it's completely unaffected by affixes. (That also sets the lower bound for FA). edit: These runs were artifact-less to remove contamination by the standart set.

Ability      V       Affixes
----------------------------
FA           20      23
SInv         9       25
Blind        37      36
Conf         55      54
Nexus        38      41
Pois         41      41
Telepathy    48      63

In general with affixes. Branded and slay weapons are about 75% more common
Westnernesse, Gondolin, Defender and HA weapons are about 33% more common.
edit2: To put this in context. The first level with a 1% or greater chance of finding a westernesse weapon in current V is about level 16. In Affixes, it's level 62!
Weapons with extra blows are about 4x more common with affixes, which probably should be verified and changed.

Also, my code is finding Morgul weapons once again. Is this by design? It's possible that it's buggy.

[buzzkill]

Is the ability the 'christen' an item still around. Then you can give your crappy dagger any awful name you wish .

[Nomad]

I'm not sure if there is a middle ground with a different choice of adjectives. For example I find 'sharp' less bad than 'cruel'. However, I would also assume that all swords are sharp so I'd still find it odd. In addition, I'm not sure what the difference between a 'keen' and 'sharp' sword would be or 'of quality' and 'of craftsmanship'. By which I mean that they are basically synonyms, with 'in-game' meanings about which is better but not really natural meanings.

Maybe it's a bit of a case of over-categorisation? (Which necessarily requires coming up with a lot of names.) Or, actually, looking at it another way, perhaps the problem is not splitting the categories finely enough.

Maybe rather than having many different affixes like Keen, Sharp, Quality that describe different combinations of hit/dam/sides/weight/ac bonuses and penalities, there should just be a single affix for each thing. "Of Accuracy" for a bonus to hit, "Of Damage" for a bonus to damage, "Heavy" for extra weight, "Of Protection" for AC, etc. If you want one ego/theme to have a greater damage bonus than another, that could be covered by applying the "Of Damage" affix twice, rather than using multiple different levels of names for "Damage" versus "Even More Damage".

That way you don't need to come up with dozens of appropriately flavoured separate names for "small hit bonus and small dam bonus" versus "small hit bonus, large dam bonus, and a bit heavier", etc. - just simple individual names for each type of bonus/penalty.

[half]

Well, I think this is a matter of taste really.

Indeed it is. I share Antoine's taste on this and I imagine many others do too. I like the idea of a 'pick one attribute from one list and/or one from a second list', and I like the way that prefix/suffix naming makes this transparent. However I really don't like the names, which I find to be very over the top. People have different tastes as to what is 'over the top', but you can presumably see the dimension we're talking about.

I'm not sure if there is a middle ground with a different choice of adjectives. For example I find 'sharp' less bad than 'cruel'. However, I would also assume that all swords are sharp so I'd still find it odd. In addition, I'm not sure what the difference between a 'keen' and 'sharp' sword would be or 'of quality' and 'of craftsmanship'. By which I mean that they are basically synonyms, with 'in-game' meanings about which is better but not really natural meanings.

A poll is probably a good idea to sort out who is in favour and who against.

I should add that this is a really great coding and design effort, so hopefully there is some way to proceed.

[fizzix]

I've been wondering about this a little bit - can someone extrapolate on how stats distribution will be useful for the new affix type? It seems it may be difficult to compare distributions of holy avengers and "cruel swords of holy wrath", or armor of elvenkind and "durable dweomercraft stealthy leather jerkins of awesomeness"?

The main idea is you can search for appearances of flags. So you can find how often you find weapons with free-action, or brand-fire. You can also test the appearance of the same egos as previously, since they appear intact.

[bulian]

On your other point, the stats module will facilitate comparison with 3.3, but we don't have stats for earlier versions. But some devs have discussed backporting the stats code to 3.0.x precisely so that we can make these comparisons.

I've been wondering about this a little bit - can someone extrapolate on how stats distribution will be useful for the new affix type? It seems it may be difficult to compare distributions of holy avengers and "cruel swords of holy wrath", or armor of elvenkind and "durable dweomercraft stealthy leather jerkins of awesomeness"?

[Magnate]

>> The problem is that the names have flavour, a flavour different from the original.
> Ok, then they can be changed to match the flavour of the original. That isn't impossible.

So my suggestion is "someone should do that"

An excellent suggestion - I heartily agree. Someone with a better feel for the "correct" flavour than me!

What did you think of my idea of "silent affixes" that modify the properties but not the name or description? That should prevent the proliferation of "sharp swords of craftmanship" [i.e. sword (+6, +6)].

Well, I think this is a matter of taste really. Personally I find it much more flavourful and enjoyable to find a Sharp Sword of Craftsmanship (+6,+6) and then a Keen Sword of Quality (+9,+11) etc. etc. It's always struck me as one of the biggest absences of flavour in Angband, and one of the main reasons I did this work. I don't see any advantage in silent affixes at all.

>> That wasn't my argument in this post.
> Yet it is *precisely* what you went on to say in response to fizzix.

This is getting into an attribution snarl.

You thought I thought "that Angband really shouldn't have more variation in [non-artifact equipment] than exists in the current ego_item.txt"

I replied that, instead, "any increase in variation in ego items should be very carefully considered". And I suggested to the excellent fizzix that you should not "include any affix in 3.4 that was not (in some form) in 3.3".

So it is a matter of how much you change, and how fast; and my suggestion is that 3.4 should change the range of items as little as possible. Wide-ranging changes to the distribution of items will be bad enough.

I can handle a "mithril banded mail" or "rusty plate mail" if necessary, or maybe even a "two-handed sword of slicing" at the outside, but if Sethos's "giant bastard sword of cruelty" or "midnight mace of majesty" pop up any time soon then I will wash my hands of you.

But why? I am genuinely baffled. Are you positing that the addition of extra affixes will make rebalancing distribution more difficult? I'd argue the opposite: if we do as you suggest, and ensure that nothing can be generated in 3.4 that could not be generated in 3.3 (regardless of how it is generated or what it is called), then we have to do a ton of work rebalancing distribution for 3.4, and then we have to do it AGAIN for 3.5, or whenever we open up the generator to allow new permutations. It makes much more sense to me to do it once, and be able to keep improving it.

Or are you arguing that there is something sacred about the previous arbitrary set of ego types (which have been added to and edited in a completely un-strategic way over two decades)?

If your argument is about rebalancing, I understand it but disagree. If it isn't, I don't think I understand it yet.

On your other point, the stats module will facilitate comparison with 3.3, but we don't have stats for earlier versions. But some devs have discussed backporting the stats code to 3.0.x precisely so that we can make these comparisons.

[fizzix]

I agree with this, and think it is an important point for the Dev Team to take note of. That said, I've pretty much given up on the idea that Sil will keep up to date with V.

Redesigns of the V code can help *future* variant designers but rarely help existing variant maintainers.

I'm sympathetic to this point of view, but I'm not sure what the solution is. On the one side there's stagnation for V, on the other side variant maintainers are left in the dust.

It's hard for me to think that moving more things to edit files is not a noble goal, and should be pursued. I originally got into Angband because it was customizable without any compilation or code changing. Sure, I couldn't add new spell effects. But, I could add things like rods of teleportation, or play with monster drops using DROP_GOOD and DROP_GREAT. Two things I remember doing before ever winning my first game.

Antoine. What if the name doesn't change but the description does? So it's not called a "sharp broadsword" but it has a description line oon the Inspect screen that says, "it is sharper than usual"? Is this still a flavor break? One of the things I miss in current V is the "strikes undead with holy wrath" lines for the less exciting, "does 250 damage per round to undead". I'd like to find a way to bring little things like that back.

[half]

I promise you, the changes being made are making it *MUCH* more difficult and frustrating to maintain a variant.

I agree with this, and think it is an important point for the Dev Team to take note of. That said, I've pretty much given up on the idea that Sil will keep up to date with V.

Redesigns of the V code can help *future* variant designers but rarely help existing variant maintainers.

[Antoine]

I also think it would be good if there was a strategy for balancing object distributions. I see a tool (the object simulator) but I'm not clear what exactly is going to be done with it.

A.

[Antoine]

>> The problem is that the names have flavour, a flavour different from the original.
> Ok, then they can be changed to match the flavour of the original. That isn't impossible.

So my suggestion is "someone should do that"

What did you think of my idea of "silent affixes" that modify the properties but not the name or description? That should prevent the proliferation of "sharp swords of craftmanship" [i.e. sword (+6, +6)].

>> That wasn't my argument in this post.
> Yet it is *precisely* what you went on to say in response to fizzix.

This is getting into an attribution snarl.

You thought I thought "that Angband really shouldn't have more variation in [non-artifact equipment] than exists in the current ego_item.txt"

I replied that, instead, "any increase in variation in ego items should be very carefully considered". And I suggested to the excellent fizzix that you should not "include any affix in 3.4 that was not (in some form) in 3.3".

So it is a matter of how much you change, and how fast; and my suggestion is that 3.4 should change the range of items as little as possible. Wide-ranging changes to the distribution of items will be bad enough.

I can handle a "mithril banded mail" or "rusty plate mail" if necessary, or maybe even a "two-handed sword of slicing" at the outside, but if Sethos's "giant bastard sword of cruelty" or "midnight mace of majesty" pop up any time soon then I will wash my hands of you.

A.

[Magnate]

The problem is that the names have flavour, a flavour different from the original.

Ok, then they can be changed to match the flavour of the original. That isn't impossible.

That wasn't my argument in this post.

Yet it is *precisely* what you went on to say in response to fizzix.

[Nick]

Will that be AngbandBase?

Um ... yeah ... kind of.

What I hope will happen is that a bunch of low level stuff (file handling, parser, rng and so on) will become a library, and essentially that library will be the evolution of AngbandBase. Variants should then be able to have hooks into that. Then there will be the UI code, and the game (whether V or variant) should be able to tell that what to put to the screen, and accept appropriate commands from it.

Knowing where to draw the boundaries between these things is the difficult bit. That and stopping the dev team messing with it

[Antoine]

I think, too, that at some point not too far away Vanilla will get to the stage where there is a portion of the code (low-level stuff + UI/ports) which will be more or less distinct, and so variant maintainers will essentially just have to make sure they can 'plug in' to that, and immediately reap all the benefits of latest ports to multiple platforms.

Will that be AngbandBase?

A.

[smileyy]

I think, too, that at some point not too far away Vanilla will get to the stage where there is a portion of the code (low-level stuff + UI/ports) which will be more or less distinct, and so variant maintainers will essentially just have to make sure they can 'plug in' to that, and immediately reap all the benefits of latest ports to multiple platforms.

I find it a curious and...quaint...artifact of the game's development history that this is not the case, so many years later.