For anyone interested in object pricing issues

[Magnate]

I'm out all day today, but I've committed the latest version of the power algorithm to trunk (r1358). If you're interested in critiquing the new pricing system, please compile and test with this version. It makes adjustments to ratings for CHR, ESP, rnexus, brands and slays on rings and launchers, additional power for multiple resists and sustains, and power for activations. I'm particularly interested in the costs of:

Foo of Resistance
Robe of Permanence
Shield of Preservation
DSM
Defenders

I think people will still consider Foo of Speed underpriced (and Foo of Slay orc/troll), and plain launchers overpriced - but most things should be ok now. Pls just bear in mind that the system is designed to be consistent with itself, not with previous prices - if things are cheaper than before it means you get less money for selling, so you'll have less to buy with. Please try a full game or two rather than reacting to individual prices.

If you want to see what power rating something has, look in the pricing.log file in your angband dir (the one where your dumps get written). I'm not sure exactly when the value routine is called, but you may need to attempt to sell the item in order for its power to be calculated. For an easy comparison if it's a weapon, divide the power by two and consider a weapon doing that much damage per blow, and if it's not then divide the power by four to compare with a ring of that much damage. Those are the rules of thumb which we're attempting to use, so if there is a big difference between how much you value the item and how much you would value the weapon or ring of equivalent damage, post here with the item and its power rating and we'll see what went wrong.

Apologies in advance if the pricing.log file gets too big (just delete it and it will restart) - I've not yet worked out how to make it log more selectively.

[Magnate]

P.S. There is one whole area which isn't covered yet and remains on my to-do list, which is weight. At the moment a 40lb suit of plate has the same value as a 4lb shield of the same AC, and this is clearly wrong. I do plan to adjust power for AC by weight, and hope to get this in for 3.1.1 (though it is non-trivial in integer arithmetic!). As previously noted I'm not planning to adjust power for weapon weight, because the tradeoff between extra blows and extra criticals is currently too difficult to calculate.

[Nick]

I think my main problem will be deciding what to wield in the foo slot - there seem to be just too many choices.

[Optimality]

I noticed in the code that your resist tables have rconf at 24, and rchaos at 20. Doesn't chaos res give conf res as well? I put rconf at 8 on my chart.

[Joe B]

Doesn't chaos res give conf res as well?

Nope! I learned that the hard way.

[Optimality]

Nope! I learned that the hard way.

I must have variant confusion . Still, I think 24 is way too high.

edit: apparently it used to, but recently changed.

[Magnate]

I must have variant confusion . Still, I think 24 is way too high.

Well, the consensus on rgra was that rconf was the second most important resist of all, after poison, so it is currently valued at 2nd most powerful.

Things may have changed, but rather than reopen the "which resist is better" debate, I'm going to calculate them empirically (relative to each other) after I've sorted out armour weight. There will still be a debate about how much resists are worth in terms of equivalent damage, but at least there will be none about their relative importance ...

[Atarlost]

I don't think you can just go by how much damage it will save you on average. Certain elements have non-damage side effects. The slowing of inertia and gravity can theoretically be valued, but the blinking effect of gravity, the effects of nexus, and the confused player state have incalculable value. rConf is considered the most critical resist because confusion is the second most crippling status effect and the more crippling paralyze isn't grouped with the resists.

[Pete Mack]

I am finding myself more and more in agreement with Rizwan's rarity argument, because I don't think it's universally held that "desirability" may reflect rarity as well as power.

I had assumed that things like RBase, SI, FA, and probably RNexus should be cheap because they are widely available, both in artifacts and ego items. (Examples: Boots of Stability, gloves of Free Action, Bows of Nazgul, though the latter are unfortunately exceedingly rare.)

Rizwan's original example of speed rings vs rings of Acid was a bit misleading in that both are quite common at their native depths; speed ring is thus expensive mostly because of power rather than rarity. Speed boots, OTOH, are both powerful and rare, and should be valued higher as a result.

My original post about RNexus was more of a devil's advocate: for pricing because I think it should be fairly cheap and common, and for stat gain because I think it is broken.

EDIT:
Note that to some extent pricing is dependent on rarity, because damage from non-weapons is valued twice as high as from weapons.

[PowerDiver]

Why don't you free yourself from this pricing mania? The ego items provided by shops are a lottery. IMO egos should not be offered in the shops, but once you go down that path there's no point in second guessing yourself. There is no reason the prices charged or offered should not be just as much a lottery as well. What is unbalancing is allowing ESP for sale in the shops in the first place. The difference between charging 50 AU versus 50K AU is tiny in comparison.

As to randarts, there is no point in trying to make a one-to-one power mapping onto the standarts. I would guess it is better to make a set of rules that say "object flag foo has prob p at depth d in slot s", and then add rules boosting or dropping the prob depending upon the artifact's depth and rarity. Then put each flag on each randart IID. Cleaner solutions almost always turn out better for reasons you cannot even guess at the start. I'd bet this situation is no different.

[Atarlost]

The biggest problem with rarity independant pricing is that items that give critical flags may bring in too much money. rBase, rPois, see invisible, and free action aren't rare flags and if they were valued according to how important it is to have them the lowly and common shield of resist fire would become a huge source of cash.

[Magnate]

The biggest problem with rarity independant pricing is that items that give critical flags may bring in too much money. rBase, rPois, see invisible, and free action aren't rare flags and if they were valued according to how important it is to have them the lowly and common shield of resist fire would become a huge source of cash.

That's right, which is why like Pete I too have changed my view on the rarity argument. The current system *does* reflect rarity, by making rbase and SI / FA items less expensive than their pure functional value would indicate - because they're common.

What I have a problem with is making speed on boots worth more than speed on rings. That just strikes me as silly. The damage example is different: +damage on nonweapons is worth twice as much as damage on weapons for the specific reason that it's applicable to every attack against every foe, regardless of which weapon you choose to use. Whereas damage on weapons is only usable at the expense of all other weapons - a large and significant choice. Even allowing for carrying swap weapons, offweapon +dam is worth quite a bit more.

Re your earlier post - I do plan to take side effects into account when calculating relative importance of resists. It's already in the monster power calc - sure, it's possible to quibble about the numbers attached to the side effects, but at least the idea is in play.

@Eddie: personally I agree with you about ego items appearing in shops, but I think it's worth trying to maintain some semblance of internal consistency with pricing. I don't have much time for arguments about at which point in the game you should be able to afford what (because that assumes far too much about the game's money supply, which is broken), but I do strive to make sure that if item X is widely held to be worth more than item Y then it costs more. (I also didn't think there was much wrong with randart generation nowadays - but I'd be intrigued to see how your alternative system works.)

[Polyonymous]

@Eddie: personally I agree with you about ego items appearing in shops, but I think it's worth trying to maintain some semblance of internal consistency with pricing. I don't have much time for arguments about at which point in the game you should be able to afford what (because that assumes far too much about the game's money supply, which is broken), but I do strive to make sure that if item X is widely held to be worth more than item Y then it costs more. (I also didn't think there was much wrong with randart generation nowadays - but I'd be intrigued to see how your alternative system works.)

This was my point earlier. It doesn't matter how self consistent object pricing is. Until the game is "balanced" (against whatever standard), having consistent prices. As long as the selling price is close to the buying price, it's not at all hard to have enough cash to buy out of depth items (shop price caps stop this somewhat, but only for things over the cap)

You are trying to make it cost a reasonible amount more. Why not use a=0? That would satisfy the relative pricing goals just as well as a=2.

I've been assuming that you're trying to make them reasonable against an eventual balance goal. Pricing consistent with that goal will produce feedback about the what's mispriced. Which will in turn point you towards some things or many things that need to be better tuned for balance. What is that eventual balance goal? (I'm not asking for perfect precision, general guidelines, etc. It'd make a good discussion, I'll open a thread for it)

Here's some things I've thought about in light of a pricing function, might need to be rebalanced, etc. (I'm not touching the dungeon based suggestions)

If the formula is balanced, shop sale price caps have no point except to stop the buying of fairly priced items. Likely the trade in value of items needs to be a much smaller fraction of the value (or fewer dropped relative to bought). This is definitely true for the portion of the game where the price caps aren't a factor. The ratio of items sold to items worn is pretty high (arbitrarily high, it's a speed of play choice, what is the game balanced for?) Also, Following up on Eddie's comments, if you can buy random egos, there's not much rationale for not buying designer egos potentially at a sufficiently high price increase.


On items that scrolls of Armor/Hit/Dam function, it's probably not worth pricing the various plusses as more than the equivalent number of scrolls. Shops should run sales...

On sale now!, Dagger of Slay Orc (1d4) (-10, -10) with 20 To Hit and 20 To Dam scrolls. (or whatever tradeoff of +stats and scrolls that minimizes expected price)

Given how pointless that seems, the first couple plusses (up to where scrolls get ineffective?) likely should be dirt cheap (cost about the same as the required number of scrolls and not be included in the power calculation). Is the alchemist prepared to handle the additional demand? Scroll prices should be up significantly. Making enchant success rate relative to the increase in prices is the only way to make it "balanced" (scroll costs 100, power based value cost increase on success is 500, success rate should be at most 20%, 20% or higher and the optimal sale metric for a consistent world is for players to use scrolls up to their desired power level). That does make some sense in that more powerful items. The other alternative (Dagger of Slay Orc (1d4) (-10, -10) or wherever the incremental power based cost exceeds the scroll costs) and spending time with a bundle of scrolls seems worse.

Also, the relative scroll prices (at least dam/hit/armor) should be rebalanced in light of the relative power each one gives.

[Pete Mack]

@magnate
I haven't changed my opinion on pricing; I just assumed that frequency was already included in the power of an feature, vis: RBase, SI, FA.

As for BoS vs RoS: BoS allow you to wear a second elven ring, which translates directly to more damage. Cubragol and Ringil have the same property.

[Pete Mack]

@Eddie

I don't think there's an unbiased way to make a frequency vs depth model as you suggest. Thranduil is shallow and common. Does that mean ESP should be too? The signal vs depth is just very noisy. Price is supposed to be a proxy for this.