Make improving spells less weird

[Stanlee 3,434]

Continuing my trend of suggestion threads that explain esoteric mechanics, today we have: improving enchantments!

 

Some of you might be wondering, in a game that's all about rolling 1-100, how enchantments over 100 work.

The answer is Benediction, it's a +5 to the casts so you can get 105. Okay, see ya!

Oh, how do we manage casts above 105? Well, that's handled by this code here

 

  public void improvePower(Creature performer, float newpower) {
    float mod = 5.0F * (1.0F - this.power / (performer.hasFlag(82) ? 105.0F : 100.0F));
    setPower(mod + newpower);

 

Flag 82 is Benediction, newpower is the cast you just made, power is the cast already on the item (at least I assume so, I don't see it establishing this.power anywhere, and if it wanted to use the cast you just made it'd use newpower).

So, how do we translate this? I translate it thusly: for every 21 power below 105, you get an extra +1 to your casts

But there's a catch: it only runs this code if it's trying to improve the power of a cast - if it's first time or you just dispelled, it's a +0

This causes me to use a system by which when a spell I'm trying to get high reaches a certain power, it's dispel time. And the cast after that dispel is a dud.

 

So here's how a graph of this looks

 

This is troublesome for a number of reasons.

1) It's counterintuitive. Usually improving takes something good and makes it better, turning your 89QL item into 90QL, but for this the best casts come only from the worst ones.

2) It causes a complete mess with Dispel. Want to improve a CoC cast? Hope you're ready to remove the WoA in the way. Or in the case of mailboxes, hope you're just ready to make a new one, you can't dispel those.

3) Dispel sucks.

 

My suggestion is thus, update the formula to this:

  public void improvePower(Creature performer, float newpower) {
    float mod = 5.0F * (this.power / (performer.hasFlag(82) ? 105.0F : 100.0F));
    setPower(mod + newpower);

This has a number of advantages

1) It's intuitive. Better casts beget better casts. 110 casts are now possible! (For instance, 108 cast item gives +5.14, +5 from Benediction, 99.87 cast required so it's still very far from reasonable)

2) You can reasonably improve casts that are dispel blocked. This gives a lot more functionality to things like anti-shatter juice, which in current iteration can cripple casting potential for things like CoC.
3) No more dispelling when trying to improve casts.

[Ekcin 2,406]

Sounds reasonable. +1

One question: How important is dispelling? I mean not necessarily for 100+ casts by a channeling 95 to 100 priest, but for a more humble one, in the low 80s, though with benediction?

[Lethyria 161]

Sounds interesting like an interesting concept, and seems like  a nice alternative to dispel changes, and more intuitive for priests. 

One small thing to be considered: Isn't the improve cast done after the power roll is greater than current power, so you'd have to change it to calculate the recast bonus beforehand. (Otherwise you could never recast a 105+ if thats the intention)

 

15 minutes ago, Ekcin said:

One question: How important is dispelling? I mean not necessarily for 100+ casts by a channeling 95 to 100 priest, but for a more humble one, in the low 80s, though with benediction?

Dispel works on different code, the dispel power required is a random number between 0 and power+difficulty on a uniform distribution so on average half the power + difficulty so shouldn't be too hard to remove.

 

Edited May 7, 2022 by Lethyria

[Ekcin 2,406]

Thanks Lethyria. But what I meant was not the dispelling proper which works normally. Rather the importance of dispelling vs. going on casting on top of the enchant until a better result appears. If I understood correctly, the best results (100+) can not or not with sufficient probability be reached without dispelling, or not?

[Stanlee 3,434]

4 hours ago, Ekcin said:

One question: How important is dispelling? I mean not necessarily for 100+ casts by a channeling 95 to 100 priest, but for a more humble one, in the low 80s, though with benediction?

 

Let's take a look at some numbers!

Using the simulation tool we can easily simulate 5 million CoC casts at 80 channeling (note: I have not toggled Benediction for this, as it causes some errors - the numbers I will use are adjusted for Bene, but this graph is not)

 

We get the following information:

Your mean cast is 38 (selection: -5 to highest. Casts below -5 (0 with bene) won't be ones you ever have on an item, so 38 is the useful mean). This gives a good number of your casts about a +3

 

Now let's go with a standard situation: you have an item with an 80coc cast, you're aiming higher, and you're wondering if you want to dispel.

That 80 gives you a +1 to your casts (1-(80/105))

Assuming no further cast improvements, expected favor to land 90+ is ~1831, 95+ is ~3474

But if you dispel and regularly get that +3 your mean cast suggests you could get, expected favor for 90+ is ~1474, and 95+ is ~2638

This would be about a 20-25% saving in favor

 

However the requirement of dispelling (which can easily rack up the costs due to repeated dispel failures) and the first casts (+0 and may land well above mean, themselves needing more dispels) means the saved favor is likely far lower (to the extent that I'd question if you actually save any). Finding more accurate numbers would require better tools than what is available. If anyone's a programmer, reach out, maybe we can make such a tool.

 

tl;dr it reads like it might be good for lower skill priests, but chances are good it isn't

Edited May 7, 2022 by Stanlee

[Ekcin 2,406]

Thanks a lot!

[ArdraTouborg 0]

Hey thanks for the super helpful reply. I'm not sure how I missed that thread. I haven't quite mastered the search function on here. I think I'll pass with him this time around. If i want any further guideline we will contact you here https://forum.wurmonline.com/index.php?/topic/190479-make-results-improving-spells-less-weird/

Edited May 8, 2022 by ArdraTouborg

[Lethyria 161]

14 hours ago, Stanlee said:

tl;dr it reads like it might be good for lower skill priests, but chances are good it isn't

 

 

That's a really nice write up and it made me think, which is actually better. So I grabbed the simulator code and I've run some simulations. These could contain errors but I thought I'd publish my results. 

For a 90 channeling without bene priest targeting 90+ 

Scenario						Simulations	Total Favor		Total CoC Casts		Total Dispels		Shatter Rate
Channeling	Bene	DispellThreshold	Bonus	TargetPower	ItemQL	Count	Mean	Standard Deviation	Mean	Standard Deviation	Mean	Standard Deviation	Percentage
90	FALSE	120	70	90	90	1000000	1884.80355	1927.13903	37.696071	38.54278059	0	0	0.024078
90	FALSE	80	70	90	90	1000000	1771.81036	1760.328537	34.702979	34.46202375	3.666141	4.892349651	0.021769
90	FALSE	70	70	90	90	1000000	1747.20379	1722.91571	33.163638	32.63694912	8.902189	10.17218008	0.020778
90	FALSE	60	70	90	90	1000000	1785.29259	1754.824708	32.64765	31.97002326	15.291009	16.51151645	0.020743
90	FALSE	50	70	90	90	1000000	1886.67903	1854.613106	33.222097	32.48616683	22.557418	23.72838891	0.020938
90	FALSE	40	70	90	90	1000000	2029.96437	1998.969093	34.556729	33.82767713	30.212792	31.30089483	0.021989
90	FALSE	30	70	90	90	1000000	2207.1165	2181.822823	36.562311	35.92449874	37.900095	38.97977077	0.022964
90	FALSE	20	70	90	90	1000000	2393.23206	2370.727933	38.840981	38.24951658	45.118301	46.15945039	0.024163

In these results it looks like dispelling at between 70 and 80 is optimal here, although the favor saving is relatively minor at ~7% but the variability drops too, all this lines up with my experience. Next I decided to look at my use case which is usually targeting  100+ with my 98.8 channelling priest with bene, cutting down on scenarios to save time.

 

Scenario						Simulations	Total Favor		Total CoC Casts		Total Dispels		Shatter Rate
Channeling	Bene	DispellThreshold	Bonus	TargetPower	ItemQL	Count	Mean	Standard Deviation	Mean	Standard Deviation	Mean	Standard Deviation	Percentage
98.8	TRUE	120	70	100	90	1000000	2886.3844	3000.008491	57.727688	60.00016982	0	0	0.013358
98.8	TRUE	90	70	100	90	1000000	2548.98228	2545.011081	49.8092	49.71814994	5.852228	7.083256003	0.011675
98.8	TRUE	85	70	100	90	1000000	2473.30198	2455.686937	47.541414	47.16263779	9.623128	10.86668899	0.011018
98.8	TRUE	80	70	100	90	1000000	2440.47793	2417.195517	46.047765	45.53853218	13.808968	15.05200888	0.010539
98.8	TRUE	75	70	100	90	1000000	2444.37644	2418.256861	45.218352	44.62989261	18.345884	19.60044113	0.010506
98.8	TRUE	70	70	100	90	1000000	2484.23439	2455.592028	45.033647	44.38502273	23.255204	24.46384952	0.010455
98.8	TRUE	70	70	100	90	1000000	2480.2323	2452.238948	44.964092	44.32027778	23.20277	24.44914453	0.010416
98.8	TRUE	60	70	100	90	1000000	2625.29794	2592.566541	45.771267	45.01729748	33.673459	34.81578814	0.010475

 

Here it seems that 80 is about the optimal dispel point (15% favor saving), with a slight improvement in total aggregate shatter chance at 70.

 

If anyone is interested in other scenarios let me know.

Edited May 8, 2022 by Lethyria

[Oblivionnreaver 5,174]

good idea yus

[Oblivionnreaver 5,174]

ayo add this