Test games played

[Arnuz]

That? With Halflings? I would probably lose without a good uncommon creature.

Is the purpose of the tests to check whether the game should be doable on expert with random wizards for skilled players? Always, 50%...?

[Nelphine]

This is a good question. Where do you want game balance for each difficulty? Can we break it down as we have been doing based on synergistic retorts?

[Arnuz]

Met a lawful perfectionist Sharee, conjurer charismatic 4L 6D and brought her up to calm with trades and presents... But she won't get in a pact ;(

Started the war with Tauron. I have 8 ghouls with resist elements. Tauron is at 2 mana? Haha, nice!

[Arnuz]

Tauron keeps being at no mana by corrupting my lands, so far the conquests are easy. Second city conquered.



Not found his capital yet but I've seen the direction his settlers take.

[Seravy]

I'd say a random wizard needs to win 100% on Normal, 100% on Advanced if played by a good player, 80% on Expert if played by a good player and maybe 30-40% on Master.
A below average wizard should always lose on Master and win ~50% on Expert, 75% on Advanced, 100% on Normal.
A non-random, but average wizard should be similar numbers, maybe a bit higher - 90% on Expert and 45% on Master?
A non-random, good wizard (say, 10 books 2 retorts matching the strategy, a race matching the strategy) should be, assuming a good player, 100% on Expert, 75% on Master, 10-20% on Lunatic.
A non-random, "best" wizard (7-8 books, 4-5 retorts in perfect harmony with the most suitable race) should be ~95% on Master, 30-40% on Lunatic, assuming the player is not just good but also doesn't make any significant mistakes.

So far I lost 3 out of 6 on Expert so it doesn't seem to meet that goal, but it's clearly not enough data.

Problem is, it's very subjective what counts below or above average... it's no longer the time when we could just point at 3 retorts and say "each of those you add means you can beat 1 more difficulty level than otherwise" - the game is too balanced for that.

[Nelphine]

You might want specifically to point and say 'this should work'.

Start with wizards that have no retorts. Where should the various combinations fall? Should 4/4/4 be as good as 6/6 and in turn as good as 10/2? What difficulty should that be 100% and 50%?

Then go through the 'standard' wizards, meaning the ones the ai are designed around. Where should they fall?

Once we have a baseline, we can go forth to see if it matches they, and determine what happens when you add more and more retorts.

[Seravy]

Ok let me see. All of these will assume the player is playing well and doesn't do large mistakes such as sending a good army into a lair full of colossus and wasting it all, or declaring war on someone they can't beat or forgetting to garrison an important city, etc.

Weak - Zero synergy and picks wasted on books or retorts that don't contribute at all plus a race that works horribly with those.
4 Chaos, 4 Sorcery, Artificer, Runemaster, Specialist with High Men.
2 Chaos, 2 Death, 3 Sorcery, 1 Nature, 2 Life, Guardian with Klackons.
10 Chaos, Omniscient, Arfiticier, Halflings.
4 Life, 4 Death, Guardian, Myrran, Gnolls
100% Normal, 75% Advanced, 50% Expert, 0% Master or Lunatic.

Average - 10+2 books, or 2 retorts that have no synergy but aren't entirely useless, or 6+6 books. A race that neither helps nor hinders it. Probably 4+4+4 books falls into this category as well.
6 Chaos, 6 Sorcery, Klackons
6 Death, 6 Life, Nomads
10 Sorcery, Astrologer, Cult Leader, High Elves
4 Life, 4 Nature, 4 Sorcery, Orcs.
100% Normal, 100% Advanced, 90% Expert, 50% Master, 10% Lunatic.


Good - 8-10 books with 2-4 picks in something else that works well with that realm, or 5-5 with 2 in retorts. A race that benefits the strategy.
8 Sorcery, 2 Nature, Sage Master, Specialist, Halflings.
8 Life, Myrran, Warlord, Dwarves
10 Nature, Specialist, Conjurer, High Elves
5 Chaos, 5 Sorcery, Chaneller, Nomads
5 Life, 5 Sorcery, Famous, Artificer, Halflings
9 Life, Warlord, Tactician, High Men
100% Normal, 100% Advanced, 100% Expert, 75% Master, 25% Lunatic.

Top - 7-8 books in one realm or 5+4 in two with 4+ picks in retorts that work well together, above average race that support the strategy - or a "good" strategy with map settings that exploit it.
8 Nature, Specialist, Conjurer, Spellweaver, High Elves
6 Life, 2 Sorcery, Warlord, Tactician, Alchemy with Barbarians
8 Chaos, Runemaster, Specialist, Sage Master, High Elves
8 Chaos, Myrran, Specialist, Sage Master, Draconians
5 Sorcery, 4 Life, Alchemy, Archmage, Tactician, Gnolls
Max power map, 3 Nature, 7 Sorcery, Archmage, Sage Master, High Elves.
100% Normal, 100% Advanced, 100% Expert, 95% Master, 50% Lunatic.

Something like this.

Standard wizards...heavily depend on how much retorts they pick and which race.
If they pick only their primary retort and a generic race like Orcs, they should be the same as "Average" above. With both primary and secondary, most should be "Good" although there are a few weaker than others.
If the AI picks both retorts plus something else with positive synergy and a matching race, it can reach "Top" but the realistic chances of that all happening are near zero until the AI starts to get extra picks.

[Arnuz]

Uh, we differ completely on our top strategy :D

I propose that at the beginning the player will declare a likelihood of victory, based on his feeling: the best player with a strategy he's unfamiliar with wouldn't really work.

In this case I got ghouls, albeit without FM and I know their limits. I've captured 2 nodes and a lair which kick-started me, and the gold from the conquests kept me rolling. I have 4 shamans cleaning up Tauron's mess. I would have given myself a 50% at start but seeing how I could go to war with a chaos wizard who starves himself of mana I'm raising to 80% (based on the history graph and the 2 wiz pacts).

Meanwhile, my war against Tauron continues, I have 12-15 tiles corrupted but I've taken several towns from him, and finally found his capital (south city in the map). I've given lots of presents to Sharee, including gold, and she's finally accepted the wizards pact.

[Nelphine]

Sorry my point wasn't for the AI to perform well, I just wanted a baseline of where you think things should be.

So one thing, you think an average should win lunatic 20% of the time. I don't think that will happen. Winning lunatic right now requires extremely dedicated wizards, and I think most average would have 0% chance of winning.

So let's try to specify category a bit more:

Where should 4/4/4 fall? What races should change that if any?

Where should 6/6 fall? What races should change that if any?

Where should 10/2 or 10/1/1 fall (no retorts)? What races should change that if any?

Should different realms make Amy difference?

Then, the next set is 5/5 with ideal retort (as determined by the AI who use that combo by default).

10 with ideal retort

8/4

10 myrran

4/4 myrran with ideal retort

5/5 myrran.



I think those are all the baseline builds a human can play. If we can determine how those builds should do per difficulty we can go from there. Note that none of these have more than 1 retort yet.

[Seravy]

Good catch, the numbers for Lunatic are unrealistic. Haven't played that in a year. Will change it.

Define "fail". Below 10% wins? Below 50%?

4/4/4 and 6/6 should be equal as well as 10/2. Not sure about 10/1/1 I think that's inherently worse and I'm fine with that. Note that in case of 10/2 a very good and relevant common starting pick can make it count as good as if those 2 picks were 2 good retorts instead, bumping it up a category. (such as Sprites, Wraith Form, now probably Blur, etc)
But if the starting spell isn't game-changing then 10/2 should be equal to the others.

Note that human ideal retort(s) do not equal AI "ideal" retort(s). The AI plays entirely differently and the sets have been designed to make sure there is no excessive redundancy and every retort appears at least once but no more than 3-4 times. Neither is true for the human.

Race and realm makes huge difference. Some realms work well together, others not so much. Same for races. But we can say every race or realm should work well in at least some combination.

Also, a lot of races or retorts do better or worse on certain map sizes, power or mineral settings etc.