Tuesday 21 March 2023

FE Heroes theoretical Grand Conquest gameplay(?)

Rare sight of an uncontested knocked out area. Third round grand conquest is always a wild one.



It has been while since I wrote anything about FE Heroes as I have no interest in covering metas and crying about how many times I ended up with 20798 lifts in Aether Raids. The most recent one was about voting gauntlet written in 2021 but just like the last AHR we had, people just don't care as much anymore. So today let's look at another mode where theoretical behavior differs so much from reality.

Grand conquest.

In the past infernal difficulty is almost impossible with the warps. As time goes by, the difficulty goes up by including more inflated units into the pool. That however, does not take account into the facts that (1) old units are still there meaning that the average opponent strength is not inflating as quick and (2) recent units also enables more disgusting combos with such potential inflating faster than pure unit strength. As a result, those 'top difficulties' became much easier. Similar phenomenon happens in most PVE modes. Remember 10th Stratum in 2016, or Veronica as final boss in the first Tempest trial?

For inferal difficulty with warps, my solution is to use units with multiple actions so that I can overwhelm a certain front constantly in a single turn so that opponent can't warp. Summer Edelgard, Lynja, Ninjorrin and many more disgusting units would do the job well as long as terrain allows early aggression. 

But I am not going into that in detail. My question is, how would grand conquest looks like in an ideal play? Here are the assumptions and simplifications:

Assume that players would score 2000 per sword consumed. Assume that teams consist of equal number of players who will play at all time and waste zero resources. WLOG assume that scores do not flow to neighboring area (because you can adjust number of times you participate in those areas to achieve the same score regardless) and ignore the 'help out' function because that help would be of negligible effect when everyone scores so much (that will be an interesting random perturbation under a perfect game from both though). Ignore also the mechanism that scores multiply with players participated in the area because I believe it's worthy to invest at least an epsilon in every area or otherwise opponent is always happy to take the free area. 

This is now the game where each player starts with K units of resources in the reserve (which do not recover, resembles the swords). In each turn the teams take turns to assign resources onto contested area (say, teams take turns to assign 1 unit every time). Each turn players can distribute at most N (= 4 x head count) plus arbitrary amount left in the reserve into those contested areas. At the end of each turn, area with opponent spending more onto it would have the ownership flipped. The score calculation would be the same as in FEH: summation of [area hold/5] plus area count at the end. 

How would the game look like then?

...the more I think about the possibilities the more complicated it could be like, that at a point I simply give up.

We simplify by thinking of a game of two players (maybe with 20 areas). At first it's of course the concept of choke points and defending the choke points, so you want to reduce the number of points that you are defending to increase the average resource that can be put in one area. You may also think of going into an equilibrium where it's not worthy to take any more lands except for the last turn, and the examination can go much longer.

However all these become useless with the extra resources K assigned to each team at the beginning because they can then breakthrough the defense with much higher firepower. This is now a game of distributing proper resources in each turn to balance between offence and defense.

Similar problems occur in real life, quite commonly actually. One practical example is to participate an auction with limited amount of money. Assuming that each participant has equal amount of money and each auctioned item has a fixed and transparent intrinsic value, how do you maximize profit in the auction? 

Going into these 'games' would be extremely complicated and far beyond what we want to discuss here, so I decided to cheat a bit.

What if teams are allowed to react against opponent's resource distribution bit by bit within the same turn? Say, when a team is to assign 1 unit of resource onto an area then the other team can assign 1 unit after the first assignment. 

Well then this will be a draw as long as the team with the second move replicates everything. Zermelo then tells us that since player has a forced draw strategy, there will be no winning strategy.

Everything is going well on a two player game as this is the classical setup. What about a three player game then? 

All classical tools broke down here and I have absolutely zero intention to head into an active research  field. But my guess is that it depends on the maps. In a two players game, the contested area is always symmetrical to both, hence the draw. In a three players game this is sometimes not the case, especially when two teams combined always overwhelm the third one.

We can generate perfectly symmetrical map (like the star graph if you like) where by symmetry will end up in a forced draw. There are also one-sided maps like three teams based in the three vertices of a V-shaped map where the two teams will simply squeeze the middle team to death.

There are configurations/states where forced draws will happen from there, which we call that an equilibrium. Equilibrium may occur with unequal scores like 9:10:11 or 5:9:16, as it depends on the contested areas only.

There will be maps with asymmetry that a team will start with negative maximum possible expectation, meaning that they are going to lose somewhere they had at the beginning. Since trading (between two teams) is a waste of resources, the team would cleanly lose that area and fight for an equilibrium afterwards. Similar strategy will be adopted by the team who are supposed to gain that piece of land who will quickly take that and go for the equilibrium from there. 

The last paragraph is surely the most shady argument/guess. One immediate question is: is it possible for an oscillation to happen? The answer is yes and obvious if you set K=0 and resources available to 3 (instead of 4) then by parity there will be exchange each turn. If we take cyclic states as an equilibrium as well, then I ultimate guess is that given enough rounds, all maps will end up in an equilibrium (a 2:14:14 in the V-shape map is also an equilibrium!).

***

I think I have my time wasted in writing all these because grand conquest will never be played in such an ideal/competitive way. It is still more fun to play the mode as it is, rather than giving the team a chatroom probably temporary discord server as well with players monitoring and reacting 44 hours nonstop.

FEH is really declining with less groundbreaking updates, lower reaction against new units due to saturated barracks, less votes in popularity contests like CYL and AHR, PVP modes completely scaring away most players and so on. I would still support the game till the end, but please give me a premium unit that is +10-worthy both in terms of art and competitiveness. Not now, but when I have enough orbs...

Wednesday 15 March 2023

15/3/2023: WBC感想

夏天……已經結束了。

在甲子園會這樣說,可現在是經典賽2023。對澳洲來說還是夏天,但他們的夏天還沒完也還沒出局。

台灣打出2勝2負的戰績,在謎一般的分組裡僅以失分率包尾出局。不過考慮到宇宙夢幻美國隊居然慘被墨西哥打爆而大失分,如果他們同樣以失分率出局的話說不定下屆就會用分差之類的決勝負了。

其實去年我這個吱迷就因為眾所周知的原因比較少看了,不過作為半雲球迷來到WBC怎也得看看。以下是各種碎碎念,相當多的觀點已經由各家媒體評論人鄉民提過了就是。

台灣的打擊能力已經在近年國際賽中得到驗證,包括P12、亞職和這次的經典賽。中職打者如林立、四爺和小可愛等也打出非常好的表現,說明種花頂尖打者打更高階投手並非無一戰之力。中職近年的洋投等級穩步上升與打者的表現絕對有極深的關係,這次古巴隊中壓制種花隊的先發正是來年樂天的洋投之一。到底打者可以進化到那個程度,讓我們拭目以待。

與打者相比,投手明顯是種花隊的短板。只帶14投打完第一天就知道是作繭自縛:如果是季賽的話即使在WBC投手限制下打四場這些投手都十分夠用,但先發不會投長加上緊急換投需求大增再加上強制投2休1的規則下調度就顯得很困難。不過你也要問,中職除了養傷中的徐若熙、古林和曾仁和以外還有多少能夠應付高階打者的投手呢?看那些平時在中職見一個K一個的在WBC每一球都彷彿用盡全身力氣就可見一斑。但在這資源有限的情況下投手群已經做的不錯,當中呂寶、吳哲源、宋家豪和agent of chaos的表現都留下了深刻的印象。老虎有點可惜……在調度有限下配到個不會接側投的捕手。中職投手養成機會和強度皆不足已是長期的結構性問題。現在正值打出成績之時,有沒有機會推動改革就看民眾的熱情能否推動球團和政府了。

正如上屆以色利暴打預賽各隊一樣,WBC每一次都提醒我們在高階聯盟混的投打對戰起來就是強上一截。對種花隊來說日職的強度也遠超中職:強度破格準MLB級的張育成不說,日職平均以上的吳念庭和宋家豪在種花隊中已經贏過一票的中職all star了。要追上日職當然是奢望,畢竟經濟人口體量就擺在那。不過如上所述,謀求進步總是可以的。

說到日職,大王啊大王……你怎又被世界的惡意卡住了?正如大艦巨砲大所言,他真的是壞了。這樣還不如回爐中職重修……

接著說說同組各隊。這一組之所以可以全員2-2,在我看來大家其實都有著類似的投手問題:能應付WBC的投手不是沒有,但就只有這麼多,用完就沒有了。近的看荷蘭有幾個美職投手,對台灣卻只能用荷職ace,因為等級差太多被打爆;遠的看英國和加拿大可以在高手如林的pool C裡面斬獲勝利(英國那一勝的先發也是來年中職洋投),對上美國也只能分別推出大學生和1A投手先發,毫無懸念的被打爆。其結果就是,那些在WBC國藉認定下拉伕出來的隊伍和不與美日職深度結合的隊伍都有投手深度的問題、導致戰績飄忽成就我們看到的大混戰。不過誰不愛看混戰呢?

再看在日本的B組。日本這次可說是夢幻隊了,大谷達比山本佐佐木的先發堅不摧加上鐵牛陣一個比一個猛,打者則不知道甚麼時候也從細膩野球換成了火力兇猛路線,相信對上任何包括美國隊在內的對手都有十足底氣。不料捷克和中國隊卻暴露了他們的弱點--慢不打。那些進四強的美洲隊是否後悔沒帶一個能丟120的投手了?不過不算慢不打的問題,日本隊幾場下來還是偏慢熱,四強決賽要留意這問題。

中、捷乃至澳洲都是弱在投手深度不足,對上日本可以抗個兩三回,再多就只能被單方面暴揍。中捷對上強一級的澳也是同樣情況。捷克還有可以三球三振大谷(笑死)和鎖澳洲5.1局的投手……可惜了。中國隊的話只能說有進步,但想要摸到WBC強度的邊的話不向大眾推廣擴大玩家基數則絕無可能。

韓國?嗯……你們還是先想想要游回去還是直送北韓吧。他們對上日本會輸這麼慘還是投手深度不足,但為甚麼一個擁有十隊聯賽國家的代表隊投手深度會不足呢?有關青黃不接的問題,我總是會想以他們的人口來說十隊與日本相比會不會太多?那台灣六隊又是否太多?這對台灣來說是一個警號。不如無論如何現在它們看見種花隊大概要躲著走了,不管賽程安排如何如何,把台韓調一下以今年台灣的表現進八強甚至四強都不是不可能。

最後說一句,大英的球衣真醜。