Apologies in advance for a very long post. Everything I say here is my personal opinion based on the experiences in the genre. All the numbers will have to be adjusted, numbers I provide are just for the sake of having more concrete examples.

As the latest Humankind beta is up and running, I have been thinking extensively about how Humankind multiplayer experience could look like. While Humankind's single player experience looks astonishing (love the strong and unique narrative of each playthrough, the diplomacy and of course the visual appeal of the game) the multiplayer implementation of it seems very trcky, given its pace and combat system.

I would love to hear the thoughts and ideas of all of you, who got to play the game's various demo versions. Here are some of my thoughts, with a little bit of contextual information.

My personal multiplayer experiences with other 4x games include several hundred hours in Endless Legend (mostly with a group of close friends) several hundred hours in default civ6 multiplayer and couple of thousand of hours in civ6 CPL (teamer games and ffa games). Here are some challenges encountered in each of the games:

1) Endless Legend - was only doable and enjoyable with a group of friends and scheduling, as the game takes quite a long time to wrap up, mostly due to the initiative-based sequential combat system, which is very similar to the combat system in Humankind. Having 3-5 combat encounters in the late game per turn, and the desire to play them out optimally would make a turn last 15-20 minutes. Which means that the entire playthrough could take up to 16h (even in a 4FFA or coop vs ai format). So then even playing with friends created bit of tension, like "Hey dude, hurry up" (several times during a long turn) and if you weren't at war you would be watching your empire and hurrying up your friends. Quite often we had to save and quit and schedule another date to play, which could take up to a couple of weeks and when you'd get back to it, the immersion would be broken and your lands would feel foreign. 

2) Default civ6 - vanilla civ6 is really imbalanced and unstable, as you had no idea in a lobby, whether the host and players are committed enough to endure the sequence of desyncs/resyncs and host drops. After the aforementioned events would occur, players are unlikely to reconnect through a steam lobby (as not all of the steam profiles are public/hard to find e.t.c). Players would rage quit quite often or feed their cities/relics/great-works/resources through the diplomacy screen. So it was practically unplayable past the early medieval era, later than that you would have 2/3 players going for the win and the rest are AI, whose combat and sim abilities are much inferior to competent players. 

3) civ6 CPL - CPL does fix the above mentioned mistakes of vanilla civ6, by actively rebalancing the civs, improving the stability and playability of the game by giving host ranks, potential lagger tags on the discord server. As the player-base grew, you could also find someone to substitute you, in case you weren't able to play your game out for X,Y,Z reasons. On top of that people can get temporarily or permanently banned for quitting/feeding depending on the frequency of these repeated offenses. This ensures that most of the lobbied games go through.

The games usually last 3-6 rarely 8 hours and most played formats are FFA and teamer (which is played in tournaments as well). I generally find it hard to maintain top shape after one long game. Assuming I play one more after, I am bound to make mistakes and play suboptimal, some of my fellow players complain when a single given game lasts too long (and they are not having the best time), resulting into forcing the concede vote, and top contenders not being able to play it out (different victory conditions take different time). So I personally think that a single match should not last longer than 8 h in a multiplayer setting. 

My main criticism of civ6 multiplayer is: in order for the game to be enjoyable, lots of enthusiasts have to work really hard to guarantee this experience. This is done without much of the support from the official firaxis devs. So it kinda felt like Firaxis made a really enjoyable single player game, and then "nailed with the big hammer couple of tweaks to gameplay to make it multiplayer".

Before we talk Humankind Multiplayer, let's establish some vocabulary in order to make the points I am trying to make a bit easier to express.

Some vocabulary:

Global Turn -  all the actions taken until next turn is pressed by all the players (or the turn is forced to end due to the time limit).

Army Avatar (AA) - representation of a given army on the global map.
War Moves (WM) - sequence of 3-4 (depending on the state of the game) moves done by each player on a given battlefield, that fit into one global turn.
SIM - city building/districting, tech-tree actions, civic decisions, random scripted event decisions.
Map Moves (MM) - moves performed by AAs on the map.
Conflict Resolution - more than one player can meet the criteria for something during the same turn. For example evangelize a belief for your religion, choice of the culture, claim of a wonder using influence.
Persistent Pop-Up (lenient/strict)- a UI element, which opens without your consent. Lenient - can be closed by clicking "remind me later". Examples: "choose a new belief in your religion", "choose your next culture". Strict - cannot be closed without completing it. Example: WM in one phase.
Global/Phase/Conflict Time-Turner (GTT/PTT/CTT) - time-turners for the global turn/each phase of the battle(like initial troop allocation, attacker moves, defender moves) and conflict queues (conflict resolution).

Finally Humankind Multiplayer:

1st Decision) Having a) dedicated Amplitude/Sega/3d Party servers vs b) games being hosted by players. While a) is more stable and is easier to debug, improve upon and control. It also costs good money every month, which probably would conflict with the one-time purchase monetization model. b) Would require additional systems in place to make it work well.

1st Strong Opinion) a) would be amazing if it is free, but no one can afford to run charity. So a) probably would require multiplayer-pass monthly subscription as means of additional monetization for server maintenance fees. As a result it cuts off a decent chunk of the player-base, which is never good.

I would opt with choice b) for but would love it being done better than vanilla civ6.

This means having a built-in proper matchmaking system, where people can join lobbies (cross-platform if possible), online players are visible and everyone has a tag: like stable host/ stable host NA only, Potential Lagger, Quitter, Sub and so on. So whenever players get into the lobby, they know whether it is worth waiting for the lobby or not. Perhaps have a formal ranking system, so we know what to expect from each other. Voice communications/Team Voice communications option.

Having a built-in lobby-voting system: on the Map-Type (and it's core settings if any: resource frequency, map terrain, sea level and such), Game Speed (if there is such an option as civ had it: online, standard, epic, marathon), Game Duration, Time Turner settings, and culture bans (Assuming there are N-players <=10 in one game, we can ban max(10-N-1,0) cultures in each era for the sake of balance).

Having a built-in additional diplomacy system: concede vote if someone is snowballing out of control (giving some other players veto rights, say most techs, most military and so on) probably should happen at fixed turns. Irrelevancy vote - give control to AI, go next (if your game goes poorly). Pause option maybe with a dropdown menu: something happened IRL, substitute player X/replace player X with AI (due to connection issue, personal circumstances)/autopause if player disconnects, vote me irrelevant, and start a scrap vote (if the game is unplayable due to technical issues and people could not agree on things), Auto pop-up "Request 10 min Break" after each 2 hours (end of the turn) (passes if at least 1 player votes yes). Save and get-together again vote: current host can mark some kind of simplified availability table and players vote.

Having a built-in post-game vote: the game is bound to end for whatever reason, someone could have quit, then no sub was found, then the game was scrapped. Or host was unstable the game was scrapped. GG WP. This would be important in order to identify good stable hosts, potential laggers, good and friendly players and ban/label - toxic/racist/sexist people and quitters.
 It is okay if balancing is done by the community via modes, votes and so on. And I think it is important that devs part-take in it, by collecting win-rate statistics for different culture builds and openly publishing it: Say if Culture 1 -> Culture 2 -> ... ->Culture 6 is too strong, it would be nice to cross reference, identify and rebalance a specific culprit culture(s). No one likes many bans as this practice takes away from the game experience.

I guess these were my 2 bits on the organizational aspects of Humankind Multiplayer, assuming the devs decide to go with the route b) (majority of the points made are still very viable for the route a)); and trust me those should not be neglected, as otherwise we are bound to have vanilla civ6 experiences.

Let's Talk Gameplay:

Duration and Pace, Implementation and Trade-offs) As I mentioned earlier, if one session lasts longer than 8h, it is problematic. If save and resume are too far apart in time it is problematic. When I say problematic. I mean for the public competitive lobbies; in private lobbies whatever people agree on doing works best. However one of the things humankind does extremely well is the narrative faster pace might mean the sacrifice of the feeling of immersion, or butchering the gameplay-driven storytelling mechanics (if you have to make a million choices each turn, you don't feel the gravity of each one of them and are likely to autopilot). From what I have seen this far at the current pacing of the beta, if everyone auto-resolves every battle and rushes through the eras, one can reach the middle of industrial era in around 6 hours (not even the win conditions). This means that the total duration of 8h per session, assuming people are playing manual battles, cannot be achieved.

 It can be solved in by having a faster progression game speed- all the costs and also the amount of movement of AAs on the map scales with the speed (more movement for AAs on the map is quite important so that the player has the same amount of territories discovered in the same eras, and war vs time relationship is the same). Here Humankind has a nice advantage over civ, as in civ war moves happen on the actual map, so giving more movement speed to units (to compensate for the faster pace of the game) just ends up breaking the units, while in the Humankind the war moves have their dedicated battlefields.

I think the majority of competitive multiplayer players have already played the single player extensively. Hence some tradeoffs to the narrative pacing can be made. More decisions/events within the same turn to compensate for the faster overall pace are likely to be accepted.

I think the time-turner that scales with the game progression (number of units in armies, AAs on the map, number of cities and Territories, as well as eras) is necessary, but more on that in the discussion on the multiplayer combat. Setting up a well thought out time-turner is crucial, so that late game global turns do not last too long.

Conflict resolution)
As I outlined before, the conflict resolution mechanic is needed for several situations in the game. For example 2 religions can simultaneously evolve "rare, but happens". If the conflict is not resolved it could happen that both religions will have same Tenets, which is undesirable. This happened quite often in civ6 upon founding a religion in a multiplayer setting as a result there were duplicates or even triplicates of the same religion on the map. More crucial is to resolve conflicts when it comes to picking culture (having 2 identical cultures would be detrimental in Humankind).

Solution a): I would propose to solve it with a persistent LENIENT pop-up. Each player fulfilling the criteria is given the position in a start-queue (start of the turn queue) according to some priority selection rule (determines the position in the queue), the first one in the start-queue gets a window saying "your have enough fame to choose a new culture", which opens right away. If you dismiss it, you get de-queued and en-queued to the tail of end-queue (end of the turn queue), so that the next person can choose in the start-queue. Only one person can have this persistent pop-up at same time in order to avoid duplicates. The end-queue is traversed when the time turner shows = current GTT length - Const*Number of Players in End-Queue, where Const is the time allocated for each player to make a choice, say 10 seconds. The players who passed it to the next turn will be placed to the head of the start-queue of the next turn.

While this solution works adding additional time-turner (conflict time-turner) would make a players feel under more time pressure (as if it wasn't enough already). It will also be quite hard to allocate the time for players and priorities of the decisions that players have to make (It is nearly the end of the turn and I need to make my phase moves on 3 battlefields and go through 2 conflict resolution queues - a bad case of "if I don't manage, my game plan might get entirely messed up").
 

Solution b) Players, that fulfill the criteria, are still en-queued according to some priority selection rule, however the persistent STRICT pop-up appears for everyone at the same time, saying "your civilization and N others, have accumulated enough fame to choose a new culture". Then each player gets to arrange N+1 cultures in their preference order. Top being the first choice. The player is asked to confirm the choice, then is asked to choose one of the 3 options, evolve now, end of the turn, next turn. The position of each player in the start-queue, end-queue, and start-queue of the next turn (this one is only partially filled up) gets resolved. If the player didn't choose anything (just stayed idle in the menu) half-way through the turn he is moved to the tail of the end-queue; if still idle at the 25% of the remaining GTT this player gets to the end of the next-turn start queue. If there is an idle player halfway through GTT start of the turn cultures get assigned, 75% in end turn cultures get resolved. Otherwise assignments are resolved when the last player made their selection.

I prefer this solution over the prior one, simply because you chose once and don't rely on other players making their choices; hence there is no need in adding an additional time-turner, like CTT.

Priority Selection Rule: When it comes to competitive gaming, it is nice to have a good balance of determinism and surprise. In civ6 priority selection was done in the 2 following ways: random - for pantheons (out of the players that hit 12 faith on the following turn, a random one will get the choice of pantheon, others can choose one a turn, again hidden and random). Second way civ6 did it, was through the slot order (how high was your slot in the lobby) - this was done to resolve conflicts for people finishing wonders at the same time, or claiming great people. Humankind could adopt either, or could follow a different route. For example more influence higher priority. Or with advancing the religion: assign some number, which is a function of domestic and foreign followers. But to make such conditional selection rules work well without making people angry, one needs to see the other player's stats next to their stars and fame score. I personally prefer conditional rules and relevant stat visibility, because you cannot devise a strategy for random. I would like to hear what kind of systems devs and other community members have in mind.

Having a good conflict resolution system is quite important for the multiplayer for the reasons outlined above. More so if you are playing a team game of 4v4 or something, as there you play the map and for your team (meaning that sometimes war civs play for sim and vice versa). While in civ6 the choice of civs is determined at the start of the game, in Humankind the civs are dynamic and determined through the choices of cultures, that individual players make. Assuming I see enemy scouts close by, I would not opt to pick a culture that excels at SIM in Humankind, instead I would opt for a war culture. Hence being able to have a strong offensive/defensive culture would be in my best interest. Since the information about my position in the queue is visible, I might opt to pick according to my priorities or pass; if it makes sense I could chose to give my teammate a higher position in the queue. This makes up for interesting team gameplay and team decision-making, that sometimes can make or break the game. Similar shenanigans occurred in civ6 with slot order, recruiting/denying/passing around of the great people.

Assuming that Humankind is played with simultaneous turns, and sequential war moves only, we have several possible conflicts that can occur on the map itself:

 1) Players A and B have enough influence to claim a territory and are standing on it at the same time. I think both should be allowed to found an outpost. The player who's outpost finishes first claims it. The other player is refunded the influence invested. While the outposts are being established, the ransacking should just subtract turns from the current progress. In a rare case when players have completed the outposts at the same turn, they can both bet extra influence. Whoever wins claims the territory, the loser gets the extra influence that the winner bet. Perhaps allow speeding up the formation of the outposts for extra influence (not refundable gambling). When a similar conflict occurs with settlers, then it is just down to betting. I would not like this conflict resolution to be down to who clicks first, or at least as little as possible.

Speaking about the influence, there was a thread, saying that influence is not all that useful at certain stages of the game. Perhaps forming diplomatic treaties and making diplomatic demands should cost influence. This would accomplish 2 things in the multiplayer setting. 1) Make feeding territories through diplomacy harder (combined with war support mechanics it will be very hard) and 2) one has to think carefully weather they can afford right now to expand/attach/make treaties, given the state of the map.

2) Different PC power/quality of connection to the host - issue: In civ6 these factors matter, as the first moves and last moves (A can move 0.3 seconds before B <=> B can move 0.3 seconds after A) matter a great deal for war moves. On top of that you could force-end turn. This mechanics would resolve the queued up movement instantaneously. All of the above are determined by how good your hardware is. Some players see it as a feature and play around it, as I have last moves you have first moves can be used differently for player’s advantage, while other players hate this mechanic with passion.

 In Humankind these factors don't matter for the sequential attacker/defender war moves, which is good and refreshing, however they matter a great deal for AAs moving on the map. While the first moves are very powerful (If I have first moves I can block you from moving through choke points/ claim high ground) in combination with force end turn (then resume) and provide a great deal of strategic advantage on the map, the last moves don't add anything. Hence the superior PC and host connection gives a player an insane advantage in terms of map control and AA movement. So I would really love to see the devs force synchronize the turns as best as possible. Perhaps the fact that AAs move at a fixed speed on the map is a good thing (although could be a tiny bit faster, especially for trains), to make the first moves less powerful. If A is 3 hexes away from the choke and B is 2 hexes away, even if the synchronization is not perfect (it can never be) and A has 0.07 sec on B, B can still reach the choke faster because B moved better a turn ago and because the moves are not instantaneous. 

Combat system:
My criticism of the Endless Legend combat system was that it takes too long especially in the late game (while being otherwise very enjoyable). Same might be true in Humankind. If you imagine a battle with siege weapons, levies, reinforcements from both sides if I am not mistaken in each global turn you will have each player manually play 4 phases with great many units to micromanage. If you have more than one battle like that somewhere else on the map it would be very difficult to make it reasonable time-wise.
Here are several propositions:
Call this option Fast-Dynamic Combat and make it into one of the game settings in the lobby voting phase. It would probably become a standard option for most of the public lobbies. And it should include the following features.
 0) Disable animation on the battlefield. It would shave off a lot of time spent (small time gains would add up to a significant amount).

1) Have an additional UI element on the left side, containing the following features:

 1.0) Battle Declaration as a Phase: If there are no active battles there are PTTs (phase time turner) for different scale battle declarations small (1,2 units in the army)/medium (2-8 incl support and immediate reinforcements (adjacent AA stack, as only the reinforcements deployed within the same global turn matter as far as time constrains are considered)), during the first X1,2,3% of base GTT duration. Past those points the battle declarations can be made and accepted, however they can have only the defender options to retreat/auto-resolve/play manually next turn and attacker options to auto-resolve/play manually next turn. Auto-resolve only takes place if both parties consent to it.

1.1) The UI on the left side contains a small representation of each battlefield and the individual PTT (as well as the reminder for the time cut-offs to start new battles). Let's call a battlefield staging, if you still need to position your units before the beginning of engagement; and active, if sequential WM can be made. Active battlefields always take priority (which decisions do you need to make first) over staging battlefields (attacker battlefields have priority over defender, to optimize the overall time spent by players engaged in the conflict). All the staging battlefields can be grouped together and have overall time turner duration = const1*[number of overall units present on all battlefields during the staging phase]+ const2*[number of staging battlefields]. After all the staging is done, we find an optimized way (probably using some graph abstraction for it) to put together all the battles (as in who does moves, on which battlefield, in which order), so they still fit in the current GTT base duration. If for some reason it is impossible the GTT is extended by the over the head time plus a little bit to account for ping. Each of these war-move phases opens as a persistent strict pop-up of a given battlefield where the battle occurs. If the player does nothing and his PTT runs out, then AI auto-resolves the phase for the player. Upon entering a new phase of the active battlefield, you get a lightning fast recap of the opponent moves.

If the above suggestion works out the turn would probably look like this: conflict resolution decisions -> attacker WM - > defender WM/staging -> staging -> globally optimized sequence of attacker/defender war moves -> other remaining decisions. If you have resolved all of your persistent pop-ups, you can take all the other decisions on the maps, like choose tech/production civics/move AA/deal with random events.

I am not sure how it works but when you go to the battlefield the camera moves to the part of the map where it occurs, zooms in on it, and takes an artistically appealing angle. After the moves are done it zooms out in the same exact spot of the map.

2) Quality of Life: in light of 1.1 and the above, maybe make the camera remember the last non-war related position. So after I am done resolving my war moves and wait for my opponents to do the same I can take all the actions with a nice feeling of continuity.

3) Joint battles: In case the team game or if players A and B have an alliance. A is at war with C. Can B join and fight C as well on the same battlefield. Generalization of it is also interesting. Adds quite a bit of theoretical complexity to 1.1) (which is already not an easy task) as well as needs the devs and community to decide what are the risks for additional parties involved if their ally loses the battle.

 In my opinion it is quite an interesting concept and makes up for many nice stories and player experiences. A was desperately holding the lines in a dire battle vs C, and his forces were depleting fast. Then B's army shows up from the fog of war around the mountain. In a swift maneuver B's forces are pushed back from the flag, forcing his troops to retreat. 

 Thank you for reading. I am excited to see thoughts and ideas, perhaps some polls. And I can provide additional illustrations/diagrams/explanations for the things that I described. 

I play at the Civ 6 CPL server since Civ 6 was released. I have considerable experience playing Civ 6 mp, over 5,000 hours easily. The Humankind sequenced combat system will turn off many mp players. It takes to long, and nobody wants to sit around waiting on other players to finish combat. The quick combat resolution should be a default setting for mp. Also, the civ builds are very unbalanced, and some just bad for the mp experience. Who wants to be next the the player that picks Huns or Mongols? I'm imagining picking Huns and all the sudden my neighbors have "connection problems" and disappear. How many players will drop off once they see Joseon get picked? Humankind has a great concept of building a civilization by stacking cultures, but I see great difficulty in making the game competitive and thus interesting to play mp. If the game isn't competitive, you aren't going to get the player base that makes Humankind a good mp game.

Thats why we should have an option in multiplayer lobby to change how many rounds of battle is in one turn, freedom to choose and now i think that 2 rounds per turn is great idea but we didnt test it so we can only imagine how long will it takes