So, at it's core I believe Humankind to be a really tightly designed game - I think the design space of "you must snowball really well but that won't technically win you the game" kind of Engine Builder really interesting and cool to see in a 4X, as someone who loves board games.  Humankind sometimes can get a bit crazy when it comes to creating snowballing imbalances, but I think the game is getting a lot closer to where most of us are going to be satisfied with the competitive imbalances in the game (though I still want Aesthetes to be really good early, as I like a good HUD and having Influence stuff be powerful is just good player communication).

That said, the game doesn't really communicate any of this to the player.  That's something where a revamped tutorial could certainly help, but broadly speaking I think the AI's behaviors are at least partially to blame for this miscommunication.  Players new to the genre or just simply trying to learn the game in question will tend to fall into two camps - those who have good heuristics and those who do not.  Heuristics like "science is powerful in a 4X" help you "solve" a lot of games pretty quickly if you want to learn them fast, so I like to rely on them.  I also like it when the AI can generally express these heuristics to the player, so that someone unfamiliar with the basic rules of 4X game design can learn them by just carefully observing the AI.

Unfortunately the AI in Humankind just kinda lies to players about how the game operates mechanically, due to some really strange and observable behaviors.

1.) The AI blitzes out of the Neolithic

This is really rough for a *lot* of reasons.  First, there is a beautiful catchup event available in the game that virtually all new players should be encouraged to try to attain.  If the AI is clicking through to Ancient on T3, a nonzero number of players are going to start panicking and feel like they are falling behind.  This pressure is reinforced if they just wanna vibe and have fun and play as a specific culture (thankfully more casual players can change this as an option), but it's just rude behavior out of the AI.  Getting folks to see A Story in Our Stars and get more pops for the Ancient Era is great, because with more pops you can actually do stuff instead of just clicking end turn a lot.  Humankind is largely an up-tempo game with a lot of fun stuff going on, so helping players set themselves up well to do that by not scaring them into leaving the Neolithic too early is just friendly game design.

2.) The AI does not seem to really value "good" cultures more than others, and misunderstands the Fame system entirely

I know this is probably going to always be a point of contention in the Humankind community, but I *like* the fact that some picks are just objectively better than others.  It means that you can, as the player, sacrifice your ability to score fame via era stars for resources *now*, like paying life for cards in Magic: the Gathering.  I'm gonna do that stuff all day.

Once you start looking around at the game you realize that there's tons of ways for you to get fame, especially competitive things.  Any time there is a competitive resource in a 4X, you have to realize that it is way better for you to secure it vs non-competitive resource development, as winning it deprives your enemy of it as well.  I think it's good and healthy game design that you can get tons of fame in tech and wonders and manufactories and stuff, so that the player can make this sacrifice of fame for snowballing.  The player needs ways to feel like making progress and doing stuff is mechanically good, and it is in Humankind, but the AI largely does not agree.

Instead of taking cultures that will help their economy grow and making quick progress through the ages and making you, the player, feel like you need to take the good thing now or the AI might take it instead... after the Ancient, you have literally zero competition for whatever culture you want.  All game.  Even on max difficulty.  Just have a good Neolithic and come into Ancient with a bunch of pops and it's basically impossible to not have a permanent tempo advantage on the AI when it comes to era progress, but that's not because you are just burying the AI in widgets - the AI is farming fame like a lunatic.

This is bad behavior for new folks, as just with the early Ancient flip, it makes people "feel" like they need to start figuring out ways to turbomax their fame immediately.  This is like telling someone playing Wingspan that eggs are worth points and watching them try to literally just play meadow cards and lay eggs all game - they have undermined their own ability to score points in exchange for having less fun.  Because of the snowballing imbalance in Humankind and the presence of competitive fame, turbomaxing era stars is actively harmful to your win% on higher difficulty levels *and* comes with the downside of tricking newer people into playing "weak," which tends to be way less fun than encouraging the player to play "strong."

The AI doesn't need to have a perfect list of cultures to mulch through, but it should be respectful of the strong ones and make players who are paying attention *feel* the pressure to make progress and secure the cultures they want.  This will, in turn, help the players learn faster and have more fun (most of the time; thankfully, the "no competitive" cultures tag is there for folks who actively want to turbomax fame if that's their vibe).

3.) The AI needs to make more cities, please, I'm begging you

There is one way that the AI can do stuff that will use the panic and stress of a new player productively, rather than the last two negative examples.  The AI likes to make very weird and crazy decisions in regards to their city growth, by which I don't mean it doesn't hyperoptimize districts like a nerd (NO MATH ONLY STICK).  The AI will just like vibe on 3 cities all game if you actively expand and do warfare and stuff.  This is not good, as cities are great in Humankind and the AI should show people this.  More cities with fewer tiles attached will communicate way better to people, especially if the AI will start conquering (and seeding!) Free Cities.  The more the AI doesn't have to cheat to get resources the better the human player can learn off of the AI's behaviors, and if the AI starts getting 4-5+ cities relatively quickly, even if it mismanages them, it will help people realize that they should be doing their best to do the same.

~~~

These changes aren't suggested to make the game inaccessibly difficult; honestly, I think if you guys can make these changes happen it should help people learn the game faster and more easily, which should in turn help them have more fun and win more often.