Last year I've watched through Exaltwitch, an Exalted actual play (you can read my review of it here). One of the big plot points of that campaign was the quest to cleanse the party's ancient ally, Three Fates Shadow, from being an Abyssal under the influence of a Deathlord (an ancient spectre that rules the underworld and is generally evil). The culmination of that arc, after it's been built up for the majority of the actual play was the party going to the highest authority in heaven, The Unconquered Sun, and demanding his help. He agreed to do it under one condition - he asked each member of the party "what are you willing to sacrifice?" and then one of them would have to make that sacrifice.
That question, at that moment, rang very hollow. In Exalted, the player characters are capital H Heroes of legends in the vein of Gilgamesh or Sun Wukong. They are the once-kings of the world who can move mountains with their bare hands and who have saved the world from tyrannical titans of old. The question being about some thing they'd have to give up in the future, rather than asking them for what great feats have they already done in pursuit of the goal felt very transactional, rather than heroic.
I understand that it was time to wrap that plot up as the game was beginning to draw to a close. Leaving that plot unresolved because the players didn't put in that much effort to fix things would be anticlimactic. Ideally, it would be communicated between the GM and the players way ahead of time. As it was now, it basically felt like throwing the character at the glory most high and tossing some coins in his face demanding "it's broken, fix it" because the players wanted it really bad.
The GM did state later that if the PCs had not offered enough their ally wouldn't have been redeemed. However, the PCs also did gamble a bit about who would have to make the sacrifice, and to make matters worse it turned out they had to sacrifice nothing since the person that drew the shortest straw was the comatosed ally, who on the count of being in that state did not volunteer to sacrifice anything. So the PCs earned a free Deus Ex Machina resolution to their arc for free.
Now imagine if the resolution was instead tied to the themes of being a tragic hero, someone larger than life, or doing impossible feats. What if Three Fates Saved would have to live the life of virtue, despite the Deathlord influences, and sacrifice her life selflessly in order to achieve redemption in death? What if she had to confront the Deathlord that held her in a binding and make him achieve peace and pass on (which would be a monumental accomplishment), or accomplish a great feat of peace, like reconciling the differences and bringing peace between the Realm and the Solars (while the party that were the Solar by now essentially wanted to conquer the Realm and subjugate them essentially). Any mighty deed that would play on the themes of "redemption", while avoiding things tied to the aspects of death and destruction.
Instead, we had Rey offering to sacrifice ever seeing her home town of Nexus (central location of the series), her father (her central relationship), and any glory she would gain in the future. Jorek offered his chance of revenge at those who have wronged him. Valeria offered giving up what makes her the most happy - her relationship with Speaks-of-Silence. Sure, any of them would be noble sacrifices and an interesting plot to explore, except the series was already drawing to a close with 13 episodes left to go, so you'd barely see any of them play out.
Worst of all, I don't think this was the GM's intended resolution for this arc. A few episodes earlier the characters were also trying to figure out how to redeem Three Fates Shadow, which eventually drove her to the coma. At that time I noticed the GM tried telegraph some solution to her problem, spelling out the clues to her condition, but the player did not pick up on it and the moment was lost.
In situations like this, especially when you're making Actual Plays, it feels important to communicate with everyone about what they want to see accomplished in the game, about how they want the game to progress, etc. This also goes both ways - sometimes the players want to communicate with the GM about what they want to see in the game, and sometimes the GM wants to talk about how some things should play out. Some might balk at that since it's a bit close to railroading, but if your intent is to improve the game, make it more fun for everyone involved and for the people watching, it might be for the best. We've done that on a number of occasions. Heck, one time we played through a module, discussed it at length, workshopped a better version of it and played through it while still having fun.
It's a fine line to thread, but it probably works for the best when everyone's involved in the process and there is a back-and-forth on how things should go down. It might not be for everyone or every game, but it can make some key moments land better.
Conclusions
When you have a long story arc in a game you hope the payoff would be worth the long buildup. The longer and more epic the arc the more satisfying payoff you need for it to feel rewarding. However, if the resolution hinges on what the players do, it might be worth having a chat about it ahead of time so everyone would be onboard with what would produce the most satisfying ending to that story. This is especially important when producing media for other people to consume.
If you don't plan ahead, together between the GM and the players, you might just have to ask your characters "what are you willing to pay in the future for a resolution to this plot in the present?"...
My group and I play a lot of games with interesting mechanics. Lately, we've been trying Cortex, a modular RPG system where you can tailor the engine to your game needs. One part of the system you can plug into your game are Trait Statements - some statement that focuses and refines a trait for the character that's meant to be challenged in the course of the game. So for example, you can have a Perception trait at D10 with a Statement "Trust No One" attached to it. This would tell you about the character's worldview. Mechanically more importantly, you are supposed to challenge these Statements to get a bonus to a roll and to change your character. So if say, you decide that you can trust someone, you would roll 3D10 instead of 1D10 for that roll, but then you would have to either change the Statement, or change the die associated with the trait, either turning into "Perception D8 Trust No One (and a bonus to something else)", or "Perception D10 I Can Count On Others".
While this mechanic in itself is all well and good, from playing various games over the years, I'm yet to see anyone embrace such character changes / sideways growth as a part of their gaming experience. Let me elaborate.
Homunculus character
More often than not in my experience, when someone makes a character for an RPG they come out as a homunculus, a small version of what the character will be later in the story. When you make a warrior that's all about being honourable and just, they start out as a honourable and just warrior with weak stats, and over the course of the game, they grow into being a honourable and just warrior with strong stats and minor tweaks here and there. If you want to play a crafter, you build a crafter and invest in them being a crafter, etc. Rarely do you see a shift from one to another, or from one fundamental set of beliefs to the next.
Medieval art and homunculus baby Jesus - "perfectly formed and unchaned"
Sure, you could come into a game with a blank slate of a character and form them as they grow. From what I heard this was especially prevalent in oldschool RPGs where most level 1 characters of a given class were about the same, a lot of them wouldn't survive the meat grinder and you wouldn't care about their backstory if they would just die one session later. This kind of attitude is literally related to the term "grognard" in its original meaning.
Similarly, you could build a character and aim for them to have a character arc where they go from a naive child to a grizzled grognard and then to a quiet farmer, but unless you are playing something like Chuubo's where you can literally create an arc for your character, it might be hard to execute.
From my experience, you generally see homunculus characters - a fully formed idea of what the character will be like, with minor wiggle room for the details. If you want to play someone else, you generally don't shift your character from one thing to another using mechanics like the above, you just make a new character.
Similar mechanics
Cortex is not the only game that has mechanics for such character shifts.
In Star Trek Adventures every character has a set of Values, which basically reflect their moral centre. Things like "The Needs of the Many Outweigh the Needs of the Few, or the One", "Holds Everyone to the Highest Standards", "Duty above all else", etc. Those are used to either challenge the characters and make the situation more complicated because of their beliefs, or to let the character challenge that value and change their worldview based on that experience.
This of course is very much keeping with the themes of Star Trek and character development. However, in the game it can feel like you should make characters that don't believe what they should be believing, and your reward for having that character growth is a simple stat readjustment. I've heard a player be frustrated with the game expecting them not to make a character the way they want them to act and constantly questioning what they believe in, and perhaps giving a mechanic to what otherwise might be organic character growth is having the opposite effect (reminds me of Freakonomics...).
In City of Mist your character is built out of themes. Things like "trained boxer", "man of steel", "diviner", "the guy with a van". These themes accrue "fades and cracks" over the course of the game if they are neglected. If you don't show up to your boxing practice, solve problems with guns or generally make that part of your character not important, you will eventually have to replace that themebook for another to reflect what has taken its place in your character's life. While this can be an interesting flow of a story, especially when replacing your themes can turn you fully superhuman or fully mundane with some serious repercussions for either, if the players are too loss-averse or make their characters just right, they might not engage with this mechanic at all.
Many Powered by the Apocalypse games we came across feature an interesting character option for late-game levelling - "make a new character". This is example from The Veil:
In most games this feels a bit strange, but there is perhaps one game where an option like this works - The Sprawl:
The Sprawl is a Cyberpunk game, which comes with its genre expectations of character life being rather cheap and expendable. Since this character level up option costs additionally a good chunk of money, you can see it as "your character gets to retire", rather than being a given for any character. It's something you work extra hard towards.
How we handle these things
I hope our group is not alone in this, but seeing as True Friend needed to be a merit it might not be universal, but we have a relaxed attitude to character building. If you need to tweak your character, just do it, it's fine. If you want to do a complete rewrite of a character for new mechanics, the GM will usually agree (we've done that once in Heaven for Everyone after a new supplement with a new character splats came out). If you want to make a new character because the old one doesn't play that well, pretty much the same applies (we've done that in a yet unpublished Humblewood game).
Couple that with us generally knowing what kind of characters we want to play (and GM being pretty much always on-board with whatever the players come up with), we rarely engage in any of those mechanics. We have character growth and changes as a part of playing our characters in the world (for example in The Living Years demigod Atrus didn't want to form a religion around himself not to impose his worldview onto foreign people, but since they came to him for guidance and after being reassured by one of the NPCs he trusts it's fine, he changed his character's outlook organically).
So perhaps it would be good to make such kind of attitudes something acceptable in more games without necessarily needing to put in mechanics around retiring an old character and making a new one...
Conclusions
A number of games feature mechanics for tweaking your character's stats and worldview. Often, however, these might not be all that useful to the players if they already made the characters exactly the way they want to play them. It's good to give the players options to tweak their characters to better suit their games as they get some hands-on experience with how they play, but making entire mechanics around it might be a bit much...