My players said I was railroading them, so I scrapped the campaign and asked them what they wanted

>The only way to make a game good is to have a plot, and to push players to accomplish certain things within the plot.
Strictly speaking, no, but you need some really experienced and motivated players. It is balls-bustingly impossible to find an entire team of self-motivating people.
>They way you "don't railroad" is simply by letting them choose how they want to go about accomplishing things, and then you make the choice that they DIDN'T do have some sort of visible consequence on the world.
Hold up. What you want to do is make the choice that they DID make have some sort of visible consequence on the world. Otherwise you have a bunch of confused people going around wondering why everything's actively getting worse not only despite, but BECAUSE of them doing completing goals that give no reason for them to believe they're making htings worse.
This kills the game and makes people think you're a shit DM.

I have one question for the OP:

How do the players in this campaign earn experience points? In other words, what specific actions are worth XP? Killing monsters? Performing (subjectively) heroic deeds? Good role-playing?

There is a reason that sandbox games usually default to GP = XP. It incentivizes exploration. You could also (and I have done this) award XP even more directly for map-hexes and dungeon-rooms explored, with bonus awards for discovering new dungeon-floors; but this winds up homogenizing the experience, making gameplay less focused and more meandering.

Awarding XP mainly for treasures found provides that uneven, "Skinner-box" type reward system: it makes the players salivate for whatever may be in the next dungeon-room, more willing to risk potential danger at the prospect of potential reward.

Long story short, OP, the essential missing component you may or may not have considered is the reward system.

>le subjectivity meme

I'm actually not mixing anything up. It's really simple, actually: even the most eloquently told, beautifully crafted railroad can fall flat if the hook doesn't grab a player. They'll be strung along for a ride they're just not into, regardless of how much work goes into it. Inversely, all players in all possible forms will be able to enjoy a properly run hexcrawl, because they'll always be doing something they want to do.

This makes the hexcrawl the maximally enjoyable way to run an RPG, from a purely objective point of view.

>I don't know what I did wrong
>I don't fucking understand what I did wrong.
I got your problem, you are playing with dumbfucks that you simply cannot please.

Solution: Play with people that are not dumbfucks.

Sandboxes work if you have at least one player who can set their own goals for their character and work under their own agency to push forward towards those goals. Other players can fall in line with that player or set their own goals, but it doesn't really matter at that point as long as at least one of them is driving the rest.

False. Different groups like different things. Some players enjoy being given a clear direction and goal and have no problems with railroading. Or maybe they feel intimidated by all the options a hexcrawl might present. Or maybe they just don't really know what they want, and an open sandbox will lead to them bumming around and not really doing anything, leading to an unfufilling experience.

The only objectively best style of game is the one you and your group have the most enjoyment with.

Hex crawls are generally outdated garbage beloved by a handful of players.

Look who's never played a good game of old-school D&D.

I just let them think they're not being railroaded give them multiple paths and let them think that the choice matters even though whatever path they pick is the one I wanted them to

The solution to railroading is self-motivated player characters. If your players have clear ideas about what kind of story they want to play out, then a railroad is unnecessary. If your players are prepared to jump on your plot hooks and chase them further than you ever imagined, then a railroad is unnecessary. If your players are willing to get fully invested in their character's goals and engage with the high and lows of their own personal drama, then a railroad is unnecessary.

Obviously, this requires players who have a great deal of creativity, improvisational skill and acting ability. Such players are rare, so a DM usually has to provide the plot and motivation themselves, as a framework for the players to build on.

>Thinking old means good
Oh user

posting to help out future GM's.

>good games of old-school D&D
>bad games of old-school D&D
>good games of new-school D&D
>bad games of new-school D&D

Well aren't you just the most adorable little dumbass?

Your mistake was stopping your railroading instead of making it more subtle.

Hexcrawls are literally a countryside-as-dungeon with all the same problems, because while is was a milestone in storytelling it is just a milestone and not the pinnacle. Every system outside D&D in the past decade have pretty much given the same GM advice dressed up slightly differently, with the clearest language being written into Dungeon World which is very approachable even if you hate the system. And just writing this probably triggers the shit out of you despite being absolutely true.

This. Sometimes railroading is just how you GM. Trying to GM in a way that you're uncomfortable with just leads to you being shit at it. Just dress up your railroading by having minor changes and applying quantum ogres.

>he fell for the sandbox meme

Players don't do well in sandboxes. You have to railroad to a point, but give them options as to which track they want to head down.

They don't have to know they all lead to virtually the same place, as long as you improvise it well enough.

I know what you're dealing with OP, I've played groups who like campaigns open ended in a Skyrim esque type thing.

So basically, you have to give them a goal: I usually default to a villain doing something evil that they have to stop (something so evil that it threatens the entire world so the PCs can't just ignore it). Then I let the PCs adventure on the open-ended map to get weapons, level up, and side quest as they work towards an overall goal.

This way, the PCs can side quest and bumfuck around while working towards an overall goal.

Another method I do are a couple quests in different locations that the PCs can choose to do in whatever order. For example: "This is Weapontown, here will be a quest for magical weapons. Over here is Magictown, here will be a quest for spells and spellbooks"

upvote

A sandbox CAN be done well. But it's good to have overarching plots within that sandbox.

In my experience, it can be mostly improvisation
>the players fight some cultists in that dungeon (unremarkable combat with a few 1st level fighters and an evil cleric)
>Decide one of the treasures the PC's take out is a relic the cult was seeking
>for some reason, these assholes start tracking the party
>players have to put it all together and decide what to do about it

Or maybe it's totally different, and the party made friends with the cultists, and are taken back to their temple as potential converts.

Suddenly you have a "plot" that wouldn't have existed if the players hadn't gone to that certain place and interacted with it in a certain way.

It should be a living world, not just a series of static quest nodes.

A good sandbox also usually starts small, so that stories can develop. Striking a spark anywhere can change the setting.

You overcorrected. You went from one extreme to the other. An on-rails plot is boring, but no plot is also boring. You can have a narrative, you just need to be flexible and willing to change it in response to player agency.

Players are idiots and have no idea what the fuck they want, full story at 11.

If your players ever accuse you of railroading, tell them to fuck themselves, because most groups literally require railroading or this exact scenario occurs, and almost every DM does it, they're just better at hiding it.

Don't get me wrong, I love my group, but every attempt at sandbox I've made with them has crashed and burned because they end up feeling like they have no direction despite me having to put in three times as much work to accommodate for the fact that I can't predict where they're going to end up as well.

What this doesn't mean is that railroading is a GOOD thing. It is merely a NECESSARY thing, and while it is definitely something that can and should be used if your group requires it, it's up to you where you draw the line, and you shouldn't railroad more than you need to. It's a toss up between maintaining a coherent story and allowing your players to feel like they have freedom, and that's where your balancing act lies.

There can be a happy medium. Some sandboxier campaigns have an overarching plot or general motivation. There might be an ancient evil, but there can be a bunch of his minions that you fight at your own leisure, following whichever plot hooks interest you the most.

You weren't in the wrong though OP. Those guys clearly didn't know what they were going on about.

Sandbox just means you don't pitch a fit when the players take things in an unexpected direction.

And one of the great strengths of D&D is that you can key a map with a few monsters out of a book (plus treasure), and have a night's play ready to go. You don't need to graph every NPC or location in painstaking detail--just have a few tools at hand for generating stuff on the fly.

You can also tell players you need to plan out the next session, and require that they give you a rough plan of their intentions for the next session. I think a lot of people forget that it's not just the GM's job to cater to the players every whim--if they want open ended exploration, it's partly their responsibility to make that possible.

Even just having a list of a couple dozen unused names is fantastically helpful. If the players decide to invite random blacksmith #204 to dinner, I can pull his name out of a hat, along with names for his wife and three kids.

>Listening to players
>Literally ever

If you set up a sandboxy setting where things moved on their own outside player input (like the autoplay-sandbox in but had the campaign involve the players being part of some police/military/etc. group where they'd have orders to carry out (not always being ordered and having free time to pursue individual interests and plot hooks, but occasionally being told "you have to go to [x] with the intention of doing [x]" by the powers in charge) and consequences for failing or not accepting the orders, would that be too railroady, assuming the party was all on board of being part of that group and being ordered around in such a way I've never GM'd before and have been thinking about the potential campaign I might do for my group. I value player agency but have been told by GMs I've played with that I have a habit of driving the plot forward more than average, which makes me worried that I'd end up railroading too much if I have a group that doesn't play the way I do.

>There is no such thing as "non-railroad". Sandboxes never work, they're a garbage concept, and no one has EVER done a good "sandbox" game.

Total horseshit.

>Players don't do well in sandboxes

On the contrary, player can do well, but doing well in a sandbox is a skill. Some players git gud at it, others just nope out. OP's players haven't gotten into the sandbox groove yet. Or maybe he's running it poorly, which also sounds plausible -- it's tough to run it right, you need to have interesting stuff happening around the map for the players to see and get involved in.

>I don't know what I did wrong
You gave people what they SAID they wanted rather than what they wanted. They wanted some wiggle room and side quests around a linear core story like The Witcher 3 has, not a full on Roguelike,

First post best post

Eh. Prolly is actually your fault. There's a fuckload of gray area between railroading and sandbox.

Railroading is when your player says "I want to do this" and you say "that doesn't make any sense--your character wouldn't do that." Or you put obstacles in the way specifically to handicap them into solving problems in the particular manner that you pre-planned that the problem be solved (the "Oh we have to get to the next level of the dungeon? K I get a pickaxe" solution should generally be viable, if stupid).

Sandbox only works if the PCs are willing to make their own plots.

If they want to be kings of a region, or invent the tank and build super war machines to crush the world or be the best damn spice traders in the world you can have fun. If they're a bunch of retards who want to kill stuff and collect loot they should fuck off to Runescape.

Completely correct.

>If they're a bunch of retards who want to kill stuff and collect loot they should fuck off to Runescape.
I really hate this attitude and it keeps people out of a good hobby. Explore kill loot repeat is perfectly fun in P&P too if done well. Stop acting like your preferred fun is more legitimate than theirs.

If nothing else, most newcomers to tabletop games nowadays come from video game RPGs and probably need to be introduced to the concepts in a familiar way before they can start to take full advantage of what P&P has to offer. And it's not like D&D didn't have that kinda gameplay long before Runescape, either.

>Explore kill loot repeat is perfectly fun in P&P too if done well
Not for the DM. You can automate the tables, so play without one.

Yeah I think you're just responding to an internet-expert. No one who actually plays RPGs categorically rules out explore, kill, loot adventures. Wraith the Oblivion and Mouseguard both have them, for shit's sake. I can't think of an RPG that doesn't.

>I don't know what I did wrong
>I don't fucking understand what I did wrong.

You picked your players wrong, they're a fucking bunch of idiots. Scrap them and reuse your campaigns.

Except the DM is functionally useless if all the players do is
>find dungeon
>kill goblin
You can roll the damn tables and auto-gen a dungeon to delve.

Have either of you actually DMed before?

Psych--no one needs to read your lies to the contrary. The obvious answer is "no."

Designing dungeons is fun as fuck. Putting challenges into them and watching how players cope with them is fun as fuck. Giving them loot so that they get back to town and then wander around wondering "what kind shit can I stir up with this?" is fun as fuck.

In fact, doing so is so much fun, for DMs, that DMs got together and expanded the games into what we now call RPGs, almost-entirely because the "got back with loot, now what?" part was radical.

It's a style. It takes practice.
For you AND the players.

You're not used to this, so that's fine.
The main thing to ensure is that the player's actions have consequences. If you've got your sandbox running right, you can basically just take a step back and let the players chase the tail of their own consequences.
But that's by the by, you've read the advice and stuff. It's practice.

Your players have been trained to follow a plot, though. They've learned to expect a certain thing. If they're not used to a sandbox, they won't have that internal sense of "we have to find the fun".
In a sandbox the players can putter around doing boring shit if they want. That's not your job. Your job is to keep the world moving and the factions fighting and the large-scale events happening.
If they want to get something done, that's down to THEM.

You're the world. They're the people in the world.
If they want direction they have to direct themselves.
They have true agency.

That's the sandbox ideal.

That's not a sandbox, a sandbox is an open expanse with toys littered throughout and no established direction beyond a vague end goal. A true sandbox doesn't even have a goal. A sandbox is not "when you don't railroad."

easy solution: make a railroad but offer them 2 ways to go about most things. Gives them a sense of control without spreading yourself too thin.

"Don't railroad us" just means "make us think we have a choice in what happens." If they actually do, that's great, but they don't have to.

>Designing dungeons is fun as fuck. Putting challenges into them and watching how players cope with them is fun as fuck.
For you, sure.

>Giving them loot so that they get back to town and then wander around wondering "what kind shit can I stir up with this?" is fun as fuck.
For you, maybe. That's not why I DM, and certainly not my goal when sitting at the table.

>In fact, doing so is so much fun, for DMs, that DMs got together and expanded the games into what we now call RPGs
...what

Nigga, are you really this retarded that you're saying "This is so much fun I love it can't you see how this is the best?" and pretending that's an argument?

He's not the one that started the "my way of DMing is correct, yours is stupid" train. See the folks he's responding to, with their blanket denials that any DM could ever enjoy that sort of thing.

user, you know why listening to focus groups lead to terrible products? People don't know what they want.

Also, there is a middle ground between no choice and no direction. Look at something like Ocarina or Time. There's an overriding story, but you can do different parts of it in different orders at a pace of your choosing.

As basically everyone already told you your mistake was to listen to your players. Sandboxes are shit and basically impossible in the majority of groups. A good DM is one that can railroad while giving the illusion of player agency. A good way to do that is to plan a session while keeping details variable. This way while the players can choose if they want to go to the swamp, to the mountain or to the desert either way they'll find a tomb with the cultists that you want them to find, just dressed differently.

I know this is bait, but palette shifting is still shitty behaviour.
You're still imposing your will on the players.

And just because you're bad at running a sandbox doesn't mean that everyone is.

A million small ideas won't be as intriguing as a single solid idea. It sounds more like your original campaign lacked flexibility. You can run a railroady story without anyone feeling like they're being railroaded by being ready to rewrite your adventure at any point. If you need to cut the session short after two hours because you need to rewrite the story around an extremely insignificant detail that players spent most of the session investigating, do it. If you play really long sessions, break out a board game for an hour or two while you fix shit.

Obviously put your foot down when you need to, let them feel like they hit a dead end, but give them a little something at the end of that dead end to keep them going.

If your campaign can't be written around a band of chucklefucks inventing a pressure bomb to stop a plot-relavent pickpocket, THEN your adventure is too railroady.

You have to tie your players to the train tracks, but blindfold them so they don't know it.

Make every choice they make lead to what you want done.

It's the only way to win at DMing.

The only way to make a successful sandbox is to combine aspects of a plot-driven game and a sandbox game.
>There's this bbeg who's going to attack the land and you should stop him
>Here's this world, decide how to do it

>The only way

Not true. Also

>The BBEG meme

Video game trash.

I always end up with a balancing act, the players start on a railroad until they get tired of where it's going so they blowup the track, then they wander around in the sandbox for a while until they want to get back on the train

>I know this is bait, but palette shifting is still shitty behaviour.

What? No. "Palette Shifting" is literally one of the best strategies a GM can adopt.

The players can *go* wherever they want, but beat for beat an encounter that's been mindfully crafted will always be more engaging for the players than rolling a random encounter table.

>Not true

Yes, true. If you're not going to put pressure directly on the players, you NEED to give the world itself pressure. An impending war, or apocalypse, or raid, or something. Otherwise, the PCs will faff about endlessly, forever. The world needs to be running forward so the PCs can bump into it.

>I don't know what I did wrong

They don't want to FEEL railroaded.
Watch the second Matrix movie. Particularly the bit where they bang on about 'the illusion of choice'. That is the important thing to take away from this experience.

You were telling them where to go and what to do, instead of letting them think they wanted to go there. Or putting things where they wanted to go.
There is a middle ground between linear JRPG with hours-long cutscenes and traipsing through a cave for hours, and open-world western RPGs where you're so busy building houses and brewing potions and grinding skills that you forget the world is in danger.

You need to occupy that middle ground.
They want to feel like they're driving the plot. Highway them instead of railroad. Tourist traps, other traffic, pee breaks, detours, shortcuts, etc. One direction, one goal, but a few choices of how to get there.

I never said "your way is stupid"

I said, objectively, sandboxes are shit. Can they be fun? Sure, anything can be fun. But they're still shit, especially when you make your rule "I will never, ever ever ever railroad in any way because anything resembling a plot is evil"

>an encounter that's been mindfully crafted will always be more engaging for the players than rolling a random encounter table.

Maybe for you, but I have a talent for improv, and a tendency to overthink shit so overplanning is my single worst enemy. My and my players favorite scenarios have all been things I did off-the-cuff, because when I do that I generate exciting battles, where when I prep an "encounter" it winds up lacking the spontaneity and starts feeling canned.

Players skilled at sandboxes don't need much of that from you, that's for playing with folks who've been trained by story-heavy DMs to need hand-holding.

>anything resembling a plot is evil

Said nobody ever. Plots in a sandbox should emerge from play, they shouldn't be squashed by the DM because he's some kind of strawman idiot.
Sandboxes, run well, are some of the best gaming experiences I've ever had.

I think you needed a very flexible antagonist with a tangible, malicious goal.
If you do this, OP, always think like the antagonist when planning your sessions. You'll want the players dead, but you want them to challenge you at the same time.

>that's for playing with folks who've been trained by story-heavy DMs to need hand-holding.
>Anything but a sandbox at complete stasis with no big events currently happening in the world is hand holding

Kek

>Read up even more on how OSR players do them and try again.
I don't think there's a more obvious way to say "I don't like what you said, but I can't actually think of anything to say to rebut it."

Different person here and I disagree with you, improv might be fun for you guys and gets some laughs, but I'd say for most avid DnD players, it'll be pretty obvious if the encounter wasnt crafted. It always lacks depth.

A good well-crafted encounter will always be better than a good improvised encounter.
To the players, its all new content either way.

>implying things nobody said

Point is, players don't naturally just stand there and wait to be told what they should do. Play with some kids sometime -- my last such experience the kids practically ran me over in their enthusiasm to get out into the imaginary space and DO SHIT. There wasn't any semblance of aimlessness or "what now?" to be seen. That's a trait that I've only ever seen in older players who've been trained by video games and story-focused DMs that they need to look to the man behind the screen to point them in the "right" direction to find the game.

>Video game trash.
Alright, what motivates the main plot for your games?

Have you ever watched Buffy the Vampire Slayer, user? Because I think that's the ideal format for a game. The most important takeaway point is that you're not constantly going after a single objective. Each episode has its own objective, and most of them are independent. But every couple of episodes in the first half of the season, the same asshole is behind what just happened. Either they just foiled one of his plots, or his plot crossed over theirs, or he just sent someone to kill them because they keep fucking things up for him. Then things come to a head, and the series starts building up to a final confrontation with him.

That's how you do a good sandbox campaign. Throw a couple of different plot hooks at your players, and whichever one they reacted to the best, that's their new nemesis. They start finding evidence they're behind every third or so plot they come across, and they'll naturally start pursuing that nemesis. Now you start coming up with plot hooks based on how they go after the nemesis instead of just what they naturally look into. Then give them a satisfying conclusion, and viola - a perfect sandbox campaign.

Well why not sandbox them until they pick a plot then go from there?

>ITT: Veeky Forums doesn't "get" RPGs.

First post is like always, best post.
OP, whenever you think of an average player, imagine yotsuba/any other

>Players skilled at sandboxes don't need much of that from you, that's for playing with folks who've been trained by story-heavy DMs to need hand-holding.

>Players don't need stress or drama pushing the game forward because buzzwords

Shove a FATE rulebook up your ass until you die of colon perforation you dumb fucking hipster faggot.

Competently skilled players can readily create their own stress and drama without the GM pushing it on them arbitrarily, you ignoramus.

>Alright, what motivates the main plot for your games?
The personal goals of my player characters.

You can't just say something like that without explaining which of these two disparate sides your weighing in on.

Players are exactly like women, user.
They're just ________Stupid_______

So do you just have all their goals in the same place or do you just put the spot light onto one player at a time for the sessions it takes to resolve their shit then have that player make a new character because they have no real reason to stay?

Usually the players have a goal that they're united in because I have a session 0 where my players make their characters together like a good gm

And what if a player has a different goal?

From my perspective as a GM a "sandbox" is to a degree merely an illusion due to limited resources as a GM. I will describe the situation then describe how it was responded to on Veeky Forums.

The players leave the town and there is a fork in the road. One direction leads through a swamp to a logging community. The other direction leads to the coast to a porn town. We shall refer to the first as Path A and the second as Path B. I as a GM have only prepared one encounter with a group of bandits. If the players travel down either Path A or B they will encounter these bandits. The players know that there are bandits in the area.

This is an example of equality of process and inequality of outcome. I as a GM with limited resources will not have all the infinite paths branching out with unique material, because to a player perspective there is at no point where material loops back in on itself and repeats (they only see the material I prepare once).

I keep lists of encounters, NPCs, shops, etc. for this purpose.

The response I received last time was that this robbed the players of meaningful choice and was railroading. I personally disagree, and even if I do agree it does not matter as the players, from their perspective, never see the strings.

I do not know how I can be expected to craft an essentially infinite number of branching pathways.

Well you see, because the player characters all made a unifying goal to keep the party together, they have enough camaraderie to spend a few sessions helping their friends accomplish any sort of personal goals they have on top of that.

As an added bonus, because I'm not shoving a man plot with some BBEG down their throat every other session, they get to pursue those at their own pace.

And if any of the player's goals is to kill someone in particular are they just always weak fuckers with no skills?

>The response I received last time was that this robbed the players of meaningful choice and was railroading.
This is true.

>It does not matter as the players, from their perspective, never see the strings.
It's a crutch. It's not hurting you in the moment, but your ability to improv in other, more critical situations suffers for it.

>I do not know how I can be expected to craft an essentially infinite number of branching pathways.
Random encounter table by terrain

What you are calling "competently skilled players" are actually "GMs trying to make a game when their GM refuses to make one."

If players were capable of what you're claiming, they'd be GMs.

>Random encounter table by terrain
Not that user, but how do random encounter tables not rob the players of meaningful choice?
I mean, the players' action have no bearing whatsoever on the results of the table.

Why can't sandbox and railroad be intermixed?

Start your PCs with a plot hook, try to get them invested in the world and the story. But if something starts dragging at their attention, take the pressure off the main plot a little and give them enough room to pursue this other interest. If they want to fall off the main plot entirely, find a way to do so - this is a story and a game at the end of the day, you don't need to actually have the world end if they don't stop the lich and decide to make cheese instead.

>If players were capable of what you're claiming, they'd be GMs.

All players are capable of it. All players have the makings of a GM -- maybe not a great one, but most people can learn to be a decent GM with enough practice.
Quit looking down your nose -- if you expect more, you'll get more.

>This is true.
Is your destination not meaningful choice? The port and the logging camp are different outcomes with different stories. Merely you received the same encounter along the way. Player choice impacted the story and the outcome.

>It's a crutch. It's not hurting you in the moment, but your ability to improv in other, more critical situations suffers for it.
Improv is to be supported by actual work. I am not talking about characters and stories (which should be a mix). I mean names and stats. Improved stats often lead GMs to either fudge or create unbalanced encounters, or worse only pretend to have stats.

>Random encounter table by terrain
This eliminated meaningful choice and is still equality of journey, merely more varied.

That's how most good DMs do things, actually. You can make a sandbox good. You can make a story-focused game good. You can even make a total railroad good -- people buy tickets for rollercoasters, after all.
But most folks will use a bit of this or a bit of that. And that's great too.

Railroad and Sanbox both seem to be terms that Veeky Forums has essentially made meaningless.

Railroad does not mean linear, it means that player choice is robbed of them in a obtrusive fashion. Having a story is not a railroad, nor is consequences. Telling players "you can't do X because reasons" is. Letting players do things and suffering the consequences constitutes meaningful choice.

Sandbox is merely when you allow players to create their own story and goals and find it for themselves. Often players can not play in a sandbox because they lack meaningful motivation as people to actually direct their own story.

>This eliminated meaningful choice and is still equality of journey, merely more varied.
I will clarify further. I am a big advocate for not using random encounters. As I said I have lists.

Drawing from lists for what is appropriate for the moment creates a far more meaningful and coherent narrative than random encounters.

I brought this up in the last time this was argued as well, and received the response that I as the GM should not decide what is a meaningful narrative and leave it up to random rolls. I disagree with this.

Whaddyaknow. This train of "you gm WRONG!" shitposting is starting to produce some actual good posts.

>Wanted a more sandbox game where I can travel to all the places of the map without some heavily-guided order
>completely self-motivated and willing to do and pursue shit that interest the character
>Will actively work with the DM to help him get ideas for what kind of plots I'd be balls to the wall into
>Gave plenty of backstory for the DM to use
>He pushed us into his narrative, now we are one with the rails
>I don't like it
>I didn't want this narrative at all

You did alright OP, you just have some shitty players. As long as you made over-arching plots as well as minor quests, you couldn't have done a single thing else.

>doctor is my baby a player or a dm?

>Improv is to be supported by actual work. I am not talking about characters and stories (which should be a mix). I mean names and stats. Improved stats often lead GMs to either fudge or create unbalanced encounters, or worse only pretend to have stats.
And this right here is why I love Apocalypse World and a good deal of the translated JRPGs we have.

Apocalypse World gives you a suite of tools and rules that allow you to improvise a proper sandbox.
Ryuutama, Tenra Bansho Zero, etc. don't even bother to pretend that they could be used for a sandbox. They are clear and up-front about the fact that there are rails to follow and how you can still make that train ride a meaningful experience without trying to derail it.