There is a lot of discussion, as far as I can tell from the inception of TTRPGs, about player freedom and character autonomy. In current times, this is often distilled down to a fairly naive statement of "railroading is bad" - but I think there is more nuance in the discussion than just that. I posit that there are three general categories of player freedom that can be maintained, and that RPGs are a balance of which to maintain in what proportions. Full dedication to all three is possible only in pure make-believe, as there are then no limiting factors and thus there is no more game to be played. With none of the three, what is left is only a story told by one individual or a procedural system running its course. It is not necessary, I think, to maintain all of them or have any sort of balance - to be a game we need enough freedom to allow for decision and many possibilities, but enough limitiation for these to be meaningful. These apply in both a mechanics and a narrative sense.
Freedom of Action (FoA) is probably the most often discussed form of freedom. I think this is what everyone defaults to discussing when these sort of topics come up, and thus I think it is maybe the least meaningfulf or interesting for me to discuss. Simply put, FoA is the ability of the players to choose the actions their characters take in the game. For much of TTRPG history, FoA has been the primary pursuit of games and considered absolutely untouchable. Characters must be able to take whatever actions their players want, within reason. When it comes to "railroading" the topic of FoA is likely the easiest to explain and note the problems in; if you limit the actions the characters can take, then where is the game?
However, I think more recently, there has been experimentation into limiting FoA. This is where some of the difference between the mechanic and the narrative becomes relevant. For instance, PBtA games have very low FoA mechanically, in an attempt to service their narrative. Narratively, characters have nearly infinite FoA - this is handled by the mechanics of the game ignoring most of this action or by lumping it into very specific "moves" which are on the character sheet. When focused on mechanics, characters are limited to only a small handful of game influencing actions.
Freedom of Consequence (FoC) is the space of possible outcomes in the game, and how they can be influenced. At a simple baseline, the consequences of actions are entirely the domain of the GM but are expected to be guided by the game mechanics. In other words, the traditional model of game has strict mechanical consequence, but very free narrative consequence (within the often limited scope of setting and genre). When discussing "railroading", FoC is probably the most important to discuss when it comes to novice GMs, and I think where much of the real debate on the topic is currently focused. In many cases, a GM limiting FoC makes any other player freedom completely irrelevant - especially when considered at a narrative level - and is the case that is often called "railroading".
When it comes to a games core design, limited FoC can be a valuable tool and is often implemented. Hit Points are an aspect of limiting FoC; in the D&D style of game, one of the only meaningful consequences in the mechanics is loss of Hit Points - the game is not able to create true consequence in many other ways. The pressures that a game can apply to a character or player are often extremely restricted like this, which creates pressure on the GM to actively put significant effort into creating FoC.
There have also been games exploring FoC in my opinion. Something like FATE gives players absolute FoC (in theory). FATEs goal, as far as I can tell, is to give players FoC in a narrative sense - which is something that is very difficult to do and often considered taboo in many game cultures. FATE allows the non-GM players to drive the narrative and determine the consequences of actions on their own. Mechanically, FATE has a similarly limited space of consequence to most games, but the system ties "damage" into a more narrative space as well to allow for mechanical and narrative consequence to be intertwined when progressing the story forwards. Allowing "damage" to be a consequence of any form of action while also being marked as an arbitrary plot-relevant "condition" means that FATE has very open FoC.
Freedom of Resolution (FoR) is the most complicated of the three freedoms I am presenting. Resolution is the game space that exists in the transition from Action to Consequence. Most games have a very strict and thus limited FoR: when an action/intent is declared there is a non-deterministic method used to determine an outcome. Most often, this is rolling the dice. The outcome of the method of resolution (MoR) is then interpreted by either the GM (large FoC) or by the game engine (small FoC) and consequence is applied. In most games and game cultures, it is not really a consideration that one could have a greater FoR than this base level limited one - the core MoR is the game, there is no lever to pull and no knobs to twist.
I think that the OSR movement is actually focussed around the idea of increasing FoR. OSR generally has a focus on "Diagetic" actions and solutions, characters solve problems using in-world methods. The method of resolution for a "diagetic action" tends to come down to GM fiat - but done well it would be following logical and rational guidelines and principles. The GM is not using their position to determine what is best for the game plot, or what would be the funniest story to tell afterwards; they are there to determine wether the player presented MoR is a viable solution to the problem at hand. I will use an example of dealing with a trap in dungeon: in a generic D20 game (eg 5e) the player dealing with the trap rolls a D20 and either resolves the situation or takes damage based solely on this one outcome, there is no Freedom of Resolution here. In a narrative style game (PBtA) the player could present a course of action (high FoA) but this action is then resolved by either just happening because it is not relevant to the plot, or still through a fairly simple die roll or other core MoR - although there is high FoA and FoC here, the resolution is entirely procedural. In an OSR style game, the players presented with a trap will describe various actions they take, and the GM describes the outcomes of those actions until the trap is either dealt with or it triggers - this may initially seem like high FoA and no MoR, but I argue that there is very limited FoA (action is "disarm the trap") and that the entire process is actually the very open and unlimited method of resolution. Dice are only used when logic dictates that an outcome to an action is uncertain and it would be unfair for a GM to make a decision either way.
I will go through some game systems to discuss how I think they fit into this model.
This is the category of games that most people are familiar with, this is where D&D 5e is, as well as similar games such as Pathfinder and Lancer.
I actually think these types of games have an interesting (to discuss) dichotomy built within them, they simultaneously have no freedom and complete freedom, depending on the part of the game that is occuring. These games mechanics are built almost exclusively around combat, with little to no mechanics for anything that isn't combat. In combat, they actually present almost no freedom. Freedom of Action is limited to some very basic universal actions (move, attack) as well as the special powers listed on a characters sheet. There is some FoA in making "tactical" decisions, but a lot of the time this makes negligable difference or is really a predetermined set of steps. For instance, it is a tactical decision to flank your opponent or not - but it is also just a strictly beneficial action to do so, thus there is no decision to actually make beyond a binary "can I flank my opponent?". As discussed earlier, there is no Freedom of Resolution in these systems, dice are rolled and the game engine makes a determination. Following this, there is also no Freedom of Consequence, as soon as the MoR engages the system is entirely procedural - consequences are limited strictly to those described within the game engine: usually HP damage, sometimes a strictly defined status effect. Combat is a strictly procedural and largely uninteresting enterprise, as the only question to be answered is that of how many resources are expended to achieve the predetermined outcome. This is, in my opinion, compounded even more by the Neo-Trad style of play which focuses on individual character stories and thus does not even allow for the one potential unexpected consequence of a character dying.
Conversly, there are almost no mechanics in these games for anything outside of combat. D&D has some limited "social skills" which can sometimes be used, and Lancer has literally nothing. This puts them in the real of playing make-believe, with not actual game occuring. Limited social skills are at this point just a hindrance, since everything is already so close to make-believe. Players can occasionally make some die rolls, but with how reliant the rest of the system is on GM fiat these rolls really just mean that the GM has to warp their results to fit the story they want to tell. There certainly is fun to be had in pure make-believe and many people enjoy it, but many people (such as myself) enjoy having some aspect of game.
There is also the weird interaction between the entirely closed combat system, and the entirely open everything else system - I intend to write about this at a later date as The Dissonance of the Combat Encounter.
This is a category that has been on the rise in the past decade and a bit, seemingly coming largely out of Forge-theory. These games have the goal of creating a good story and narrative through play, and create mechanics around that. Known systems in this genre are Powered by the Apocalypse (PBtA), Forged in the Dark (FitD), and FATE. I think these all explore slightly different approaches and fit somewhat differently within the framework discussed.
PBtA games function under a model of back and forth narrative between players and GM until a "move" is triggered. Moves are defined moments where the narrative is cut and mechanics are engaged, which occur from specific narrative actions. A move is resolved with the roll of a die which is then used to refer to a very specific set of consequences designed for that move. This system thus has complete FoA, but has a divergence upon resolution. If the action is not covered by a move, it continues as pure make-believe. If an action is covered by a move, that move is triggered, and mechanics are engaged. The move mechanics have have no FoR and no FoC, resolution is procedural and consequences are extremely limited within a single move (although the space for consequences allowed across all moves is limitless). Generally, the consequences of moves are more extreme than what would occur through normal back and forth narrative to create a loop where players are encouraged to trigger moves as often as possible. Consequences are also usually immediate to the scene. Thus the games feel generally free, with controlable moments of restricted mechanics that push open narrative forwards.
FitD games are derivative of PBtA, but have evolved to be their own category. I think that largely they work the same, but there is difference in FoC. In PBtA games, consequences are an immediate thing to progress the narrative, whereas I find in FitD games consequences tend to be long term and heavily gamified. FitD games have two basic consequences for failure, either take damage (stress) or "progress a clock". If things go poorly, you also take one of a select few "traumas". From this perspective, consequence is more limited and restricted, in my opinion, in FitD than in PBtA. There is also the aspect of the "downtime" portion of the game, which is a heavily restricted set of actions to cover everything that isn't the core activity of the game. For instance, one can engage in a "vice" to reduce stress levels. This sort of thing has almost no FoA (there is a limit on actions, so basically a resource priority decision), there is no FoR or FoC - this action is always the same for everyone, their actual chosen vice is irrelevant.
FATE I mostly discussed earlier in the Freedom of Consequence section. The interesting thing about FATE is that it tries to maximize FoC for the players, they are the ones who largely determine consequences of actions. As a narrative game, it has full FoA like the games above. It has a single MoR, and thus limited FoR. Consequence has a dichotomy between mechanics and narrative. In order to support the high level of narrative FoC for the players, the game actually largely restricts the mechanical effects of any actions. Almost any action will: "deal damage" (mechanical, not narrative), buff a future roll, or hinder opponents. Any narrative consequence can be boiled down into one of those three mechanical results.
I don't have much to add over the discussion in the FoR section, but the key point is that OSR games seem to be trying to open up Freedom of Resolution. These games remove powers, moves, and the like from character sheets to open up their FoA. Their FoC is generally fairly limited mechanically, and they don't particularly care about the narrative aspect. The core aspect is that most situations presented in an OSR game should be solved by exploring methods of resolution, players present options and the GM determines outcomes. A good way to think about this method is that it is a series of puzzles with no defined solution. Combat in these games is fairly restricted and limiting, but I think that this not as important to the discussion because the approach should be that combat is actually a consequence of failure; combat is a way to determine the damage of failing a puzzle, it is not the intended primary form of gameplay.
Reign is the system I have been running the most in recent years, and it fits my needs really well. Part of that, I think, is the amount of freedom it gives players and how it does that. Reigns FoA is medium, the game has a list of skills and generally meaningful action will engage one of those skills. The skill list is, however, large and covers many possibilities. In contrast to 5e for instance, there is much more mechanical support for social actions. In 5e there are 3 social skills which are defined by their intended consequence (deception, intimidation, persuasion), in Reign there are many more skills and they are defined by the approach being taken rather than the intended consequence (graces, plead, jest, etc.). Thus skills in Reign provide a method to approach the open intent of an action; For instance, Graces could be used for deception, intimidation, or persuasion depending on the situation and nuance of the characters actions. Thus, FoA is fairly good, as the skills largely provide a likelihood of success of an approach, not the ability to achieve a specific outcome. Reign uses a single MoR, and is procedural until consequence. What I think makes Reign worth playing is the Freedom of Consequence. Although not as open as FATE, Reign has the most FoC I've seen in a more traditional approach game. There are multiple metrics of sucess that can be measured by the dice rolled in Reign, and the players can be given the choice of which to prioritize. Fundamentally, the two metrics are Speed vs Quality, and they are achieved through different metrics on the dice. The decison between doing something faster vs doing something better can be a very interesting one, and gives that level of control over consequences that most games generally lack.
I think this model can be a useful tool when considering how mechanics affect fundamental goals in our game design. A more narrative game probably wants higher FoA or FoC, whereas a tight and balanced combat game wants to be more limiting in those aspects. I think Freedom of Resolution is the least explored of the three, as I think only OSR is even touching on messing around with it. In the future, I might think about how to implement high Freedom of Resolution in a game - I think it is an interesting design puzzle to solve. The model is not all encompassing for a game, it cares only about the process of player decision to consequence of action. There are certainly things that fall outside that scope, such as Freedom of Expression (although I do think that FoE is affected strongly by the freedoms in this model).