Sunday, July 25, 2010

Level Scaling - The Why

I'm going to spend a few posts discussing the issue of a much maligned game design trope - level scaling.  I recently got into a "discussion" (if you could call it that) on the BioWare forums about the mechanic.  If you've read those posts, some of what I say here will be repeated from there, but I feel it is something that actually deserves to be examined with some thought.

One of the reasons level scaling is much maligned is because it is examined on its own as an individual mechanic. Breaking down a game into its components is a good means to analyse a game technically, but the problem can be that the individual criticisms don't take into account how the game works as a whole. There's a "can't see the forest for the trees" problem occurring for many people.  Interestingly enough, this seems to be the issue that many US game reviewers had with the game Alpha Protocol.  Here is a very interesting article discussing the vast divide in reviews of the game from the US and Europe. I have to say, from my perspective, it seems like the US reviewers were expecting an FPS or a Mass Effect. Alpha Protocol is neither; it is a great game in its own right, albeit with some technical/polish issues.

I imagine Thorton would use these options on an Internet forum too...

Generally, one of the core aspects of an RPG is considered to be the process of leveling up your character.  Through their actions, they become more experienced, which subsequently lets them become more powerful. In most games, a large portion of this power is demonstrated through combat, as the player's hero becomes more capable with a weapon or magic, but also with other supporting skills. Players have to fight at some point, so they need to do something to make them more capable in dispatching enemies. If you can find me an RPG with no mandatory combat, I'll be surprised and impressed. (Actually, I think that's called an adventure game) Players like having this progression, as they like reaching the end of the game knowing that their former rat-slaying weakling has dispatched trolls, giants and dragons on their way to saving the day/world/land of cute puppies.

As such, the enemies increase in difficulty as the game goes on. Take Baldur's Gate 2, where you go from killing goblins, slavers and trolls to facing off against an entire drow city, beholders and dragons.  Or Diablo 2 (why did I pick two sequels?), where you go from beating quill beasts and creatures that run away when you kill one of their friends, to a giant charging slug of doom, jungle-midgets-from-hell, demonspawn (and demonspawners) and a prime evil in the form of Diablo himself.

What's 3 feet tall and delightful to kill? These things.

So as we progress through an RPG, our character becomes more powerful, and as such, the enemies need to become more powerful as well. If the former doesn't occur, the player has a reduced sense of achievement. If you don't believe me, then you're at odds with everyone who has ever played World of Warcraft, or done something that was not entirely fun for the sake of earning an Xbox achievement or PS3 trophy. People (gamers in particular) like achieving things and getting rewards, no matter how small they might be.

If enemies don't become more powerful, then we fall into a terrible situation where the combat becomes repetitive and laughably easy, both of which inspire boredom. The last thing a player wants to be while playing a game is bored. Hence why developers/designers do everything they can to stop the player from being bored. So therefore the encounters the player is involved in have to become harder, so the player takes advantage of their improved skills, tactics, items or anything else that might affect gameplay.

Having established that it's essential for a player to have that sense of achievement, there's a few means of delivering that. For starters, you can pit the player again new enemies, which will look different from previous enemies, most likely behave differently, possibly use new abilities or weapons, and definitely be tougher (ie take or deal more damage) than the previous enemies. The designer could have the player face off against a horde of weaker enemies they've faced previously, demonstrating how much more powerful they are by handily dispatching an army of enemies that previously posed a threat. It is also possible to put the player in fantastical "boss battles" with amazing enemies that require special tactics to win is one excellent means, and also serve as natural climaxes within the flow of the game's plot and its difficulty curve. I'm sure you can think of other ways to make encounters more challenging too.

Take 2 of a fight against the same boss. This time, you win.

So fights have to get more difficult as the game goes on, which is more than achievable by pitting the player against increasingly difficult enemies.  That's been happening in games for years. But in the past few years, games have become increasingly complex. We're no longer always pushed down a linear path. However, this poses a problem. If we're no longer dealing with a linear path and there are several major quest lines to pursue, how do we keep combat challenging across the length of the game?

Let's say we're dealing with four key-quest lines as part of the main plot. By necessity, if we're making the game non-linear, the player must be able to pick any of these four quests from the start. If they can't... well, the game isn't non-linear, is it? At which point you're simply providing the illusion of choice, and beating the player into a pulp if they don't take your "suggested" path.  In that case, my bother giving the illusion of choice? Just force the player down the linear path that they have to take anyway. It is a lesser evil to pretend that the player has choice when they don't than to simply not present it as an option.

The illusion of choice...

So in our scenario, the player can pick any of the four quest lines from the start, and will likely gain a lot of experience (and hence combat prowess) from finishing one of these. If they then go to another one of those four areas, unless the difficulty of the combat encounters is increased... that's right, boredom sets in.  So we have to scale the difficulty of the encounters to match the player's level, which sounds like a job for level scaling!

Unfortunately, level scaling doesn't always work out for the best, and has numerous problems associated with it. Those issues are what I'll be addressing next time.


  1. Increasing challenge is probably more a matter of matching increasing player skill. If character progression allows choices or at least novel abilities, then hopefully it matches that curve. If the numbers just get bigger it's not even an interesting problem ;)

    Addressing the question more directly, I dislike automated level scaling (particularly literal scaling where the power of the creature is set algorithmically). It's very difficult to maintain the tuning of a really good boss fight under those conditions (note the wonkiness of more number-based Icecrown Citadel boss fights with the stacking buff and gear, even while crossing the normal-heroic divide).

    I don't feel non-linearity really justifies the loss of good encounters, either because they get outscaled or because you never implement them in the first place (As much as I love DA/ME/Elder Scrolls etc, none of them have any really great encounters to my mind). In Classic Week I dealt with this by offering limited but real choice (Vileshadow and Juggernaught were both difficult encounters, but that difficulty would be strongly affected by the order in which you did them). There is a better way, but it's more work.

    D&D, particularly 4th Edition, handles the level curve quite gracefully. A level gives an expected power standard, which player choices can put them above and below. This midpoint is baked into the tuning of the enormous antagonist resource pool, which the GM then uses to make an encounter. Ultimately balance is up to the GM, but the system helps them at every step. The key is that different resources are appropriate at different levels, NOT the same ones simply scaled algorithmically.

    Similarly I think the best approach for achieving non-linearality while maintaining encounter integrity is to simply redesign the encounters for the expected power of the PCs when they reach them. So if you have four major quests, you should have four versions of each end encounter (and any other major encounters on the way). This is a lot of work, but still less than what we do already for dialogue and story choices.

    At least for me. I find encounter design very efficient in terms of gameplay delivered vs work put in, while dialogue is an expensive chore ;)

    - Mengtzu

  2. Curses! You jumped ahead several posts on this topic and just divulged one of the workable options for implementing scaling - albeit one that actually requires a fair amount of work. :-) Also, I don't think I ever find dialogue a chore, although I won't argue that it is frequently time expensive. (What can I say? I love writing)

    In all seriousness, you've made some good points here. I agree that development plays out more meaningfully if it is skill based as opposed to pure numbers. Doing an extra 15 damage with an attack is far less exciting than having an entirely new attack that also does an extra 15 points of damage. I won't go into too much more detail on other points, as that will take away from subsequent posts I going to make!

    Thinking about it now, highly memorable encounters might actually deserve a set of posts like my ongoing antagonist series...

  3. I have yet to try Alpha Protocol, but screenshots like the one above is so demoralizing to look at. I'm tired of RPGs categorizing dialogue choices into their archetypal nature. When the dialogue is created from that perspective, there's no nuance. It becomes more mathematical and less emotional.

    "If you can find me an RPG with no mandatory combat, I'll be surprised and impressed"
    You might be interested to know that could do this in Deus Ex and the earlier Fallout games. Deus Ex 3 is supposed to have this feature as well.

    I just realized this comment had nothing to do with the topic. Oh well...

  4. Discussion is never bad, whether it relates directly to the topic or not.

    You've raised an interesting point on Alpha Protocol's dialogue system and I'm curious... how is that approach any different to having three or four lines with fully written answers that effectively boiled down to those emotions? Typically that's what we end up in even if we have a silent protagonist and detailed response lines. Nuance in the writing doesn't typically play into the actual result of the response, so it's more imagined than real, and worse, has the potential for the player to pick a choice they didn't actually want to because their interpretation of the nuance of the line is different to the writer's.

    Now, I love the "old-school" RPG silent protagonist with lengthy prose responses, I'm just playing devil's advocate here to demonstrate a point.

    That said, I love Alpha Protocol's conversation system. The fact that it operates with a time limit requires that the conversation be broken down into those one-word options. And the short time limit is what makes the system so great. It forces the conversation to flow in a natural and cinematic fashion and it forces you to make an instant "gut-feel" decision rather than giving you the freedom to contemplate a decision for minutes. That alone makes it a superb and unique roleplaying experience, and does an excellent job of removing the potential "meta-game" feeling you would have if you could take your time to make a decision.

    In short, I guess I'm saying don't discount Alpha Protocol because of that dialogue system. I viewed it sceptically before I played it too, but after experiencing it, I loved it.

  5. I wasn’t aware of the conversation timer. That means the simplified dialogue serves a real purpose. I could tolerate that in a fast-paced action game, but I’d hate to see it in a RPG like Dragon Age (I think Dragon Age 2 is getting this treatment). Games that utilize a categorized dialogue system also tend to use this as a way to rank your character by your responses (Mass Effect), but that’s a discussion for another time (and something I dislike even more).

    I think categorized responses restrain the writing and design. The characters become less believable as their behavior is configured after their emotion “weak points” rather than their character. This gives me more of a “meta-game feeling” than any other system, as the mechanics of the dialogue are exposed.

    How do you know the effectiveness of these “emotional responses” when you can’t see the content? How can there be any room for a player’s knowledge and intelligence in a system like that? What I knew something special about the person I was interrogating, would there be any chance of me ever being able to exploit that information? I don’t want to ask myself if is a good choice. I first want to evaluate the dialogue option – then evaluate what emotion will come out of it.

    If I choose a renegade choice when speaking to a NPC; I don't want him to respond one way or another because of the nature of the response. Dialogue options shouldn’t be about their “raw” emotion, it should be about the content within and the emotion it carries. This system (AP) makes the NPC respond exclusively to the nature of the dialogue option. I don’t think you can boil down any response to one singular emotion, at least not without it being arbitrary. The conversation carries the emotion, not the other way around.

    PS: please do something about the width on this comment box. I'm getting claustrophobia writing like this.

  6. edit:
    I don’t want to ask myself if is a good choice
    I don’t want to ask myself if insert-emotion is a good choice

    I had it encapsulated in tags so it disappeared.

  7. Interesting points, and I think I see where you coming from, in that it seems you'd rather have the line itself inform the emotional content rather than the other way around. Maybe it's just that typically I try to break dialogue down into the emotional content that the writers were going for, but this is perhaps a symptom of me constantly analysing the design rather than a failing on their part.

    Personally, I'd probably say that it's more about attempting to cater for specific player mindsets rather than pure emotion. In my Fate of a City mod for NWN2, I tried to have 5 choices for every dialogue catering for Good, Lawful, Neutral, Chaotic and Evil in the D&D archetypes. Thus for each choice the player was making, I had to ask myself: "How would a person with that mindset act in this situation what kind of response would they give"? What's more, I made the player's alignment shift because of their responses and alignment actually made a difference to how certain NPCs treated you and even the ending of the game. That said, there were many different choices and you had the potential to change your alignment a long way and then even get it back again depending on how you played. Some people didn't like the fact that I made alignment and roleplaying such a huge factor (some didn't like being actually *forced* to live up to their paladin oath of Lawful Good), but for me as a designer it provided a means for players to see the effects of their choices in a way that I felt facilitated roleplaying rather than meta-gaming despite the fact that the alignment stats are prominently displayed in NWN2.

    Also, it's funny you ask about a player using their intelligence and knowledge... with a number of NPCs you talk to, if you've gathered sufficient intelligence on them (in the form of dossiers that act like a "codex"), then at certain points you will have the option to use that intelligence to make a special dialogue choice that you wouldn't otherwise have access to.

    How do you describe the personality of the (presumably different) characters you've played in RPGs? And how do you write a dialogue system that supports all of those personalities and the millions of other personalities that other players will create to play through a game? Ultimately the problem is that we can't have an infinite number of dialogue choices or free form dialogue for the player, and as such designers are forced (at least to some degree) to produce categories to support different play styles of a broad player base.

  8. “in that it seems you'd rather have the line itself inform the emotional content rather than the other way around. “

    There's a difference between designers creating categorized dialogue versus a player's awareness of it. Just because you write dialogue options in broad/archetypal types doesn't mean the player will consciously experience them as such. Unless of course, you inform him. It's when you enforce that design by having immersion-breaking elements inform you "this is a renegade choice".

    If you make the categories visible then you are making a meta-gaming inclination to the player to play with them in mind. The dialogue becomes less about substance and more about these categories. Often these games are designed to support this kind of dialogue, like Mass Effect with its renegade/paragon choices (going as far as risking to handicap your character if you don’t tilt one way or another). You are also limiting yourself by categorizing by design (Mass Effect), contrary to categorizing as means to help yourself (Dragon Age). By doing this you don’t allow any chance of nuance in your dialogue. The borders become very apparent.

    Maybe I’m wrong; maybe I’m the only one that feels this way. I just feel inclined to play with these categories in mind when they envelop the dialogue. It takes my focus away from the character and conversation to the categories themselves. I’m not a min-maxer. I feel more immersed in the game when I play the character (choices, dialogue) with my own personality.

  9. I see your point, and I agree somewhat. However, there are two issues at play when comparing ME and DAO. First, DAO frequently has more than three choices, so it's more likely that you can pick one you like. However, more importantly is the second issue that DAO doesn't have the paragon/renegade system that makes it highly undesirable to pick the neutral option.

    For me, this is the big problem with Mass Effect's dialogue system. My "canon" Shepard is paragon, but occasionally will perform a renegade action. I don't recall picking a "neutral" response often, if ever. I'd argue that it's not the categorisation that hurts so much, it's the meta-game that punishes players for not consistently picking the same choice. Alpha Protocol actually does this well because there are no "wrong" choices. If you annoy someone, you'll get a different benefit than if you're nice to them, so you're never punished unduly because of the decisions you make.

    The difficulty for the designer is presenting concise choices to the player so they will have an idea of the tone of the character's response. As soon as we don't have the full line of the text (and sometimes even if we do) the exact nature of the tone can be difficult to tell. Short lines mean we need more "meta-information" telling us how our character will react, and this can be a failing. However, long lines also have a weakness, and I've hated a couple of choices in The Witcher where I was forced to pick between one of two really lengthy lines. I hated both response, so had to choose based on the decision they were making, it was a forced meta-choice.

    There needs to be a balance between the organic nature of a response and the meta-information provided to the player, and that's still being worked on by designers. To be honest, that's one of the reasons I'm interested in (and hopeful for) Dragon Age 2, because I get the feeling that by having icons to provide the tone of a response, we may not be stuck with "righteous do-gooder" or "heinous villain", nor hamstrung by conversation trees that rely on our tendency towards either extreme. Of course, that's just conjecture on my part, but I'm hopeful that they will deliver.

    All that said, I love seeing the full line of dialogue and picking based on that as opposed to getting a summary and meta-information, but it seems that many gamers can't stand to have a silent protagonist these days. Oh for the days when gamers could use their imagination.