Development Info Tim Cain at Reboot Develop 2017 - Building a Better RPG: Seven Mistakes to Avoid

[mondblut]

fairly good RPGs with simplified combat

You have to choose one.

 

[Ruzen]

I thought the presentation was good. It's his accumulated experience over many years. I didn't understand the mass dislike at first but after reading pages and pages about people complaining Cain's "geometric shapes" description. People kinda stop thinking after that or people didn't even listen to all the presentation and miss the point. I really want to argue here but I don't know where to start. Most of you overreacting.

 

[Sitra Achara]

Tim giving up on complexity shouldn't be a shock - he did already give up on women.

 

[existential_vacuum]

Hey Vault Dweller, how many geometric shapes does your new game have so far?
I bet not even 20, so much for it being a good rpg, r00fles.

Codex right now

EDIT: fixed image hosting fuckery

 

[DraQ]

• Mistake #1 - Steep Learning Curves: Tim thinks character creation in Fallout, Arcanum and other RPGs was too complex. He's experimenting with creating a completely numberless character system that uses geometric shapes to visualize attributes.
• Mistake #2 - Letting Math Trump Psychology: Revealing the influence of the years he spent developing Wildstar, Tim wants to develop mechanics that are psychologically satisfying and addictive, even at the expense of mathematical elegance. For example, he says the player's first attack against an enemy should always hit even if his overall hit percentage is the same regardless, and that rather than allow players to increase their critical hit chance, they should only be allowed to increase their critical hit damage.
• Mistake #3 - Conflating Player Skill With Character Skill: This one will be familiar if you've watched some of Josh Sawyer's talks. Aiming and hitting in an action-RPG should not be determined by character stats. On the other hand, things like the impact of recoil can be affected by stats, as well as the aforementioned critical hit damage.
• Mistake #4 - Misunderstanding Randomness: Here Tim lays out his frustration with the sorts of people who can't believe they could miss a 95% chance-to-hit attack three times in a row. His conclusion is that when people talk about "randomness", they often mean selecting a token rather than rolling a dice (ie, events can't repeat themselves).
• Mistake #5 - Forcing Linearity: This one is pretty self-explanatory. Tim says games are not movies, using Fallout's Tandi rescue scenario with its multiple solutions as an example of the sort of non-linearity he prizes.
• Mistake #6 - Being Non-Reactive: Tim seems particularly interested in the sort of reactivity where characters in the world have different dispositions based on your character's background, clothing and attributes, as seen in Arcanum. He also loves having different end slides based on the player's choices in the game, using Temple of Elemental Evil's evil ending as an example.
• Mistake #7 - Telling Horrible Stories: Tim uses this to emphasize again that games are not movies. Not every character in a game has to be important or advance the plot. Tropes likes the Chosen One protagonist and amnesiac protagonist are tiresome and should be discarded.

1. There is no such thing as steep learning curve. There are only dumb players and badly conveyed information/lack of feedback. Fallout's chargen was dead fucking simple, because the structure of statsheet was flat, the point pools to distribute between primary attributes and skills were distinct, and everything was adequately explained. The only problem with FO's chargen was that some skills were fucking worthless or varied wildly in utility between different stages of the game, but that has nothing to do with complexity, only with piss poor match between stats and content. Arcanum's chargen was worse, but that's not the fault of it's complexity, only of its needlessly muddled hierarchical structure with shared point pool for different levels of hierarchy. You don't need geometric shapes to visualize numbers between, say, 1 and 10 or even 100 and if you do, then who the fuck is buying you games?
   
2. I don't even know where to begin describing how much of a reeking pile of wrong this is. First, good systems should be symmetrical and oblivious to the difference between PCs and NPCs - this would mean that the first attack against player should also auto succeed which spells doom to all sorts of non-tanky builds, heavy hitting enemies and so forth. Second it pretty much spells out HP inflation because otherwise this first attack would be too weighty and lead to cheese (like giving shitty combatants the most powerful weapon to execute guaranteed 1hks against enemies). Of course this in turns pretty much unravels ambushes and so forth. If you are not satisfied with simple random roll, then come up with something nuanced. Even in low fidelity system not concerning itself with minutiae of an attack (such as nested hit volumes), you can do much better than random roll. For example:
   • you can assume that intentional undefended melee attack always hits and that intentional, undefended, unmitigated melee attack always kills (or at least cripples if you're detailed enough for that).
   • defense is active (so target that is paralyzed, unconscious, etc. cannot defend itself), is possible only within certain angle (that can be modified by stats and status effects) and can be conducted only limited number of times per turn (TB) or has brief recovery time (RT), stat based and may also depend on angle between consecutive attacks.
     
   • mitigation is passive (things like armor, weapon being wrong tool for the job - for example dagger VS something big, some status effects and attempted defense may be mitigating factors) and may reduce or nullify effectiveness of an attack.
     
   • Ranged attacks can miss even without defense based on various factors including distance and can also be mitigated by those factors.
     
   • Fast ranged attacks (bullets, lightning, arguably crossbow bolts) cannot be defended against
   There, you have a simple system much less reliant on random rolls, that nevertheless isn't pants on head retarded and minimizes HP inflation and level dependent immunity to effective tactics. Randomness is a spice, you only add enough of it to make things interesting and somewhat unpredictable, not to make a roll play game (fortunately the randomness also gets reduced if you add more of it). The problem isn't math trumping psychology. The problem is simplistic RPG systems trumping deeply ingrained intuition.
   
3. To a degree that might be reasonable. If you want to make an actiony game then you probably don't want straight to-hit rolls if the player is aiming the attacks. OTOH you probably want to keep overall influence of stats on actually hitting (reliance on damage scaling is dumb and leads to shit gameplay - it also applies to crit damage) so hooking stats into as many elements of offensive action as possible can get you an effective to-hit check without having player miss attacks they have clearly seen as hitting the target. Alternatively you can keep simple to-hit mechanics and try to obfuscate with the right animations, but making it work may be much more complex than doing it the "hard way". Of course if player doesn't control the character(s) directly, this point doesn't apply.
   
4. Player misunderstanding randomness is player's problem. The real problem is when the system isn't fine-grained enough to prevent being fucked by a bad roll alone, and when player has different intuition of what should happen than simplistic random function would dictate. That's been discussed above, so the only thing worth mentioning is that attack results aren't like a bag of tokens. Bad abstractions breed bad abstractions breed retarded gameplay.
   
5. Can't argue with that.
6. Can't argue with that.
   
7. Mostly yes, but amnesiac protagonist (together with fish out of water) has special place in games - it allows player who knows nothing about the setting to control their character and make decisions in a reasonable manner as far as the premise is concerned. You don't have such problems in film or literature where audience doesn't control anything. In games this trope is a tool performing specific vital task and shouldn't be discarded.

 

[Don Peste]

I don't get how Fallout character creation was any complex. This small sheet? Anyway, I haven't had time to listen to the talk yet, but I'm sure it was interesting.
Most of us liked everything Cain & Boyarsky made. If now they end up making a proper Mass Effect-like RPG, but with an actual interesting story, conversations, choices, NPCs... because that's the higher they are allowed to aim... Well, that can be a good game, too. It's not going to be Fallout, or ToEE, or Arcanum, but it can be good, too.

 

[DraQ]

-RPGs ... without stats

 

[Black]

Mass Effect-like RPG

Doth not compute.

 

[Infinitron]

Player misunderstanding randomness is player's problem. The real problem is when the system isn't fine-grained enough to prevent being fucked by a bad roll alone, and when player has different intuition of what should happen than simplistic random function would dictate. That's been discussed above, so the only thing worth mentioning is that attack results aren't like a bag of tokens. Bad abstractions breed bad abstractions breed retarded gameplay.

I should note that unlike the other items on the list, for this one there was no clear implication that Tim intended to implement such a mechanic in his game (although he might). If you listen to it, you'll understand.

-RPGs ... without stats

By "stats" I meant primary attributes, not skills, perks, various derived statistics, etc.

 

[DraQ]

-RPGs ... without stats

By "stats" I meant primary attributes, not skills, perks, various derived statistics, etc.

Stats are stats are stats.

 

[DraQ]

Having attributes range from 1 to 5 would be acceptable (most of the uneven number ones didn't usually do jack shit) if the underlying system was deep and reactive. For a real-time game it could be done on a completely deterministic level and still be deep and compelling if done right.

Mostly agreed but even pure action games are rarely deterministic. Even DOOM used RNG.
A little randomness is good, it prevents rote repetition and forces contingencies.

 

[Quillon]

I immediately hated the triangle system since it reduces flexibility, maybe perceived flexibility but that's what matters anyway. But if they'll go with this for the game they are working on atm; I hope, at least the triangles aren't fixed like one triangle for physical(str-agl-en) and one for mental(per-cha-int) prowess so that we can put f.i. str and agility in different triangles and make the 2 stats "great" instead of the possibility of one "great" stat for each prowess. I still wouldn't prefer it tho. For one thing this prolly is making the stats fixed at the start and restricting upgrading during the game. I don't know, maybe skill/perk system has to be varied and great for this to not suck.

 

[Roguey]

I don't get how Fallout character creation was any complex.

That's way too much reading for a console player who wants to choose a few options and start playing. A first time player is going to be looking at that screen for a long while.

 

[FeelTheRads]

In fact a console player can only comprehend one stat at a time as seen in Fallout 3 and NV. Doesn't matter you can't see what that stat does to any other skills, making the choice pretty much arbitrary.

 

[J_C]

I don't get how Fallout character creation was any complex.

That's way too much reading for a console player who wants to choose a few options and start playing. A first time player is going to be looking at that screen for a long while.

But there were some pregenerated characters for those type of players. Why isn't it enough to have both pregenerated characters and a detailed char-gen? But no, the whole game has to cater to the dumb kids.

 

[Roguey]

But there were some pregenerated characters for those type of players. Why isn't it enough to have both pregenerated characters and a detailed char-gen? But no, the whole game has to cater to the dumb kids.

http://www.rpgcodex.net/forums/inde...-mistakes-to-avoid.114809/page-7#post-5075662

 

[CyberWhale]

I don't get how Fallout character creation was any complex.

That's way too much reading for a console player who wants to choose a few options and start playing. A first time player is going to be looking at that screen for a long while.

Sure, that's the reason for those psychological tests they've added to F3/NV. Problem? Even more reading.

Why not make a small tutorial area composed of two parts:
1. a linear one where someone (present or distant NPC) is guiding you through basic options
2. non-linear multi-solution but enclosed one where computer tracks your gameplay style and using those metrics creates a character sheet

After its generation, you could still modify it for your liking, the same way you can do it now (after answering those questions).

 

[FeelTheRads]

But there were some pregenerated characters for those type of players. Why isn't it enough to have both pregenerated characters and a detailed char-gen? But no, the whole game has to cater to the dumb kids.

http://www.rpgcodex.net/forums/inde...-mistakes-to-avoid.114809/page-7#post-5075662

Yeah, I think that analogy is not exactly correct.

I, for example, am perfectly fine riding in the back. I imagine similarly there are people who are perfectly fine just playing with pre-generated characters.

Which of course is not enough for developers who want to please everybody, though... the typical "our game is perfect for all kind of players!!!" utopia that everyone's obsessed with it and that's a recipe for failure.

 

[Roguey]

Sure, that's the reason for those psychological tests they've added to F3/NV. Problem? Even more reading.

It's staggered out though.

Yeah, I think that analogy is not exactly correct.

I, for example, am perfectly fine riding in the back. I imagine similarly there are people who are perfectly fine just playing with pre-generated characters.

Which of course is not enough for developers who want to please everybody, though... the typical "our game is perfect for all kind of players!!!" utopia that everyone's obsessed with it and that's a recipe for failure.

The person who's fine riding in the back is just going to watch a Let's Play for free, not pay money to play the game.

 

[Lurker King]

What's with this trend of making CRPGs for people who don't like CRPGs.

They're making action games that they arbitrarily define as RPGs. This way they can sell their game to the dumb masses and the much smaller RPG audience.

And still deflect any unfair comparison to actual action games.

 

[Azrael the cat]

Rule and corollary:

1. Don't denigrate the achievements of great developers because of the shit they made after they were no longer great. One truly great game >>> a career of okay-ish games.

Corollary. Don't expect a developer to not be shit, simply because in another time, another situation, they made a truly great game.

 

[FeelTheRads]

The person who's fine riding in the back is just going to watch a Let's Play for free, not pay money to play the game.

I disagree. I don't see how playing with a pre-generated character is the same as watching the game. You still play the game, you just made a safe choice at the start. Not much different from choosing a difficulty level.

Now, sure, there are those who want to do everything themselves but will cry if it's too difficult for them and want everything scaled to their level, including the highest difficulty setting which is now basically "normal", but at some point you have to say fuck you to certain groups of people.
Not everything has to be inclusive. And not everything can sell Skyrim numbers no matter how hard you try. More likely you end up with disfigured mutants like Obsidian does.

 

[Hellraiser]

1. There is no such thing as steep learning curve. There are only dumb players and badly conveyed information/lack of feedback. Fallout's chargen was dead fucking simple, because the structure of statsheet was flat, the point pools to distribute between primary attributes and skills were distinct, and everything was adequately explained. The only problem with FO's chargen was that some skills were fucking worthless or varied wildly in utility between different stages of the game, but that has nothing to do with complexity, only with piss poor match between stats and content. Arcanum's chargen was worse, but that's not the fault of it's complexity, only of its needlessly muddled hierarchical structure with shared point pool for different levels of hierarchy. You don't need geometric shapes to visualize numbers between, say, 1 and 10 or even 100 and if you do, then who the fuck is buying you games?

 

[Lurker King]

Don't denigrate the achievements of great developers because of the shit they made after they were no longer great.

Nobody is denigrating their achievements; we are denigrating the developers themselves.

 

[DavidBVal]

WTF... Arcanum problem was the character development? has this man completely lost it?