What do you consider to be some of the deepest games ever, or at least by you?
Considering depth is my standard of quality, you could just ask what I think the best games are.
I’d probably go with like, Guilty Gear AC+R, Project M, Starcraft Brood War, Quake, most of the best stylish action games. Probably throw Go in as well, that really stands out among tabletop games. Continue reading
What’s wrong with the term fail state?
It’s attached to the definition of game for many people, and it doesn’t mean anything real, so it causes semantic fuckery when people try to argue about what constitutes a “fail state” and whether a given game has one.
So what counts as a “fail state”?
Here’s an obvious one most people will agree with, game over. Meaning you reset the whole game, do it over from the beginning. You’ve lost the entire game. It’s all over. Multiplayer games have this as well. You can see this in tetris, contra, street fighter, and a bunch of others. Kind of went the way of the dinosaur except for short games and multiplayer. Continue reading
Do you think third person stealth games are inherently weaker than first person ones? Even when you get x-ray power in first person games like Dishonored, you still don’t have the same advantage of being able to see all the enemies around you, that a third person camera, inherently brings.
Depends on how important you think the game of information is in stealth games. early MGS and other top down or side-on stealth games don’t really prioritize this aspect at all (such as monaco or mark of the ninja). Though monaco and mark of the ninja (at least in NG+ mode) both hide where out of sight enemies are, only illuminating areas in your field of view (executed surprisingly well in mark of the ninja, not the type of thing you’d expect to work in a 2d game). Continue reading
What do you thinik of Blizzard’s collaboration with Google to developing sophisticated Starcraft 2 AI using Google’s Deep Mind?
I’m looking forward to it. I’m excited as hell. However I think they should try a version where they train it with the visuals on a 250ms delay, because unlike chess, this is a realtime game. Humans realistically operate with a delay between when something actually occurs, and when it reaches our brains. An AI playing a realtime game can frequently employ strategies that are not only better than human ones, but which are literally impossible for humans. My classic example is SFA Akuma, which will walk up to you, and if you press a button, shoryuken, if you block, throw, and if you shoryuken, block you and punish. It destroys the RPS loops that define the game. It’s effectively not even playing the same game arguably.
The AI in Starcraft itself is already way better than any human player (just way stupider), with up to 3000 APM depending on what it’s doing. It can operate every unit individually if it wants to, using units like Ghosts to hard counter mech builds with the lockdown ability fired from every ghost individually onto each individual unit, when in real starcraft, ghosts are practically useless because no human player can possibly micro like that.
You’ll notice that the lockdown ability didn’t return in starcraft 2 where ability units like the high templar are rigged to only have 1 unit cast the relevant ability when multiple of the same unit are selected, because it would have made the above tactic really really easy.
The point is, in realtime games, unlike turn based games like Go, computers are frequently able to beat humans in simple ways that don’t really reflect the way the game is normally played. A 250ms delay might produce something that looks more like optimal human play rather than play that is borderline rigged cheating.
Like, we want AI to beat us in playing roughly the same game we’re playing, ostensibly, but if you remove this human limitation, then it’s like you’re effectively playing poker without hiding your hand, not very interesting.
You often mention the RPS model in some of your examples and answers but what does it entirely mean? I mean yeah, it’s Rock-Paper-Scissors, but how is this important in the context of video games?
In single player games it’s debatable. In multiplayer games, it practically is the game, assuming there’s hidden information of any kind, and that hidden information can have a disproportionate effect on the other player.
Most multiplayer games can be modeled as a huge web of rock paper scissors interactions. My Street Fighting for Beginners guide lists a few that are common in various fighting games.
https://critpoints.wordpress.com/2015/03/20/street-fighting-for-beginners/
I went over rush > greedy > turtle in the last ask, so here’s some other counters in Starcraft (because I’m not really familiar with other RTS games). Melee units usually beat missile units, air units beat melee easily, and missile units beat air. But of course it’s not really that simple. Missile units can beat melee units if they’re on higher ground, or far away. Air units can beat missile units with harassment and focus fire. Melee units aren’t ever really going to beat air units though.
Counters can be flexible. Fast moves don’t always beat slow ones, slow ones can win if they’re timed correctly. SFV even rigged the system to make it so slower medium and heavy moves can beat the faster light punches and light kicks if they trade on the same frame, making it so frametraps consisting of stronger buttons are better.
Counters can also have different levels of payoff. In SFV, if you block high on wakeup, and I hit you with a combo starting off a low, that’s usually a really powerful combo that can lead into another knockdown. If you block low, and I hit you high, then you take peanuts for damage. If I decide to throw you, then I get a moderate amount of damage and another knockdown. If you try to dragon punch me and I block it, then I get to use my best combo. On your end, if you block correctly, then you’re in a blockstring at worst, or can punish me at best. If you throw tech or jump versus my throw, then you’re home free. And if you uppercut my attack, then I take a strong single hit and get knocked down.
Good multiplayer games are built off webs of counters with situational risks and rewards, that also can counter more or less consistently based on the situation.
Good single player games can take a lot from this principle to build deep gameplay, but true counters aren’t good singleplayer design, because the player needs to be able to always win.
True counters are enforced in multiplayer games by hidden information. This can come in the form of the reactionary blind spot, or the fog of war in Starcraft. Real-Life Rock Paper Scissors only works because of the reactionary blind spot. You throw close enough to the same time to prevent the opponent from reacting and changing their throw. Sirlin’s game, Yomi, uses a hand full of cards your opponent doesn’t know, and laying down both cards simultaneously to get this effect.
Not all multiplayer games work this way, like Racing games most of the time, but all of them featuring direct player interaction do (so racing games feature this when your cars are close and trying to pass one another). I call the other style of game, “efficiency races”.
Back in your talk on RPS model you talked about true counters. What exactly is a true counter that makes it different from a regular one?
Alright, it’s not an official term or anything that people use. I was just trying to distinguish between the way in multiplayer games you can have things that just flat-out counter each other (like rock paper scissors) and in singleplayer ones, you can’t really do that. You can’t have an enemy that just straight-up counters the player’s actions, otherwise it turns into a guessing game, rather than a game of skill. So enemies in single player games can’t truly counter the player character’s choices, they can at best have soft counters that make things harder on the player depending on the player’s choice of action. If you play RPS with a computer, the only way to make it fair is if you see their throw first. A lot of single player games end up like this, they throw rock paper or scissors, and you need to react quickly to throw the one that beats it.
Enemies in single player games need attacks and behaviors that limit the effectiveness of easy player strategies, like running away, or running past the enemy. These can be framed as counters. Player likes doing this, finds it really easy to do, give the enemy a behavior that counters that so the player needs to play a bit more honest. However as above, it cannot truly counter it by totally shutting it down, it needs to be flexible and create a gameplay challenge in its own right. You can give the enemy a whip to pull back in players who try to run away, or give them a bullrush ability to rush down fleeing players, but in both cases, you need to make it so the player can dodge, block, or counterattack these things. So you can have these type of loose counters aimed at making certain player strategies less effective, but you can’t make an enemy who can just pick an option that beats whatever the player is doing.
Something brilliant about the Elites in Halo is the way they dodge. They never dodge your hitscan fire pre-emptively, only after being hit first. Many enemies in other games, like God Hand, Curse of Issyos, or Dark Messiah, don’t follow this rule, which means that sometimes they randomly don’t take damage for all intents and purposes and there’s nothing you can really do about it. Elites, they take a bit of damage at the start, then you can react to them dodging after the initial burst of fire, and keep shooting at them as they dodge.
What do you think of the book “Game Mechanics: Advanced Game Design” by Joris Dormans?
I skimmed through it, and it seems pretty good. One criticism I’d have is the use of the Machinations flow diagram system for describing game logic, but it still has a lot of great textual examples backing it. I think that these flow diagrams never really work out too well (Raph Koster’s game grammar was kind of a disappointment, the one used in the book seems a bit more realistic though). They don’t have strong descriptive or predictive capabilities most of the time. I’d honestly prefer code or written word descriptions usually. But I mean, it’s not really taking away from the book at all.
Overall, the book features tons of good and useful information and examples of different game mechanics and means of implementation. I’d totally recommend it!
One critical thing it leaves out though is rock paper scissors. It loosely alludes to it when it goes over rushing versus turtling (RTS games have the early game counter loop of rush > greedy > turtle > rush). Greedy builds spend all their early resources on maximizing the rate of resource acquisition. Rush builds spend those resources on fucking people up as early as possible. Turtle builds spend those resources on defending against a rush. It’s pretty obvious to see rush beats greed and turtles beat rush. Greedy builds beat turtles because the turtles don’t have as strong an economy, and lose out in the long run. In Starcraft the meta has kind of settled on mixed turtle builds being the best early on (far as I know). The book sort of alludes to greedy builds on page 69, but it doesn’t say the word greedy anywhere and kind of forgets about this early example when it goes into turtling versus rush later on.
Without the dynamic of rock paper scissors, I think the book misses something really important to any type of multiplayer game, but it makes sense given the simulation game background the book appears to be coming from.
Actually, now that I think about it, I don’t think the economic modeling system used by the book is entirely appropriate to describe platformers, or action games or FPS games either really. It’s kind of limited to games that are more strictly about economics. Hmm. It does mention Dan Cook’s Skill Atoms, but it talks about those more in a structure similar to locks and keys in order to teach the player a chain of skills, rather than looking at the game dynamics from the perspective of a player who is already experienced and merely engaging with emergent systems composed of familiar components and familiar skills in an unfamiliar arrangement.
So, still a good book, but I guess it’s not comprehensive. Good if you’re building an economics game, or game that features economics somewhere, useless for building a racing game, or most action games including fighting games and platformers.
What do you think about games having menus that freeze the in-game action? Should more stuff happen in real time? I vastly preferred how TLoU handled the backpack feature, than MGS3. gilgamesh
I’m kind of ambivalent on it. I’m fine with a lot of games freezing time for menus. In most cases, I don’t think it seriously matters.
Cases where I’m not fine tend to be games like Skyrim, where you can pop into a menu and heal yourself with an infinity of healing items. If you could do that in Demon’s Souls or Dark Souls, that would be horrible.
Basically, when you have a menu that pauses time, you’re effectively making any action that takes place inside the menu instantaneous. This is why all the games with weapon wheels slow time while you’re looking at the weapon wheel. Because they want switching to be instantaneous in game time.
If healing is instantaneous, it’s not risky. If you can carry an infinite number of healing items and healing is instantaneous, then you have infinite health. Megaman has subtanks or etanks you can use to heal in menus, and that’s fine.
If healing is limited, then instantaneous healing means you effectively just have a slightly bigger health bar. If you can carry unlimited healing items, then you have an unlimited health bar, which sucks.
Menus not pausing means that whatever’s in your menu, you better have set up before combat comes, or take a risk setting it up when combat’s happening. Witcher, they don’t want you brewing potions mid-combat, they want you with that stuff set up ahead of time (menus in witcher do pause, but you can’t brew potions I believe). In Dark Souls, similar deal, you gotta put on the right equipment ahead of time or pay the price in the moment.
I think it ends up being more about tone than substance most of the time. You can choose to have them real-time to make switching things in combat riskier if you feel that’s important to the tone of your game, but I can’t think of any specific game dynamics that rely on the player not being able to instantly do whatever in the pause menu.
Do you think the beat em up genre could do with a level of shaking up in the AI department? I find that the enemies mostly repeat the same moveset, and don’t act all the dynamic most of the time. Although the same could probably be argued about the enemies in a lot of martial arts films.
Here’s one big change: Record when the player deals damage, and perform attacks with inverse proportionality to where damage is taken. Keep updating this as the battle goes on. So if the player is only trying to deal damage during one particular cycle, or punish one particular attack, start using that attack less, forcing them to get damage elsewhere. This can help force continuous aggression from the player and improvisation instead of really safe play.
Beyond that, better AI is really a matter of thinking up specific behavioral characteristics. Making observably good AI is hard. Halo enemies dodge only after they’re initially shot, or if you throw a grenade or fire a slow moving shot. They can’t dodge pre-emptively or on reaction to your shots, so their dodging behavior is predictable instead of random. Continue reading
What do you think of the combat in the Mass Effect series?
I played about half of Mass Effect 1 and beat Mass Effect 2, but it was so long ago I don’t really remember it. I remember it was REALLY boring cover shooting. Mass Effect 1 was a poorly optimized abomination that managed to run poorly on my much newer than the game computer, when ME2 didn’t. Enemies don’t move a lot. They shoot hitscan bullets, there’s the usual regen health/shields. You play peekaboo with cover and try to tank enemy hits while shooting them, because they take damage permanently and you don’t. They also don’t take hitstun, and all your weapons use the same ammo (ME2), or have cooldowns with unlimited ammo (ME1). There isn’t a lot of reason to move around during combat, because enemies don’t move around much, and you’re basically always using mid/long range effective weapons. Occasionally enemies throw grenades, I think. I don’t think you have grenades, only cooldown abilities, some of which might resemble grenades. Continue reading
Why do you strongly dislike using controllers for first-person shooters? What exactly is wrong with using a controller as opposed to a mouse and keyboard? While I’m at it, what do you even think about console shooters (shooters made strictly for controllers).
Basically, unlike fighting games, controllers are a completely inarguably inferior method of control for first person shooter games. In fighting games, and a lot of other (digital input) genres, using a controller over another input method is usually a matter of preference. In first person shooters (and RTS), mice are the best way to control the game, period. Continue reading
Celia Wagar's CritPoints 