Cain on Games - Tim Cain's new YouTube channel

[Shadenuat]

World design isn't optional.
By making more things in game optional, you are just making a not game.

 

[Roguey]

Zuggtmoy is a Chaotic Evil Demon Prince of the Abyss...
I don't see how that would've worked out.
Would she honestly team up with a bunch of (admittedly) powerful mortals?
More likely she would turn on them.

You can beat her into submission.

 

[NecroLord]

Zuggtmoy is a Chaotic Evil Demon Prince of the Abyss...
I don't see how that would've worked out.
Would she honestly team up with a bunch of (admittedly) powerful mortals?
More likely she would turn on them.

You can beat her into submission.

Sure.
But demons are Chaotic Evil by nature.
It's easier to just kill them.
Do you honestly wanna trust a Tanar'ri?

 

[Drakortha]

Man, he’s gotta be *really* running out of ideas now

He's suffering from what's called Youtuber "content creator" syndrome.

Basically people make a channel and set out to put their thoughts and experiences out there, and hopefully be able to share some knowledge. It's noble enough. But then for some unknown reason once the deed is already done and all personal knowledge has been exhausted they become compelled to keep making more videos. Even if it means regurgitating shit they've already done, or twist it in some way. Or sometimes they just start making shit up to justify making more "content".

Cunts just don't know when enough is enough. The Simpsons was a funny enough show. But we didn't need fucking 36 seasons of it.

 

[__scribbles__]

I talk about how dramatically technology changed (and continues to change) in the game industry and how that is a source of excitement and stress.

 

[StrongBelwas]

Technological change was excited and stressful.
When Cain was a kid, they would get little handheld games like Calico Football, then pong unit you hooked up to your TV, and the year after that the first console. Things kept coming and coming. Fun for a consumer, but you could buy something and a few months later it was obsolete. Really hit it's stride with PCs, you buy a PC a few months something comes out better in all aspects.
Cain will discuss his experience since 1981 of trying to make games for a constnatly shifting technology target.
The difference between making a game and writing a book or making TV/Movies is that the pace of technological change in those don't hold a candle to the pace of change in gaming. They've made better microphones, better cameras, figured out ways to light the movies better, but speaking broadly, directing is the same as it was say 20/40 years or so. Every time something in gaming changed technologically, it would make old games not work and add so many other options for future games. You can do realistic graphics now, how many? You can do cutscenes now, how many? Felt like an exponential expansion where you had to keep up or die.
Grand Slam Bridge may look graphical, but it is all text based. Used ASCII characters they didn't need to define and put in a text mode. This meant it works on any PC, and still works. Very interesting way of doing a game.
After grad school and diving back into games with Bard's Tale, now Cain was doing games in VGA. It was the new hotness, had to put in support for EGA and even older CGA. VGA gave them 256 colours at 32200 resolution, EGA only gave 16, and CGA gave them 4, and those 4 were predefined black white magenta and a purple color Cain can't name. Good luck making graphics in that resolution when you can't even pick your colors.
By the time he started Fallout in early 94 (Had to work out exactly when he started, probably January), but the very first meeting with Steve Jackson Games was March 94. Nobody besides Cain would be assigned until August. Fallout was 640x480 resolution with 256 colors, between four to five times the resolution of 32200. But to use this mode, called Super VGA, you had to bank switch, as the computer just couldn't access all the memory it needed. First few scan lines would be on one bank, next lines on another bank, had to tell the video card when to switch. When they did Fallout, every video card bank switched in a different way, and you had to learn that and do it in Assembly. No alternatives. You also had to ask the consumer what video card they were using, and that is assuming the consumer even knew what their card is. Sometimes it was different, do you have a Trident 9600 chipset A? Chipset B? Chipset C? They all bank switched differently. You can try auto detecting (not reliable) or ask them to guess by displaying a picture(pray they don't crash) and if they see something you could move on. Thankfully, VESA came out, but it did not come out in time for Fallout to take full advantage of it. Worked better than trying to guess. They discovered that Mac kept breaking all of the old software, and Apple cared very little about games (Could do a video about that, cares too little about Apple to bother.) Fallout would get invalidated by every new Mac update. Easier to play Windows Fallout in an emulator/partition on Mac than playing the actual Mac version.
Arcanum comes out, 800x600, still 256 colors, VESA now made everything easier. Cards shipped with VESA compliant driver. Just tell the card through a set of VESA interrupts and your resolution and colors and it would set it up. VESA would supply your bank switching routines and figure it out for you. Cain had dropped mac versions, but the windows and directx versions kept coming out while they were making the game, hard to keep up. Yes, Windows had backwards compatibility, and that helped and still helps. But not perfect, can be janky, cause issues in HUD and framerate. But when they were making Arcanum in Windows 95 and Windows NT, but then Windows 98 came out, then Windows 98 Second Edition, then Windows 2000, then Windows ME, and finally Windows XP. Had to make sure the game worked on all of these, did your operating system calls suddenly stop working? Windows cycled a new version every year, the rest of the time they ran Troika a new XP version came every year. Hard to keep up with.
All of this is just for 2d, Temple was the first use of 3D video graphics usage. The backgrounds were still 2d, but they had 3D collision mesh and 3D characters. They had to go through the 3D API from the card, and the cards all went 3D a bit differently. When they did Vampire, it was totally 3D. Learning a new way to make games, worry about polygon count and texture sizes. Math became very important for level designers and programmers. You could have gotten by in the video game industry back then in art or narrative with limited/no math. Not that they went 3D, had to know linear algebra, had to know vectors, had to know matrices. Huge shift in the industry. Cain's engineering school helped a lot here, Cain knew programmers that stayed 2D. Knew one that stayed 2D programming for the next 20 years of his career.
Hasn't even touched on languages, every game was C until Vampire, which was C++. Cain kept with C++ until Pillars which was C#, and then went back to C++ for Outer Worlds.
Made his own engines from the very beginning up to 2011, with the exception of Bloodlines being on the source. Obsidian had Onyx for South Park, then switched to Unity for Pillars. Unity didn't give you source code, you did what they told you you could do, if you couldn't you asked them, who knows if they would do anything. On Outer Worlds they had Unreal, had to learn new way of doing things, but you had access to the engine and could change things you needed to. But you were responsible for that change, whenever you updated Unreal you had to fold your changes back into it.
TL;DR Technological change never ends in gaming, get used to it. Unreal will not be around forever. New engines will come out, they will get displaced or change so dramatically you have to relearn everything. Exciting to learn something new every time, stressful because you must learn something new every time. Embrace it and love it.

 

[Roguey]

Man, he’s gotta be *really* running out of ideas now

He's suffering from what's called Youtuber "content creator" syndrome.

Basically people make a channel and set out to put their thoughts and experiences out there, and hopefully be able to share some knowledge. It's noble enough. But then for some unknown reason once the deed is already done and all personal knowledge has been exhausted they become compelled to keep making more videos. Even if it means regurgitating shit they've already done, or twist it in some way. Or sometimes they just start making shit up to justify making more "content".

Cunts just don't know when enough is enough. The Simpsons was a funny enough show. But we didn't need fucking 36 seasons of it.

He's answering questions people ask him. You can always just stop watching.

 

[Stella Brando]

...how many decades can you keep scavenging cans of food?

In the games, about 20.

 

[Infinitron]

I talk about the three programming languages I used to make all of my video games: C, C++, and C#. Overall, C++ is the best and most useful, but C# is the easiest to learn.

 

[StrongBelwas]

Started with BASIC on the Atari 800, leading into assembly. Pascal was first compiled language, write code, run it through compiler, get executable. BASIC was interpreted, relatively easy compared to compiled language, but Pascal was a lot faster and had more features.
Had to learn a lot of languages for his degree, first exposed to C and C++ academically. Learned C academically in last year of UVA, but had an overview class in '84, year before that in 1983 was when he started working on Grand Slam Bridge, written in C. As doing a lot of work in C both academically and professionally, sort of became his go to language and picked up the tricks. Recursion, making structures with pointers, arrays, etc.
From Grand Slam Bridge all the way up to Bloodlines, Cain used C. Very good grounding in how to use it for games. However, the Source Engine for Bloodlines was in C++, so they switched.
Has been asked why he didn't change from C to C++ sooner, including a potential publisher very directly and very rudely saying Cain was basically an idiot for not making the change while they were shopping around Arcanum. Cain looked at C++ and C, usually the same compiler could do both, and compared their output. He found that optimization throughout the 90s worked better on C, and the executables were smaller. That combination led him to work with C. Bloodlines comes along, they don't have a choice, but Cain is pleasantly surprised to find optimization had much improved in C++. But Cain would still find issues. Even in the 2000s, Cain would compile his code with optimization flags and it wouldn't work, then he would switch off the optimization flags and it would work fine. He'd look at the Assembly and find it was rolled out a loop incorrectly or put in an incorrect jump and then fix that. Shocked Cain to the core in the 2010s to learn there were professional programmers who did know not you could do that. If they ran into a problem, they would switch off all of the optimization and then switch them on one by one until they broke and then avoid using whatever one caused the problem, or they'd use pragmas to switch off optimization in certain parts of the code, but they never tried to figure out what exactly was causing the problem.
Started using C++ a lot from Bloodlines onwards, except for Unity games. C++, to Cain, felt like C but with classes. Yes it has more features, but in Cain's experience other programmers want to use those features when they aren't needed and they lead to overly complex hard to read code. They'd dent it's a problem, but six months later there would be a bug, they'd be looking at the programmer's code asking what they were thinking, and the programmer can't remember what they were doing because the code was so obtuse. In Cain's experience, inheritance and operator overloading are the prime causes of code being hard to go back to and read. What happens is you end up looking at code that you think is calling a function but it's actually overloading or inheriting something else. These are great techniques, but in the wrong hands they are abused.
As an example of where two excellent programmers would argue about C++, at Carbine two programmers Cain thought very highly of were arguing about whether or not they should use the standard library. One said they were wasting time and reinventing the wheel, they should just use STL. Another programmer said they should not use STL as many of them did not come with source code, making them hard to debug. Both of them made an excellent case and Cain found it hard to decide. Thinks what they eventually did was find a very plain language STL with source code and told people to use that.
Source was the first engine Cain ever used that wasn't one made in house but they went back to in-house for Wildstar at Carbine. Game was coded in C++, Cain experimented in C# for the tools. Faster to develop for, easier to do UI. However, it's optimization was terrible. Wrote the same programs in C#, C, and C++. The C# was an order of magnitude slower than the other two, the other two would do something in a second, C# would take 30 seconds to a minute to do the same thing. C# of an image processing program he made was extremely unimpressive in it's speed.
Going to Obsidian at 2011, Onyx was all in C++, but Cain still had to learn it because it was made at first for Dungeon Siege III and they were repurposing it for South Park. Figured it out, but the next project was Pillars and Cain realized he'd really have to use C#.
For Pillars, The Unity engine is either java or C#, engine was opaque so you couldn't really code it, but you get calls out. When your class starts or there is a tick, you can call out. If you needed to apply a status effect you can call a status effect event. Turns out, you had to do things like that, because they had optimization issues where people tried to do everything on the update. It's easy to wait until the tick update and then check if you have any status effects. To know that, you'd have to keep a list of old status effects or put a flag on some status effects. People were doing stuff like that in Unity's update function instead of what they should have been doing, listening for an event and have the class that has the required thing make an event. C# was not only slow, it was being heavily abused to become even slower. Used for Pillars and Tyranny, but jumped back to C++ for Outer Worlds.
C++ and Unreal were not only faster, but you could go into the source to fix issues.
If Cain had to pick a language today, he would go C++ with C as a close second. C++ is very powerful, compilers and optimization are very good now. Easy to abuse it, but if you are careful you can write really good code. Would pick C second because of his familiarity with it, compilers for it were very good but he wonders if that levelled out. C is very close to the metal, if you have to write out memory locations, C is great and let's you do that. Compiled code is 1:1 to assembly so you can see how it would be done if you wrote it by hand.
After that, C# would be his third one with a caveat. By far the easiest to learn, but may have been because he learned C and C++ first. But he also saw students and other programmers taking to C# very easily.
If you were to start now, grab C#, grab Unity, learn how to code that way, but you should eventually move onto C++.

 

[NecroLord]

Learn to code.

 

[Axioms]

Tim Cain is based! C++ forever. Unity into the garbage heap of history. Who knew?

 

[__scribbles__]

I talk about my thoughts on designing games that do not end when you finish the main story line as well as games with new game plus features.

 

[StrongBelwas]

Obviously, Cain prefers the choice of ending the game because all of his games have done that. Part of the reason is that when you have an ending, that is when you can do the reactivity that you can't do in game. Some reactions have immediate reaction (shop keeper doesn't like you, companions leave the group), but end game slides are more distant. Towns become major political players, villages fade away as people leave them, etc. Telling the player some of their consequences aren't immediate.
Also, having a game end encourages replayability. Witnessing your choice and consequences and outcome of your build is fun and makes you want to play again.
Occasionally leans towards the idea of playing a game where you can keep playing after the main quest, but there are two big caveats. Have to bear in mind that if you have content past the main questline ending, and you want to show reactivity for that, that is going to be money. Whoever that is, publisher, kickstarter, someone else, you should get that agreed on with them. It's going to affect everything you do in production/bug fixing/polishing/ and beyond. Second caveat, in pre production, you need to have planned for the player continuing afterwards from the very start. Going back to Cain's Setting------>Story---->Mechanics, you are basically saying you want the story to end outside of some side quests, which will affect your mechanics. If you don't plan to have people continuing your game and then later in development someone decides to ditch the level cap and let people keep playing, the player is going to travel to places that have no reaction to them defeating the main villain. Level design probably has to be changed, now that the villain is gone maybe some areas may need to be altered.
System Mechanics need to be changed or people will keep buying levels and gaining perks until they run out of things to buy. Walking gods that have every ability at max, going to get ridiculous. Some people suggest diminishing returns, say after 100 in a skill you can keep increasing it, but in the example of Archery it gives 1% more hit chance every five points instead of 1% every 1 point and eventually that goes up and up. At some point it will top out within a reasonable few thousand hours, maybe it goes to about 200% hit chance and getting above that isn't practical. Maybe once someone gets to 200% to hit chance, it goes to the critical hit chance ahead. Just has to be accounted for it at the start.
Also questioned about New Game Plus, feels like it is in the same area of design. Caveat he has never done that, but finds the idea interesting. If you don't reset everything on the player, you can also not resetting all of the world. Maybe change the location of some items, or characteristics of the villain. Different ways to handle partial return of skills, maybe the player can just reallocate it, or just restore their XP. Usually achievements aren't reset and he's seen achievements for doing New Game Plus.
Sometimes he sees features locked behind new game plus, maybe you can't play a particular class or race that is locked behind it. Doesn't really like the player not having the choice at the beginning, but if you really like new game plus as an element of your game he can see doing this. Could also see using New Game Plus to call out to the player about skills or features you put in that weren't strictly important. Wonders if he should have kept the crazy encounters in Fallout like UFOs and Time Lords to some form of new game plus, otherwise people wonder if they are canon. Considers something like New Vegas's Wild Wasteland trait.

 

[StrongBelwas]

I talk about chocolate and my daily chocolate eating meeting at work...and what this has to do with game development.I mention my blog Chocolate I Have Known: https://chocolateihaveknown.wordpress...

Do not underestimate how much Cain is into chocolate.
Links his chocolate blog, thousands of entries. At one point Cain could eat a bite of chocolate and and tell you within 5% what the kcal percentage was and could sometimes tell you the country of origin if it was single country.
When he was in grad school one of his roommates said he should keep the label of a particularly delicious chocolate. Filled six albums before he went all digital, hence there being there thousands of entries on the blog.
Once Cain hit his 40s, he couldn't eat all that chocolate, plus his doctor told him to cut back on it. Still wanted to try new chocolate, so at Carbine he started the chocolate eating meeting. 15 minute meeting at 3 o clock every day. He'd send an email out, he'd break up a chocolate bar and have it on his desk showing the label. Don't leave any meetings your in, but if you are free feel free to come on over. Some people came every day, some people came every now and then. Some people switched to dark chocolate from white chocolate, some got really good at judging chocolate. It was a nice mental break in the middle of the day, and Cain had noticed that back at Interplay that there were people who seemed to know each other despite not sitting at lunch together, not working on the same projects, or sit near each other. Cain found out this was from smoke breaks, a lot of connections being made there. At Cain's chocolate meetings, the same thing would happen. People would talk about new movies, what games they were playing, what places in the area they should check out. People realized there were people from every department coming in;' programmers, designers, sound guys, QA, etc. They'd ask how new tools were working, if a feature went in, recommend a cool song to the audio team, complain about getting stuck in geometry to the level designers. Probably one of the most productive 15 minutes of the day, someone always seemed to get something out of it. Another programmer would let someone know a feature they were planning to spend the day on was already implemented in another module, probably saving a great deal of work.
Cain would meet coworkers at chocolate store, a new store would open up and they would plan an afternoon around it. Take pictures, take pictures inside, go to lunch afterwards and talk about the chocolate they bought.
Took a trip to Paris in 2006, and went to La Maison Du Chocolat, a shop he always wanted to see. During his 2016 Sydney conference, worked in an entire afternoon to go to five different chocolate shops. Took pictures at all of them, employees might have thought he was a little weird but he was so excited to go to a store that he ate chocolate from half a decade ago.
People at work would come back from vacation excited to bring Cain chocolate. They would check Cain's blog and if it was new they would buy it and bring it to Cain's next chocolate meeting. Probably extended the chocolate meeting for years. When new people joined Cain's team he would invite them to attend the chocolate meetings. Some of them were pretty young, they'd think the director inviting them to their office was a little weird, but they'd see everyone hanging out, maybe up to 12 people, and they'd learn a lot.
Cain reads about onboarding of new employees and sometimes it feels very sterile. They buy them donuts or take them to lunch, nice, but feels forced and fabricated. At the chocolate meetings, a new employee would get comfortable, ask a question(i.e not understanding Perforce) and probably save a lot of time that would have been spent fumbling.
Was short, but a good mental break and tended to stimulate the team. They'd go back to work on some feature that wasn't checked in yet.
By the time he got to Obsidian he knew the meetings would have this side effect. Enjoying chocolate was still important but it was very useful. People who went through normal onboarding practices would never open up, but a few minutes in the chocolate meetings would start chatting about their family and background.

 

[Cologno]

Probably not, but wouldn't be hard to figure out for anyone with a modicum real life street sense. Pretending otherwise you're just playing or dumb.

 

[NecroLord]

is chocolate code for cruising?

Tim doing the down low...

 

[gaussgunner]

is chocolate code for cruising?

Or eating da poo poo?

 

[NecroLord]

Does Tim like his chocolate brown?

 

[S.torch]

Looked up information on radiation for Fallout, reread Lord of the Rings for Arcanum.

People who said Arcanum is not High Fantasy & wasn't influenced by Tolkien SEETHING forever.

 

[Infinitron]

 

[__scribbles__]

I talk more about damage numbers, this time about how to select a good numerical range for those numbers.

 

[StrongBelwas]

Picking your range of damage numbers means setting subsequent ranges for health/bonus stats/ weapon damage.
A lot of what he said about skill and attribute ranges are valid here so a lot of it will sound familiar.
Big numbers make people happy, they like seeing 20/200/2000, not 2. If you think it doesn't really matter between doing 2 damage to a 10 HP monster vs doing 2000 damage to a 10,0000 HP monster, you're thinking with your brain not your heart. A lot of people don't think about it. You can dismiss this, but perception is important. This is why a lot of old arcade games have the basic points for shooting an alien start at 100 so your final score inevitably has at least two 0s at the end of it. Why not just make each alien 1 point, and end with 211 points instead of 21100? People like the big numbers. The goal of the game is to make people have fun.
Random numbers seem to confuse and anger people, what they really want is 'fair' or 'uniform'/well distributed. If Damage is considered the same way, they don't want proportional, they want a feeling of power.
Big numbers have the advantage where if you want to do anything as a percentage (Such as a weapon that increases damage by 5%), if your damage is 1-10, 5% doesn't mean much. Even if you hit 10, the additional 5% is 10.5, probably not what you want.
The way Cain does damage value is he picks a range of what he wants to do, such as say 1-20, because he wants some percentage values in there. Than you just spread out the design from there, if a sword does 1-20 then he can have a club do 1-12 and a dagger do 1-5. He can have damage bonuses of +1-+3. If he wants Strength to also give bonuses, then he has to consider if you could get half of the total range just from weapon and strength bonuses. Now it's more like 11-30 instead of 1-20. You may be fine with that, you may not like all that impact coming not from the weapon, the character could pick up a dagger and still do a lot of damage.
Maybe you decide to swap to 1-100 after that, but now you are dealing with an enormous range. Maybe more randomness than you want because someone could do 3 and someone else could do 97. Perhaps instead, you do d10x10. No one will ever do less than 10, and on average you will be doing 55. Percent bonuses work on these larger numbers. So multiples of ranges like that might work.
Sticking to that idea, and spreading out the design, how much HP should the enemies have? How long do you want combat to last? If you want relatively low monsters to go down in 3 hits, than in the 1-100 range, with 55 average, probably give them 100/110 HP. Definitely won't go down in 1 hit, 2 hits probably will get them, 3 hits should, maybe 4 or 5 hits at the most. What about bosses? If a regular creature goes down in 3 hits with around 100 health, should he have 1000 health? Should he have armor, with damage threshold or damage reduction?
The converse to this is how much damage is the creature doing to the player. You may want those in parity, but there are more monsters than there is the player, so they'll need more health. Maybe some monsters do less damage, but then what if you give some monsters magical spells or weapons, or other enemies that buff the monsters?
Pick one damage range, and calculate everything. Put numbers in for all of those and see if you like the output. If you don't, go back to the first thing you did, maybe damage range, and tweak that, then see how you like the new outputs. Maybe it's good everywhere except the damage on bosses and player, than you could tweak how armor changes that. Lots of designers Cain knows use spreadsheets for just this reason.
You won't create a perfect design in 10 minutes, you'll work on how all of your choices affect future choices and if you don't like the effect, change the choice.
Maybe you do all of this, like the numbers, then you put it in, and you start getting feedback like the bosses are too easy. Do you give the boss more armor, add damage threshold just to bosses, increase the health? No one right solution, very much try and see.
Keep in mind what Cain said about percentages. If you don't want percentages or big numbers, that's fine, but expect a bad reaction from many players.
Tl;DR : Pick what you like, see where it leads in your design. If something you don't like is a result, go back and start tweaking. Try it and see, that's what all the designers do.

 

[Roguey]

I agree that single-digit damage numbers are disappointing, but I prefer the two to three digit range, not so much thousands and thousands of damage.

Also prefer small ranges. Too much chaos and luck involved with extreme swinginess.

 

[deuxhero]

The real problem with single digit numbers is that they do not handle division or even subtraction well. If an ability removes a third of ~10 damage, a value that rounds up is much more powerful than one that doesn't (3 to 4 is over 133% effective) while if a value rounds down (always or closer) that will be noticeably less powerful which makes weird breakpoint stuff. If you introduce percentile values as real things, you might as well have gone with higher numbers in the first place.

I actually like how Bannerlord made the default human HP 100 and made bonus HP relatively rare. That made it so you instantly know how effective an attack is: ~33 HP is about three hits to kill a human, while 55 is a 2HKO. Triple digits also discourages +2% damage to targets wearing pink when on fire bullshit that always pops up once you get to that kind of value (and the last digits aren't cosmetic).

Of course, the whole discussion is predicated on HP with damage subtracting directly is actually the right choice for game design. A lot tabletop RPGs have moved to wound systems where damage over a certain value gives a character a wound, each wound gives a character penalties, and a certain number of wounds takes a character out of the action. It works better for certain kinds of combat (superheroes or pulp where attacks are either shrugged off or devastating and random minions absolutely aren't playing by the same rules as the player characters) than others (a prolonged boxing match where damage is cumulative and all combatants have equal plot armor). Savage Worlds does this and began as a western, showing it actually works fairly well for guns: A .38 Special or .380 always has a 34.38% chance of downing a human minion of average Vigor on a gut shot, 66% chance two to the gut, and one to the head has an 81.25% chance, which is far superior to the minion losing arbitrary amount of HP lost from an arbitrary HP total on every attack.