Dragon Quest (1986) for the Nintendo Famicom is 64 kb, and is said to take about 10 hours to play from start to finish (though personally it feels much longer). It was developed by a team of 15 people.
Pokemon Red/Blue (1996) for the GB is 373 kb and takes between 30 and 100 hours to finish (although you might spend much, much longer playing it). It was developed by a team of ~35 people (depending on how you count), including 5 people credited for map design, 6 for monster design, and 4 for game scenario. There is only one person credited with game design, but you can imagine that with such a small team the reality is probably more collaborative.
Pokemon is in every way a deeper game than Dragon Quest. Comparing the stats for both games, we can see that: the file size has increased by a factor of six; the team responsible is double the size; the playtime is reported to be between 3 and 10 times longer. If we take a step back and look at Pokemon compared to Dragon Quest holistically I think it is reasonable to say that Pokemon is a significantly richer game than Dragon Quest. I’m a big fan of both games, but Pokemon is a much deeper, more engrossing, more personal experience. The experience of playing Pokemon feels more expressive, more nuanced, and more full of interesting little mysteries, questions, and rumours.
Pokemon Gold/Silver (1999) had a similar team size, with a similar focus on monster design, map design, and scenario. The ROM weighs in at 710 kb and, in a marvel of technology, it essentially includes the entire previous game’s world. Once you become the champion of the Johto region, you unlock the ability to return to the Kanto region and go on a second gym tour. I cannot convey to you the impact that this had on me when I played the game the first time. The game is reported to take between 40 and 200 hours to clear. By virtue of the structure of Pokemon games, as well as how much code and content these games shared, I think it is reasonable to say that Gold/Silver is richer than Red/Blue by some factor. Gold/Silver includes all kinds of subtle improvements over Red/Blue: tweaks to inventory management; the battle system; new types; pokemon have genders (for better or for worse); there is breeding; the game has a day/night cycle, etc… The ROM only doubled in size, but the depth of gameplay feels like it grew by a factor of ten!
The steps from Dragon Quest to Pokemon Red/Blue, and then to Pokemon Gold/Silver each involved less than 10x increase in what I would like to to call “size”, but subjectively I would say they each resulted in a greater than 10x increase in what I have been calling richness.
Pokemon Sword/Shield (2019) weighs in at a whopping 12.4 gb (18,000 times bigger than Pokemon Gold/Silver). It is said to take between 30 and 100 hours to finish. The game was developed in 2 years by a team of roughly 600 people. This team includes more than 100 members of the 3D graphics section, and the credits include thirteen “3D Graphics Section Cooperative Companies”. That is to say, we have no way of knowing exactly how many people were working on 3D graphics. Suffice it to say that at least 1/6th of the entire team was graphics. To the game’s credit there is also a significant “World Planning Section” that includes around 70 people, with 16 on monster design. I personally enjoyed playing Sword/Shield, and I thought that the setting in particular was quite charming and well thought out.
You can probably see where I’m going with this: I think that if a ROM is 18,000 times bigger, or a team is 20 times bigger, or a budget is however many times bigger, then we should expect to see a corresponding increase in… something. In particular though in my case I would like to focus on the ROM size rather than the team size, or budget. I personally believe that if a game is some number of times bigger, then it should also be some number of times richer, deeper, or gooder.
I realize that this is a naive premise: that the size of a videogame should map to the richness of that game. It’s silly, I understand that. Imagine if I made a similar claim about the weight of a work of art.
Michelangelo’s “David” (1504) weighs roughly 5,660 kg while “Mona Lisa”, produced in the same year, weighs only 8 kg. Am I claiming that we should expect “David” to be 3 orders of magnitude “better” or “richer” than “Mona Lisa”? That’s absurd too. It’s reductionist thinking to the extreme. The impact of a work of art has roughly nothing to do with it’s size, weight, or carbon footprint.
At the same time, it isn’t wrong to say that Mona Lisa may be (ironically) a denser work. That is, if we think about the amount information packed into the Mona Lisa per gram, compared to David… Hmm.
But that is comparing apples to oranges. Imagine instead a series of large naked old testament man statues, each larger and more ornate than the last. Is there not a point of diminishing returns? At some point I think it becomes reasonable to say “no, I will not make a bigger statue”.
—
So, I admit that this is a silly premise. However, I think that underneath the layers of silliness there is a kernel of solid criticism here. Without defining exactly what “good” means, and acknowledging that this is not a nuanced lens through which to think about videogames or anything else in general: I think a lot of people might agree that Pokemon Gold/Silver was at least “twice as good” as Pokemon Red/Blue. To me it’s a marvel that they managed to fit so much love, such an interesting world, and so many curious interacting systems, into such a relatively tiny ROM. On the other hand, I don’t think anyone could really say that Pokemon Sword/Shield was 18,000 times “better” than Gold/Silver.
I would not like to claim that the relationship I desire between the size of a game and it’s richness should be linear. In other words I would not expect a game 18,000 times bigger to be 18,000 richer. In fact, there might not need to be any increase in the so called “richness” of the output at all: but we should expect some kind of positive outcome. Because otherwise we’re just accelerating the heat death of the universe, burning down the forests, and releasing our precious entropy into space… for nothing! If I’m doing categorically the same work as you, but it costs me millions of times more resources, then there should be some expectation that my results will be better along some metric.
For example, GB Studio games are much bigger than GB games developed in C or assembly. I think the minimum ROM size is 512 kb, which is bonkers. If I ported my puzzle game Ark to GB Studio the resulting ROM would be 16 times bigger. That would be bad. The thing that GB Studio buys us with that increase in size is democracy: a lot more people from a lot more varied backgrounds can make games using low code tools and engines like Bitsy, Twine, GB Studio, or even big engines like Unity and Unreal. That’s worth it! That’s a good trade-off.
So what’s the trade-off that justifies Sword/Shield being 18,000 times bigger than Gold/Silver?
Well there is one enormous, shiny, glaring metric along which many might be tempted to talk about Sword/Shield being some number of times “better” than Red/Blue… which is to say graphics. Someone might argue that the game looks “orders of magnitude better”. But that argument would essentially boil down to the claim that 3D graphics look better than 2D graphics. That’s a tough sell. The way something looks obviously depends more on things like cultural context and the quality of the art direction, and should be basically independent of the medium, right? At least, over time that should be the case: there might be a culture that for some number of decades values oil paintings higher than acrylic, but that’s a transient state. The idea that 3D graphics are “better” than 2D graphics by any factor, is absurd. It is like saying that sculpture is just better than painting.
I reject the idea that Sword/Shield looks better than Gold/Silver. Or no wait, maybe I don’t: it was a rather lovely game at times, and Gold/Silver honestly looks kind of bland at times. So instead I’ll say that I reject the idea that Sword/Shield categorically looks better than Gold/Silver, or that this improvement accounts for the 18,000 times increase in size. It looks nice, but I think that fact is independent of the spend. It could have looked nice and without needing to be 12.4 GB.
—
OK, but what if the argument was not that 3D graphics “look better” but instead that they are “more immersive”. The addition of specular lighting, depth of field, grass physics, and robust facial expressions make the game more realistic and thus more immersive. I think that the this kind of thinking has taken hold in our society and is really at the root of my whole beef.
Consider this clod of dirt:
I’ve mentioned elsewhere that I think videogames have a “resolution problem” that the real world doesn’t have. Imagine that the clod of dirt pictured above is freshly torn from the rich loamy forest floor. It is teaming with microbial life; probably contains a worm; or a pill bug; might include tunnels carved out by ants; almost certainly is laced with roots and mycelium engaged in a the chemical equivalent of a trade war. It is a master work of level design. You could study the contents of a clod of dirt for your entire life; you would never run out of things to learn and discover. Perhaps the demiurge would disagree but, from the perspective of a humble human, no: for our purposes reality has infinite resolution.
The real world does not have a resolution problem, and literally any human could spend a decade exploring literally any patch of dirt and grass on the surface of the earth. The real world is infinitely deep…
myself in Brevity is the Darksouls of Witt
This clod of dirt might just be more interesting than any videogame that ever has or ever will exist. Just like a potato chip (10 calories) contains more energy than a AAA battery (8 calories), only it is locked up in fats and difficult to access, this clod of dirt offers an inexhaustible universe of play… for those with the mechanism to unlock it.
I am not exaggerating when I say: it is quite possible that we exist in the richest, densest part of the entire universe. We live in a bustling soup of information in an otherwise mostly empty void.
Why then do we waste so much energy trying to simulate things when we already live in the most interesting part of the universe? Why are we so impressed when our videogames look “realistic”? The inescapable reality is that we will never be able to accurately simulate anything. Our videogames do not and will never look “realistic”. All it takes is a moment of close inspection to undo any multi-million dollar attempt to make a videogame look “realistic”. Just walk closer to anything than the designer intended you to, and it all falls apart immediately. Eyeballs become glassy and hollow; hair is made of thick tubes bouncing and crashing against one another; the textures of rocks and grass repeat endlessly; the texture of wood is a photo plastered across a flat plain; your arms clip through the walls. Billions of dollars are spent every year on making videogames more realistic, but because of the resolution problem all we can ever achieve is a superficial, hollow realism that falls apart when prodded. So stop doing it!
—
I like videogames! I like Super Hexagon, and Pokemon Crystal. I like Minecraft, and Starcraft, and maybe even Warcraft. I like Starseed Pilgrim, and The Beginner’s Guide.
I wish I could like Horizon Zero Dawn (2017). That game really felt like it had a novel setting, but it didn’t really click for me. I’ve heard that it cost $45 million dollars to make, and was 70 GB. When I installed it the minimum requirement was 100 GB. The team that built that game was more than 2,000 people and who knows how many uncredited contractors. It takes between 20 and 60 hours to finish. The recent sequel, Horizon Forbidden West (2022), cost $220 million dollars to make, wants 150 GB (of SSD no less), and offers 30 to 90 hours of play time.
There is a kind of obligatory spend mentality going on here. I feel like people are making what they think they should make, rather than really deciding what’s right. They are enacting a ritual to summon the sales figures. A friend of mine pointed out that if you make a game with motion-capture faces for dialog and such… you basically have to make the hair move realistically. Imagine the dissonance of having motion capture faces if the rest of your game looked like Half-Life. Similarly, there are games where customizing your character’s look is an important part of the experience. You can’t be punk if you can’t dress the way you want, and you can’t be cyberpunk if you’re not punk. In such a game if you give yourself a ponytail and a trench coat, they damn well better flap in the wind when you ride your motorbike down the rain soaked streets of New Neo Tokyo.
This line of reasoning, to me feels like it’s own antithesis. If motion capture is going to be indirectly responsible for bloating your game’s budget to $220 million dollars then maybe don’t do it. What game could possible need motion capture faces?
I’m not saying it’s bad to focus on graphics, either. I’m saying its bad not to focus. I’m saying it’s bad to try to simulate reality in a videogame for its own sake. Your goals should be the goals of your game. If you are making a game where you climb, “Shadow of the Colossus” style, over human faces then OK maybe you need motion capture and hair physics to get that feeling really tight. Imagine Aloy’s braided lock crashing towards you at break neck speeds as you climb the enormous seems of her vest! That’s gameplay! But if your game is just about talking to NPCs, and rolling, I feel like we’ve been doing that just fine without motion capture for like 50 years.
What about “Baldur’s Gate 3” though… is the argument that the dialog is so poorly written that it can’t stand on it’s own without mo-cap faces? I haven’t played the game, but I’ve heard it has good writing. Do these same people think “book” would be better with motion capture talking-heads? The comment I’ve heard is that the motion capture faces improved the overall experience. Fair enough! But it is then our responsibility to ask the question “how much did this improve the experience, and was it worth the cost?”. If we don’t ask these kinds of questions we are doing ourselves a disservice. I think in almost any case if you imagine the budget your favourite AAA game going towards more of that game, rather than hair physics, you will be imagining a better game. A better use of those resources.
As an example of a well-managed relatively recent project, consider the budgets of botw and totk.
BOTW released the same year as Horizon Zero Dawn was 14 gb, was developed by a team of roughly 1000 humans, and takes between 50 and 200 hours to beat. TOTK (2023) was 18 gb, was developed by a team of about 1500 humans, and takes between 60 and 250 hours to beat. That’s not even a 1.5x increase in the metric I care about, ROM size, and yet totk could easily be describe as 3x bigger in terms of just the explorable surface of the game. Looking at the two game more holistically, side-by-side, in terms of richness… love it or hate it, it is not a stretch to call totk 10x richer, or deeper than botw. I played both and prefered totk’s moment to moment gameplay but think botw had the better overall structure, so I’m not even that deeply biased one way or the other.
In addition, comparing botw, totk, and Horizon Zero Dawn isn’t even apples to oranges. They are so similar in so many ways. How did Nintendo do it? How does Nintendo keep doing this?
They build inward, complexifying, deepening, enriching, rather than building outward, embiggening, expanding. They are efficient and focused.
I’m not saying we should forbid the “big budget flashy hyper-realistic horse dick-physics” game development style. What saying is that I want to see a AAA studio make a $220 million dollar MUD. Or a $220 million dollar Dwarf Fortress. Or a $220 million dollar successor to Pokemon Gold/Silver. I want 100 GB of Gold and Silver!