Study of the coin-op games
The coin-op environment is very special and very different from the business model of the games developed by EA. Let me describe first the business model of coin-op:
Coin-op business model:
Developers like “Williams” develop a coin-op game. That part is done very much like any game developed for the PC or consoles, except for the fact that they control the hardware and in the past used to make very different implementations. Now a days the costs of developing large custom chips is driving the adoption of more generic hardware solutions, mostly putting several PC components together with most often a Motorola or AMD CPU. Then the product is put into “testing” which means that it is set up in one arcade and the developer evaluates how much money the product is making. If the product makes a good return (over $300 per week for a game that costs about $2500), then the product is finalized. This means fixing all the bugs and going into production. Those units are then evaluated independently by third parties and the income level of the machine is published, leading Coin-Op operators to buy them. That’s how the “developer” makes it money, from the profit of the direct sale of the machines. A successful product will be sold a few thousand units, and a margin of 30 to 40% on the unit sales will have to cover the development costs and generate the profit the “developer” needs.
About 50% of the time, the game does not bring the expected returns. The game is then either modified to bring the required level of returns, or most often killed.
When the product is successful and operated the coin-op operator either runs the game in an arcade or in a store where it rents the floor space for half the coin incomes.
This explains that everyone involved with the game needs to maximize the income of the game per week. The developer because it will determine how many units it will sell, the operator because he keeps 50 or 100% of the income.
Now from the point of view of the player, the initial investment is small, and to get the player to drop more coins, the game needs to be rewarding very quickly. This is why understanding the coin op games is interesting to the study of game design overall
My favorite pinball. I have owned it many years…
Different aspects of the coin-op game:
Attract mode:
While the game is not played, it needs to advertise itself to the potential players. In general this has been a challenge because of limited content available for that part of the game. There is a cost to ROM and HD storage, so the attract mode has to re-use as much as possible of the actual game content. Another aspect of the attract mode is that it needs to make you want to play, and at the same time not be something that will delay the player from dropping the first coin. If the attract mode is too much fun and too varied, the player might not feel the need to interact.
The first experience:
As soon as the first coin is dropped, it is a race for the game to do what it needs to do to convince the user to drop the second coin. The investment of the player into the game at this point is very small, typically less than $1.00, so the designer should not expect that the user will try to figure out complex menus or options. The player needs to play as soon as possible and have enough fun in two minutes to be spending its next $1.00.
This is where we could learn most from the coin-op environment, and it will be probably very useful in the Internet in the future. In most of our games, the player has invested $30.00 to $50.00 up front. So we know that he really wants to play the game, and will try to decipher our environment, but really we should try to make it as easy as a coin op game.
The re-playability:
Obviously, once a user has been playing the game, you don’t want to have him ever stop. So the game needs to be as re-playable as possible, and if there is a goal, then it needs to be more and more difficult to reach. High scores list have been developed for that purpose, to drive the knowledgeable users to play the game more (there can be only one…), those lists also reset every so many games, so there is really no way to be for ever at the top of a list.
Auto tunable free play:
Long term
On coin-op and pinball machines, there are often ways to get more “free playtime”. On pinball you get extra plays, on other games you most often get extra time, in the end it is the same. This free time is a strong reward to the players, but at the same time it is reducing the profitability of the machine. For that reason, the operators monitor it extremely carefully. Most machines have a mechanism to automatically tune the game, to fix the amount of free playtime. Every 100 games, the percentage of free playtime is calculated and if it is different from the targeted time (usually 10%) then some fundamental variable of the game is changed to make the game easier or tougher. This is done constantly and drives pretty quickly to the desired result. It is easy to see on the pinball machines where the free play score adjusts itself to the average strength of the players in a specific location.
Short term
Above I described how the game adjusts itself to reach the level of profitability expected. In this section I want to describe that in games with an obvious finished goal, the game auto tunes to make the user feel that it almost reaches that goal, and that if he had done just a bit better he would have won the free time.
I think it is most visible in the driving games where you get extra time when you make it under a banner: you always fail just one second short, it feels like you almost made it. This is a strong motivation to put more money.
On other games, there are always supporting mechanisms that are getting enabled to make you catch up if you are late, or slow you down if you are making too much progress. In a fighting game, the moves of the automated opponent will vary based on how well you are doing. Even if you are a beginner you will not get wiped out completely, but you will get your money worth of fighting. In driving games, you can always catch up when you are behind. Very quickly it seems you can win, but yet you loose 90% of the time.
Special case of the pinball machines:
The match feature for the score, one could think that you have one chance in 10, but in reality this is controllable by the operator.
The score that you need to reach to win an extra play is not printed on the pinball machines anymore, it changes everyday and adapt itself to the players.
How are the users using the game?
To guarantee a longer lifetime of the game machines to the coin op operator, the machines have many internal controls that can be adjusted by the operator. They are mostly difficulty levels, and major game variations to reflect the fact that the game might be used with $0.25 or $1.00 and therefore the amount of game play delivered might need to be dramatically modified.
We think that the internet will help us understand what the user do with our games, but the coin op developers have been tracking a lot of data about their players ever since they have put computers to run the games. They keep statistics on how the game is played, how long it is played, what the score curve looks like, how much the key features are used, how many people play together, how much money each player puts in on average. This is then used to figure out how to improve the income of the game, or as feedback to develop the next game.
The operator checks those statistics to find a possible problem in the game when the income drops sharply. Most often it is due to some mechanical failure and sometimes due to users who figured out a new trick.
I suggest we develop better tools during the development process to build more statistics about what the players actually do in the game, and use them to understand where we see a difference with what we expected.