Tuesday, June 11, 2013

Revisiting RollerCoaster Tycoon

On a recent trip to visit my girlfriend Steph, she excitedly told me that she'd procured a copy of RollerCoaster Tycoon. Never being one to shy away from 90s nostalgia, I settled in for an afternoon of theme-park building on a lovely Boston May day where the temperature didn't quite crack 55 degrees. Like every game fondly remembered from adolescence, you never know how well it might hold up as an adult; for every Chrono Trigger, there's a Diddy Kong Racing.

To be sure, some elements of RollerCoaster Tycoon haven't aged terribly well. Its graphics look halfway decent at first given that the game was a product of last century, until you remember that the likes of  Final Fantasy X came out just a couple of years later. The music, exactly like that at a real theme park, consists of four or five tracks on infinite loop. And, at least in the early stages, the game is trivially easy, but that might be because we--not the game--are older.

That said, RollerCoaster Tycoon had one nice surprise up its sleeve: it's a delightfully gamer-y game. I didn't notice it when I was twelve, and I didn't expect to see it now, but practically every decision in this game is highly mathematical. What's my return on investment on this massive roller coaster I'm plopping in the middle of a lake? What's the expected value of increased customer happiness from this elephant-costumed entertainer, and is does it outweigh the $40/month cost? How do you optimize revenue from park admission--does the parabola's maximum lie toward low cost and lots of people or high cost and few visitors?

One especially satisfying mechanic in RollerCoaster Tycoon that in retrospect does seem way ahead of their time is its persistent, dynamic actors. It's easy to write a mathematical model for a number of guests over time--it's akin to a common engineering problem of a reservoir filling and unfilling at different rates. Plenty of older games would have written a differential equation, solved it as a function of time, and asserted that you had that many guests in your park.

But tracking each guest individually by mood, money in the bank, preferences, and position in the park is a much tougher problem. (Essentially, it's like assigning every drop of water its own residence time, tracking its position and momentum in all three dimensions... and figuring out how happy it is on top of all of that.) RollerCoaster Tycoon solves the problem transparently and doesn't fudge numbers the same way that some of its successors do. The game is perfectly content to let the model run, but you can step in and manually control each of your customers--put him by a bench if he's about to throw up, or stick him in front of your overpriced hamburger shack if he's hungry. The result is a highly adaptable experience that's exactly as interactive as you want it to be.

In contrast to its transparent handling of customer dynamics, the scoring in RollerCoaster Tycoon is a bit murky. Number of guests in your park is easy enough to figure out--it's a function of the amount of stuff in your park, its admission price, and your advertising campaigns. So far, so good. Another number sometimes used to evaluate your park is "park value," which seems to have to do with cash in hand plus some depreciated value of your attractions, but has no apparent formula or useful predictor.

Stranger still is "park rating," which is likely also a function of the amount and quality of stuff in the park, plus upkeep, maintenance, guest happiness, and Chris Sawyer only knows what else. Steph and I managed to get our park to the maximum rating of 999 for a few months, but we noticed that the rating seems to fluctuate by plus or minus 20 points for no particular reason. Other events happen stochastically but without a reliable method of forecasting how often. For example, the underlying mechanics behind your rides breaking down doesn't always make a lot of sense, with the most tortuous roller coaster in our park malfunctioning at about the same rate as our haunted house.

None of that probably bothered me as a kid. After all, this is a game marketed to preteens who like to ride roller coasters (and/or build things--I was never much of a roller coaster person), not twenty-something engineers who like to delve into game design. Maybe that's why Steph and I found it so easy now: since we're now interested in how the game operates in addition to making gorgeous theme parks, figuring out optimal strategies becomes that much easier. (Our optimal strategy in a nutshell: build one of every ride, charge a nominal fee for ride admission, a ton of money for park admission, and outrageous prices for umbrellas while it's raining.)

Yet, most importantly, making gorgeous theme parks is still fun, fourteen years later, and in game design, that's the biggest goal of all.