Design May Care: Postmortem For The Ten-Second Car

Posted by (twitter: @nintendoeats)
August 27th, 2013 7:58 am

AKA: balance your game!

title screen

DMC:QFTTSC is a game about stealing parts from cars. It is also about many things besides, but I want to focus on that because it’s the part that is kind of borked. Everything else (the writing, the sound, the programming) came out more or less awesome sans a few minor errors which were quickly patched up.

The crux of it is this: I had 4 hours at the end of development in which I could EITHER balance the game or do the sound. I chose to do the sound, which paid big dividends, but I wound up with a game that is absurdly easy.


Why is it easy?

It is apparently possible to finish the game without adding a new engine or nitrous. These are how you give the car more power, so plainly that is an issue. I also gave the player too much time to get parts in most cases, so the tension which makes up a lot of the gameplay is frequently absent.

Well that’s dumb! What balancing did you do?

As I was designing the various available parts, I focused on using real world numbers. This meant looking up how much a typical exhaust weighs, or engines of various capacities, and then using those numbers as a guide. This gave me a realistic spread (to some degree) which is what I wanted.  Where it all broke down was in the testing area.

Test Track

The testing algorithm calculates the car’s distance and speed in real time, but keeps it out of frame until the very end so that I didn’t have to spend time making it look realistic. Because of this, I could use a relatively simple formula to calculate the car’s speed at any given time. I balanced this formula by taking the stock car and testing it until it took about 40 seconds to do the standing quarter mile. I then put in a much more powerful engine and tweaked until that took about 12 seconds. I figured that if the player got a couple more improvements for their car and one good enigne, they would be set. I tweaked the time to get an engine accordingly, and job done.

That’s not so unreasonable, but I gather it didn’t work

You gather correctly. The problem is that I adjusted the formula to make weight more important, and reduced the base weight of the car to compensate. This means that swapping out the brakes or exhaust for lighter ones makes a huge difference. Remember, the car with stock parts was still about 40 seconds! Since the weight reduction was factored in over time, a tiny reduction would continuously affect the car over the whole lap. The end result is that James May’s Phat Pand can do the standing quarter miel in 10 seconds…with a 90 horsepower engine. If I had spent my last few hours testing various configurations I would have written a more complex algorithm and weight would have been a more reasonably balanced factor. But I didn’t.

What’s the lesson here?

Don’t just test numerical extremes, and always test tweaks to ALL of your factors. I don’t know if I made the wrong call (the sound may be more valuable than a balanced game), but I definitely regret not spending even 15 minutes more on balance. Next LD I will be including balance time in my schedule.

Oh yeah, when making an LD game you should always have a schedule. Always. Have. A. Schedule.



Leave a Reply

You must be logged in to post a comment.

[cache: storing page]