Game Design Part 2 – Balance

Pyromania Main Menu

In my second piece on game design (you can read the first part of my game design series here) I’ll be using Pyromania as an example. Pyromania is a mod for Descent 3 which brings back the classic Pyro-GX and changes gameplay in several ways.

RSMax’ Pyro-GX model, version 1, in a mockup of Descent (1) level 1 briefing screen.

The mod was created simply because Descent 3 was too dull to play. Zero-gravity fighter ships, instead of behaving like powerful, fast spaceships, reminded of speed-limited cars in city centre traffic. That might had been the inspiration for Descent 3 design by the way.

To understand Descent, its spaceships and the fully-3D concept of the game series, let’s go to the beginning: Descent (1), which was released in 1995, was a fully-3D deformed-cube game, which was rather different to fashionable 3D shooter games of that time, Doom and Heretic/Hexen and later Duke Nukem 3D. Neither of those “groundpounder” shooters were really 3D: they were “2.5D”. There were no 3D models and no fully-3D maps. Enemies, weapons (and pretty much any object in the game) were drawn with sprites. In Doom, there are only eight angles from which a given monster can be viewed (each corresponding to a compass axis). Each view is supplied with a sprite, four facing straight, and four for diagonal views. Also in Doom-engined games, maps are not 3D – one cannot, as an example, go under a bridge or have a second floor in a building (forget third and fourth – all there is is the ground at varying height).

Doom II. Notice how enemy corpse sprites are always turned towards the player.

Descent broke these limitations, having both sprites (for projectiles/explosions mostly) and real 3D models (for missiles and robots and player ships). Descent was a major thrill when it came out.

Descent robot firing on the player.

It also had a deformed-cube engine which did allow stuff like shoot-through corridors, bridges and multiple stories in a building. Best of all, nobody was limited to riding on the ground – all robots and human player ships moved in any direction, up and down and upside down backwards. Just like in a flight simulator, except Descent ships can stop and hover in mid-space, unlike planes, which require forward motion to generate wing lift thrust.

Descent automap was fully 3-D too. The green dots represent hostages, humans abducted by robots infected by an alien virus. A bit of a twist on the old “alien abduction” gag.

Descent was clearly derived off Star Wars (the Pyro-GX looks a lot like a deformed X-Wing fighter). The movies in turn derived their space/air combat from WW-II-era dogfights between piston-engined, propeller-driven planes.

Needless to say, it’s a bit weird for a future spacecraft to be decidedly low-tech. Still, there it is: no all-aspect radar, no beyond-visual-range missiles, no inertial navigation systems or anything much except blasting robots in twisted mines.

Descent 3 HUD. That is all there is. Chaff comes as an extra countermeasure later on.

Anyway, the first snag with Descent 3 was its movement physics. Ships were too slow. Tricord once made a calculation of speed relative to average Descent cube and average Descent 3 segment and relative to robot speeds. The analysis was that Descent 3 Pyro-GL was effectively about half as slow in real speed compared to the original Descent Pyro-GX. Needless to say, dogfighting wasn’t too exciting – slow ships gave the impression of running against a pillow, even on afterburner.

Eagle’s Pyro-GX model, quite detailed. This was the final Pyro-GX model for the game.

The second trouble in Descent 3 was balance. Ah yes, balance, that favourite word of mediocre game designers (and sign of mediocre game design). You see, in the real world, there’s no such thing as “balance”. A jet fighter plane can rip another jet fighter plane to shreds in two seconds of gunfire or with an accurate missile shot. There are factors which influence the outcome of combat, such as supercruise (a plane’s ability to go supersonic without afterburner), launch altitude and speed matter for BVR (beyond-visual-range) radar-guided missile launches (the faster and higher, the more kinetic energy and manoeuvre freedom a missile will have), radar signature (how visible a plane is to a radar, larger allows missile lockon at a longer distance) and radar/avionics sophistication and power (a better, smarter phased array radar sees “stealthy” planes such as F-22 80 KM away).

Boeing X-32 cockpit mockup. Plenty of instruments here.

Thrust:weight ratio, thrust vectoring, control system, aerodynamics and wing loading determine how agile a fighter is in dogfight. But in the end there might be a lucky chance (say, a gust of wind helping a plane dodge a missile or deflecting a missile by the tiny degree necessary for a fighter to evade) and skill which can make a relative bathtub win a fight. Of course an inferior combat vehicle will drag one down, but in the real world things aren’t ever “balanced”. Current generations of weapons and fighter planes are so lethal, the outcome of combat isn’t as much the result of a pilot’s superior skill, even, as the mistake of another (such as trying to evade a missile too early or bleeding too much energy in a turn). Accurate missiles are another issue – if you think current Western missiles are any better than Russian makes, you’re wrong (and the same applies to radar technology, which is promising to kill any advantages “stealthy” F-22s and F-35s have by 2020). Current generation of Russian Su-27 derivatives, as another example, in spite of being bulky and heavy, makes for some very agile fighters with 3-D vectored-thrust engines, stuff Western plane designers didn’t even dream of.

Souhoi T-50 taking off. Hit the image for an F-22/F-35/T-50 combat evaluation article.

Even in WWII, the introduction of Me-163 Komet rocket and Me-262 Schwalbe jet fighters was anything but “balanced” – the Me-262 outperformed any Allied prop fighter by a huge margin while the English were still experimenting with jet engines on a plane. The Me-262 carried heavier 20-mm and 37-mm autocannons (and rockets against bombers), whereas the P-51s and Spitfires had .5″ machineguns. Heavier cannons had less ammunition and could fire less rounds/seconds than multiple .5″ machineguns, thus favouring accurate aim over spraying. Neither plane helped the Luftwaffe much (too few, too late), but that’s a different story – technically they were far superior to anything the Allies had.

Me-262 replica at an airshow.

All of this is anathema to your balance-obsessed game designer type. To him, the best game is the dullest game where every weapon looks different, sounds different, fires different, but acts more or less the same on any side. “Balancing” is responsible for crazy stuff like slow rockets in Quake II and even slower non-inertial (unaffected by running speed) rockets in Quake III. Even Quad Damage in Quake III isn’t quad by default (it’s only triple). Descent 3 suffers from that sort of dullness too: ships are too slow, weapons are too slow and weak in spite of impressive graphics, robots are too daft and slow. Robot weapons are also weak. So what you have is a game that, surprise, is too slow and dullish to play. Real air combat is anything but that. Whereas in WW II planes were slow, had low thrust and bled airspeed too much, starting with 1950s things had changed very quickly, pushing maximum speeds beyond the speed of sound, missiles were developed, and plane-mounted radars and RWRs (radar warning receivers) had changed air combat too. Best of all, jet fighters became quick and (eventually) agile, capable of gaining speed quickly, unlike old propeller planes that threatened to stall at any overly sharp manoeuvre or an overly steep climb. The trouble with Descent 3 though was, it was too much like WW II air combat – slow and ponderous, rather than sharp, quick and agile as jet fighter combat is.

Lockheed F-104 Starfighter, a Mach 2 1950s fighter which was infamously known as the “Flying Coffin” for its high fatal accident rate (53% of all F-104s were lost to accidents).

So in Pyromania all of this was countered by introducing more powerful, more realistic weapons, better robot AI, and bringing the old Pyro-GX back with its original armament (including the Spreadfire and Helix cannons). In a way, rather than starting with a preset cliche of ships and weapons, gameplay was designed from scratch by combining original Descent Pyro-GX gameplay with more furious combat and the gameplay of jet fighter combat. A lot of stuff was still missing (AWACS, ECM, radars, BVR missiles, FLIR, etc.), but the general concept was fast and furious combat in the style of jet fighter dogfighting, with additional twists like individual armament custom-tailored to a ship. As an example, the Blackstar ship carried up to 18 rockets as its basic missiles, all could be released quickly in a volley (the same as real-world rocket pods), whereas the Pyro-GX had up to 20 classic Descent concussion missiles, each fired with a slight delay, slower, but doing more splash damage.

Blackstar looking mean.

Small, fast and agile (and fragile), Griffin would carry homing/hypervelocity two-stage SLAM missiles against heavy AGMs and smart mines of other ships, and so on. Rather than trying to bother much with balance, each ship had a selection of lethal weapons which could be used in particular tactics favouring a certain style, and each ship also had its weak points which allowed others to down it quickly, given the skill.

Griffin exiting the Centroid mine in a classic Descent-style explosion sequence. The map is a Descent II map port made by its original author.

Gameplay became more refined and more thrilling, as evasion and combat tactics became more varied and a pilot could become a virtuoso of his own tactics, and he was also given more speed, more energy and more agility with which to evade others.

Therefore, the thrill was in increasing complexity of choices, variety of tactics, and rewarding plain adrenaline-generating combat skill, where a split-second reaction mistake might mean death. It might sound extreme, but really the mod itself was fairly easy to learn to play – yet skill and smart tactics and plain essential dogfight intuition were rewarded.

The photon frag torpedo was one of the many new fun weapons, this one also being somewhat suicidal if fired too close. Or even if it exploded not too close by.

Compare this with plain Descent 3 where everything tended to boil down to hitting others before afterburning energy ran out and hiding around obstacles. Even weapons were weaker and “safer”. “Safer” weapons? Hello? Pyromania had suicidal weapons, even. In singleplayer, Descent 3 really played more like a lazy chore than a fun 0-G ship combat game. Pyromania introduced a few tweaks for robots (some becoming more powerful, all becoming more accurate and getting full-fledged weapons) and overall just made things more like a real war. Which, ironically, is what combat gameplay is about. Not “balancing” and listening too much to players complaining that anything’s too fast, too deadly, too disorienting. That’s a fast way of spoiling a game, sucking up to the lowest possible denominator.

Another improvement was the Pyromania Anarchy ship chooser – stock Descent 3 Anarchy (deathmatch) mode has no such option, a ship has to be picked in the pilot setup before joining a multiplayer game.

A game has to train skills and its thrill is in rewarding development of a player’s abilities; making it too easy and requiring no skills is sucking up to a casual, bored, unskilled and untalented fellow who’ll toss the game out after a while anyway. But the trouble is, of course, mediocrity is a shortcut to disappointing the players who would appreciate and play the game if it were challenging and developing/rewarding enough for them to improve their skills and grasp a game’s tactics/mechanics.

So really too much “balancing” and “sucking up to dimwits” is a sure-fire way to spoil a game for anyone. And for a company to crash into bankruptcy eventually. Or worse, persisting. In a fashion, the best way to design a game’s gameplay is by looking at real-world gameplay – like the DID designer said about F-22 ADF in an interview, real-world air combat and war have so much gameplay, why water it down?

Digital Image Design’s F-22 ADF. So realistic, engines had to be turned on individually.

Crafting a game world that appeals by how believable a thrill its world is, how realistic its physics are, and how engaging most of all its gameplay (which, remember, must teach – games are all about developing one’s own abilities, mind, and skills) is, ought to make most of a game’s long-lasting love affair with a player. Randomness and chance also play a certain role here, but more on that in a different post…so check back next week, don’t forget to read the first part in the series, and let me know your thoughts in the comments below!