Quake III: Gold includes the base game along with the Quake III: Team Arena expansion. The original, closed-source versions of the game still work for online multiplayer. Many users may prefer to use more recent, improved engines based on the source code release. The resulting game is not only worthy of its lineage, but it may very well be the best Quake yet. While Quake III Arena's focus may be its multiplayer deathmatch component, it does have a single.
reader comments
with 64 posters participatingGoogle's AI subsidiary Deep Mind has built its reputation building systems that learn to play games by playing each other, starting with little more than the rules and what constitutes a win. That Darwinian approach of improvement through competition has allowed Deep Mind to tackle complex games like chess and Go, where there are vast numbers of potential moves to consider.
But at least for tabletop games like those, the potential moves are discrete and don't require real-time decision-making. It wasn't unreasonable to question whether the same approach would work for completely different classes of games. Such questions, however, seem to be answered by a report in today's issue of Science, where Deep Mind reveals the development of an AI system that has taught itself to play Quake III Arena and can consistently beat human opponents in capture-the-flag games.
Not a lot of rules
Chess' complexity is built from an apparently simple set of rules: an 8x8 grid of squares and pieces that can only move in very specific ways. Quake III Arena, to an extent, gets rid of the grid. In capture-the-flag mode, both sides start in a spawn area and have a flag to defend. You score points by capturing the opponent's flag. You can also gain tactical advantage by 'tagging' (read 'shooting') your opponents, which, after a delay, sends them back to their spawn.
Those simple rules lead to complex play because maps can be generated procedurally, and each player is reacting to what they can see in real time, limited by their field of view and the map's features. Different strategies—explore, defend your flag, capture theirs, shoot your opponents—all potentially provide advantages, and players can switch among them at any point in the game.
This complexity makes for a severe challenge for systems that are meant to teach themselves how to play. There's an enormous gap between what might be useful at a given moment, and the end-of-game score that the systems have to judge their performance against. How do you bridge that gap?
For their system, which they call FTW, the Deep Mind researchers built a two-level learning system. At the outer level, the system was focused on the end point of winning the game, and it learned overall strategies that helped reach that goal. You can think of it as creating sub-goals throughout the course of the game, directed in a way that maximizes the chances of an overall win. To improve performance of this outer optimization, the Deep Mind team took an evolutionary approach called population-based training. After each round of training, the worst-performing systems were killed off; their replacements were generated by introducing 'mutations' into the best performing ones.
Beneath that, there's a distinct layer that sets a 'policy' based on the outer layer's decisions. So if the outer layer has determined that defending the flag is the best option at the moment, the inner layer will implement that strategy by checking the visual input for opponents while keeping close to the flag. For this, the researchers chose a standard neural network trained through reinforcement learning.
Let the games begin
With the architecture in place, FTW was set to play itself on randomly generated maps in teams with one or more teammates. The goal was to get it to 'acquire policies that are robust to the variability of maps, number of players, and choice of teammates and opponents, a challenge that generalizes that of ad hoc teamwork.' The amount of effort required for this system to learn is pretty staggering; the researchers refer to going through 45,000 games as 'early in training.' Distinctive behaviors were still being put in place by 200,000 games in.
The researchers could track as FTW picked up game information. 'The internal representation of [FTW] was found to encode a wide variety of knowledge about the game situation,' they write. The agent first developed the concept of its own base, and it later figured out that there was an opposition base. Only once those ideas were in place did it figure out the value of picking up the flag. The value of killing your opponents came even later. Each of these had the chance to change future behavior: once FTW had figured out the location of the two teams' bases, most of its memory recalls focused on those areas of the map.
Some of these things ended up being remarkably specific. For example, in highly trained versions of the system, the neural network portion had an individual neuron dedicated to tracking whether a teammate had possession of the flag.
In the outer layer, many of the behaviors ended up recapitulating strategies used by human players. These include base defense and camping in the opponent's base. Other strategies, like following a teammate if they have the flag, were used for a while but later discarded.
With the training done, the researchers set a group of FTW players loose in a tourney with human opponents. By about 100,000 training matches, FTW could beat an average human player. By 200,000, it could beat a Quake expert, and its lead continued to expand from there. In the tournament, a team of two humans would typically capture 16 fewer flags per game than a team of FTW bots. The only time humans beat a pair of bots was when they were part of a human-bot team, and even then, they typically won only five percent of their matches.
What’s in a win?
That's not to say that FTW excelled in every aspect of the game. For example, humans' visual abilities made them better snipers. But at close range, FTW excelled in combat, in part because its reaction time was half that of a human's, and in part because its accuracy was 80 percent compared to the humans' 50 percent.
But FTW wasn't reliant on speed for its wins. The researchers artificially inflated its reaction time to be similar to that of a human and found that this only reduced the bots' edge, with teams of humans now able to win about 30 percent of the matches. That still left FTW with a significant edge, suggesting that there were some aspects of its overall strategy that gave it an edge.
While the computational resources needed to run bots through over 200,000 games of Quake are pretty massive, it's still impressive that FTW could start with the input of just a few pixels and no overall picture of the game before managing to figure out not only the rules, but also strategies that could consistently produce wins. While Quake III Arena is now two decades old, it does provide a high-pressure, multi-agent environment that represents a far more general problem.
For now, the Deep Mind team is still thinking about a few limitations of the FTW system. One of the problems was that the bot population tended to converge on a set of similar approaches, something that's only really effective if all agents in the environment are the same. In many situations (including many multiplayer games), the agents can be specialized, requiring solutions that remain more generalized. The genetic approach used in the outer layer of FTW's reward system also tends to focus very quickly on a limited subset of effective solutions.
Those concerns suggest that Deep Mind is looking at how to make FTW even more flexible than it already is.
Science, 2019. DOI: 10.1126/science.aau6249 (About DOIs).
Quake III Arena provides the ultimate deathmatch experience.
By Jeff Gerstmann | @jeffgerstmann on
When most people think of first-person shooters, id Software immediately springs to mind. The small Texas-based company has been involved in almost all of the first-person shooters that are considered classics. Quake spawned a rabid fan base on the Internet that still watches id's every move. So when id revealed that the next game in the series, Quake III Arena, would be specifically designed as a multiplayer game, fans weren't quite sure what to think. But id's purpose became increasingly clear when it released a succession of Quake III Arena technology demos for public scrutiny. The resulting game is not only worthy of its lineage, but it may very well be the best Quake yet.
While Quake III Arena's focus may be its multiplayer deathmatch component, it does have a single-player mode. When playing alone, you can go up against artificial intelligence-controlled bots. The bots do their best to act like human players, and on the higher difficulty settings, they put up an excellent fight. Each bot has different characteristics that govern the way it fights. The portly biker chick Lucy tends to duck a lot. Xaero, a Zen master and the final boss of the single-player mode, is also master of the railgun. The single-player mode is a lot like the kind in an arcade-fighting game, such as Mortal Kombat. You'll move through several different competitive tiers, each with different arenas and bots. At the end of each tier is a one-on-one showdown; these fights take place in smaller, tournament-style arenas. The bots are downright chatty - when you get a group of them together, they'll hold small conversations with each other, which are displayed in text onscreen. In team games, you can order bots around, or even let a bot take control of your team and tell you what to do.
The single-player mode is fun up to a point, but the multiplayer mode is where the real action is. Quake III Arena moves very quickly, and it has a real pick-up-and-play design to it. The game feels slightly simplified as compared to other recent shooters, but for the most part its simplicity is a good thing. The game has been stripped down to its first-person shooter essence, and any extraneous weapons or power-ups that might have gotten in the way of great deathmatch gameplay have been omitted. However, a few things stick out as having been overly simplified. For instance, as in most shooters, there is an auto-switch weapons option that lets you automatically switch to the new weapon when you pick it up. It's a convenient feature because it lets you rearm yourself with a stronger weapon almost immediately instead of having to press a key or spin the mouse wheel to select it. But when you're in the middle of a close-quarters shotgun fight, you really don't want to accidentally switch to the rocket launcher or railgun if you happen to walk over one. In that case, you can disable the auto-switch feature. Similarly, if you're running away from someone who's firing BFG blasts at you, the time it takes to manually select your new rocket launcher could be the difference between life and death. A weapon-priority scale that lets you decide specifically which weapons you'd like to switch to automatically, as in Unreal Tournament or QuakeWorld, would have easily solved the problem.
The level design and texture design throughout Quake III Arena are quite good, even if the levels feel a little generic. None of the levels ever feel like real-world places; they are one futuristic castle setting after another, occasionally broken up by a level composed of a bunch of floating platforms in the middle of a black void. That said, the textures, especially the animated ones, are really amazing, and the game engine allows for curved surfaces, which deliver a more plausible feel to the architecture. The levels have obviously been designed with gameplay in mind - most of them are great fun to play in. However, the game seems to focus on levels that are great for only six to ten players. With more than ten players, most of the maps seem really crowded. Aside from standard deathmatch, you can also play team deathmatch, a one-on-one tournament mode that acts like a scaled-down version of the popular rocket-arena mod that appeared in Quake and Quake II, and capture the flag. Unfortunately, capture the flag seems almost like an afterthought. There are only four CTF levels included in the game, and all of them seem suited to smaller teams.
Like the great-looking textures, the game's 3D models and special effects are very impressive. The character models and skins look terrific, and the animation really brings a lot of personality to the different characters. The various models are highly varied, and range from giant hopping-eye creatures to skeletons. Models based on the characters from Doom, Quake, and Quake II are also available. Curved surfaces help improve the quality of the level design, but other effects like fog and great colored lighting add even more atmosphere to some of the game's levels. The weapon models also look good for the most part, and they're easily identifiable in your enemies' hands. Unfortunately, the sound in Quake III Arena is inconsistent. All of the weapon fire is outstanding. The sound of rockets whizzing by your head as you dodge from side to side is especially impressive. The different character voices (each model has its own set of sounds) are pretty good, though you may become annoyed with Orbb the eyeball's screeching noises - all the more reason to kill him quickly. But the real problem with the audio is the announcer, who says things like 'five minutes remaining' or 'impressive' when applicable. Id obviously tried to duplicate the sound and style of the announcer's menacing voice from the Mortal Kombat games. But instead of hiring an appropriate voice actor, id simply took a typical voice (specifically that of level designer Christian Antkow) and pitched it down a few octaves to make it sound sinister. Instead, it just comes off as amateurish. Given the frequency with which the announcer speaks during the game, you'd think that the voice would have received a bit more attention.
Regardless of how it sounds, after playing Quake III Arena, it's easy to see that moving toward a multiplayer-centered game was an excellent idea. With cable modems and DSL connections slowly gaining mainstream acceptance, the bandwidth is there for a lot of people to finally enjoy a good deathmatch game at a high speed. Fortunately, provided you find a good server, the game also manages to be playable over a 56K modem connection.
All things considered, it isn't a stretch to call Quake III Arena an outstanding game. But whether it's superior to its rival Unreal Tournament is less certain. If you're interested in sophisticated team-play modes like those featured in Unreal Tournament, then Quake III Arena comes up short. However, Quake III Arena provides the ultimate deathmatch experience. If deathmatch is the particular style of play you're interested in, then Arena is your game.