Blizzard-made maps have been heavily criticized since the start of the WoL beta, and with good reason. The common element they all share is that they tend to be polarized in their features: they feature incredibly hard or easy to take expansions, and rush distances that are often too long or too short.
This has a profound effect on how the game is played at the top level. Every matchup imposes specific constraints on map design, so the most extreme maps also tend to be the most imbalanced. All competitive maps need to roughly follow these constraints to be considered fair and balanced.
The two most important to keep in mind -- and the ones that this article focus on -- are the constraints placed on map design by PvZ. This matchup requires an easily walled natural to allow a Protoss player to fast expand, as well as a somewhat closed off third base that can be defended with forcefields. While the second condition is somewhat less important with the introduction of HotS, it is still relevant enough in today's metagame and needs to be considered when designing a map meant for competitive StarCraft 2. There are many maps that have broken these constraints: Dual Sight, Crossfire, and early versions of Bel'shir Beach.
The latest map to defy these constraints was, of course, the first version of Daedalus. Its original version went against both these rules, with an extremely wide open third as well as an extremely vulnerable natural; so vulnerable, in fact, that FFE isn't viable and modern gateway-based builds need several adjustments to hold a Nexus. No natural expansion has been this hard to defend since the days of Crossifre and the early versions of Bel'Shir Beach. Neither of these problems are helped by the extremely short rush distance. Similiarly to Bel'Shir Beach, changes have recently been made to the map in order to make it less miserable for Protoss.
Daedalus version 1.0 was heavily criticized by both the community and pro players. Many spectators, commentators, and pros had asked for its immediate removal from the tournament and ladder pools.
This article will analyze the games played on the first version of this map, to figure out if the complaints were valid and the map really did require the ramp changes to be made. The sample size of games is extremely small, but the way the games were played shows interesting trends. Only GSL code A games were considered in this analysis, as these are supposedly the highest level games played on it. On Daedalus version 1.0 Zerg is 8-3 in PvZ.
Let's start with the early game. In Code A Protoss players either used outdated and overly safe builds that left them in terrible positions heading in the midgame, or extremely cheesy openings. This was because the standard Mothership Core expand appeared to be unplayable on this map and FFE was absolutely out of the question.
As a result, the most popular non-cheesy builds were 2010-style three gate expands. These builds died years ago because they are incredibly uneconomical, while being unable to put any kind of meaningful pressure on the Zerg. Other variations include one gate tech into an expand, which similarly delays the natural nexus for too long, while also being susceptible to all the early zergling speed builds that are so popular on this map. Speedling builds are a fairly minor investment on the Zerg's side when compared to the inefficient Protoss builds, which results in protoss players being farther behind than on the average modern map.
Simply looking at the games played confirms this scenario. Here is some brief analysis:
Seed vs Rogue, Seed vs Pet, Paralyze vs Life: All these games feature a Zerg player who opts for early game zergling or roach/ling pressure and ends up ahead, despite doing no damage, because the Protoss is forced into an overly defensive build and/or position. This huge economic advantage escalates much faster than in an average PvZ, and the Protoss easily gets overwhelmed.
Super vs Solar game 1, Myungsik vs Symbol: In these games, the Zerg doesn't pressure but the Protoss is still forced into overly safe builds that can't put on any pressure, tech quickly enough, or expand quickly enough to keep up with the Zerg.
sOs vs Shine: This game is the only one out of the entire sample in which the Protoss survives the huge deficit taken in the early game, and eventually sOs is able to win in the long run.
The extreme forms of cheese used by Protoss are below. These builds are easily scouted and stopped especially when they are so predictable. As a result PvZ one base builds haven't been viable since around early 2011; this option is not a factor in modern top level play:
Super vs Solar game 2, Panic vs Roro: The Protoss tries a proxy that gets immediately scouted. Everything else after it this is irrelevant.
Yonghwa vs Leenock: This game features several serious mistakes by Leenock. Despite having a ling burrowed at Yonghwa's natural, preventing him from expanding and forcing him into a one base all in, Leenock is still caught without both detection and units by a fairly slow DT rush: the first warp in occurs around seven minutes. This game is the biggest anomaly thus far; a player of Leenock's caliber will never lose in this way more than once.
Trust vs Symbol, Ruin vs Sleep: In both these games, the Protoss tries to block the zerg's ramp with a forge/gateway/pylon wall; this is usually fairly successful, but it's obvious that such builds can't be considered the standard by which to judge a map.
Analyzing these games also brings up interesting trends in the midgame.
Zergs generally opt to invest in Lair tech armies to utilize the wide spaces of the map to their advantage while Protoss players are tending towards unsafe, tricky, or outdated builds. Unsafe robo builds, old stargate-based all-ins, and fake thirds have all been used: standard macro styles are much more rare.
This results in the supposed Zerg advantage appearing stronger than it is: Protoss players are either attacking into roach/hydra with sub optimal compositions and timings -- like seven gate void ray all-ins as a followup to overly safe openings -- or trying tech heavy builds that get busted while taking a third by the same roach-hydra builds. In some games, the Zerg early game advantages escalate so quickly that the roach/hydra army can kill the Protoss on two bases, before he can even consider taking a third.
Protoss players played like they were forced into playing extremely risky styles, like Colossus builds that skip stargate tech in order to get as many ground units as possible. The result of these high-variance builds is an extreme vulnerability to the tech switches which are so prevalent in HotS ZvP.
The PvZ balance discussions about any map should center around three points:
Can the Protoss fast expand safely and reliably with a standard build?
Can he take a third safely and reliably with a standard build?
If the answer to the first and second questions are no, are the necessary adjustments reasonable? Is it fair to ask a race to reinvent every build used in a matchup, exclusively for a single map?
The third point is the most important and overlooked. It may, however, have been the saving grace for Daedalus Point. The few games played show that the Protoss hadn't been able to develop good enough builds to play out an even game. This might be due to a large number of factors that have nothing to do with the map: time to prepare (the map was a very recent addition to the pool), inefficient or inconclusive practice ("I didn't find a way to expand in practice, so I'm just going to proxy gates and hope for the best"). Again, the key issue of the map isn't exactly "are standard builds not viable?"; but rather, "does this map force such a change in standard builds and styles that the end result is unfair for one side?"
Developing an even slightly safer opening, for example, would have deep impact on the map's metagame: Zergs would have tried overly aggressive builds, but the Protoss players' adjustments may have been enough to hold those aggressive builds and give them the edge. In this scenario, the openness of the map has actually given the advantage to the Protoss side. We never saw a fully standard game on this version of the map, but it's very possible that such a build might exist. If it does, then the Zergs would stop using aggressive openings, causing the Protoss to play greedier, and so on. Eventually, the early game would settle around a smaller number of builds, and the variance in the games played would decrease to something more stable. Of course, it's also possible that the map was just imbalanced and the games on it would never have been stable, as was the case with Crossfire.
The same considerations are valid for the midgame builds. Protoss has essentially three to four stable midgame styles in HotS:
- Stargate/robo into either phoenix/colossus/blink stalker or colossus/void ray
- Stargate into chargelot/templar; and
- Three base blink builds (which are the least used among these).
When considering Daedalus, blink styles are particularly interesting and relevant, as they might be the ones that best fit the map. The idea behind them is that by having a huge amount of sentries and blink stalkers in the midgame, defending a third base against many kinds of roach/hydra/ling attacks is somewhat easier than with more tech based builds.
The mobility and power of blink stalkers also allows the Protoss to hit strong three base timings against any kind of zerg tech switch or mutalisk rush. On paper, this looks like the perfect style to play on a map like Daedalus; however, it's less popular than the others since it's not nearly as well mapped out (and arguably trickier and harder to pull off). If it is possible for Protoss to go blink every game on the old Daedalus -- and have a 50% win rate with it -- then the map can be considered balanced. Zerg styles then have to develop to counter this new Protoss build and so on. A good example of a map that played out and evolved like this is Entombed Valley: Protoss players tried to maximize their pre-hive timings so far and cut enough corners that low tech roach heavy builds suddenly became efficient once again. Note, however, that the map itself was always considered reasonably fair and balanced.
The sample size of games was extremely small, but it appears that the initial version of Daedalus Point might truly have been more unfair than is necessary for the ZvP matchup. In the process of developing maps and advancing the game, it's always important to remember that testing extreme maps should not be done in individual premier leagues like GSL and WCS. There have been crazier, more imbalanced maps before in StarCraft 2 -- such as Arkanoid or New Polaris Rhapsody -- but instead of being forced into the top individual league shortly before the start of a new season, they were played in team leagues and/or minor tournaments first. Team leagues like Proleague in particular allow the development of a wider sample size to test maps out, while giving players and teams more time to study the map and avoid an imbalanced matchup/map combination: we can cite Arkanoid here once again.
If Daedalus Point was in the Proleague map pool, there wouldn't have been complaints. There would be no ZvPs played on it until it was figured out to the point where it could be safe to send a Protoss out onto it. This did not happen, and players were asking for its immediate retirement before it could be fully understood. Instead of its retirement, Blizzard decided to change the ramp to be close to the standard, and Protoss now can (somewhat) easily wall their naturals.
Only time would have told if this map really was as imbalanced as it seemed, but one thing is certain: innovative and extreme maps need a long time to be fully understood and can help immensely in the development of the game across all matchups. The testing required for this development and understanding, however, is immense and forcing such a map into the pool of most premier tournaments so early on in its lifetime is a huge mistake. That said, removing it instead of changing it did remove some imbalances, but also limited the potential developments and tweaks to how the game is meant to be played that we haven't found out yet.
The TeamLiquid Strategy Team extends a special thanks to Nony, for engaging us in discussion and providing a strong counter point to the assumption that the original version of Daedelus was unequivocally terrible, and that it could lead to developments in the ZvP metagame. Without his sober second thought, much of this article may not have been written nor considered.