Thanks to the recent patch 4.0, the Starcraft community has not been this active for a long time. As expected, there are many discussion threads regarding balance on forums, and it is always a joy to watch the zealots of different perspectives calling the oppositions out. Based on these conversations, it occurs to me that temporal considerations are not on top of everyone’s mind, and I am going to discuss it in this post.
Temporal considerations and causality
What do I mean by temporal considerations? I am referring to the effects of how certain change could have exponential effect in the later stage of the game. The effects are often unexpected, or at least underestimated, and more importantly not welcomed.
A classic example is Queen attack range buff way back in Wings of Liberty. Queen’s anti-ground weapon attack range increased from 3 to 5 in patch 1.4.3. This was the one change that led to the rise of the infamous Broodlord Infestor era. No change was done to Broodlord and Infestor themselves, but the improved Queen allowed Zerg to get to the ideal composition much easier than before. Because Queen is accessible at the early stage of the game, its effect starts to influence the game early on, and that shapes the subsequent interactions. Queens’ effects are even multiplied when players increase the number of Queens. I believe no one had imagined how detrimental the change would be when it was first announced, and I doubt Blizzard would have done it if they had foreseen that.
The interesting thing is, with hindsight, we would point our fingers the Queen buff as the culprit of the Broodlord and Infestor. Hence, many would have suggested the Queen attack range should at least decrease to 4 as a solution to the problem. However, imagine Queen’s range had always been 5, and we were discussing how we could solve this Broodlord Infestor problem. We would have make suggestions more specifically toward Broodlord and Infestor themselves. This suggests we often underestimate the causality of early game interventions on the subsequent interactions.
Implications on balance discussion
What are the implications of understanding temporal considerations on balance discussion? The certainty of our claims about an identified problem should decrease according to how late the problem occurs. That is, we should have higher certainty in our argument about a problem in the early game than a problem in the late game. There are virtually infinite number of factors that could contribute to the problem identified in the late game, and this is well illustrated in the Broodlord Infestor and Queen issue I mentioned. In other words, players have lots of room to make changes in their plays through out the game that could result in different situations in the late game, and we are less certain whether it is a problem that could be resolved without the intervention of a patch. In contrast, when an identified problem happens early in the game, we have a clearer picture of what other options the players could have taken to resolve the situation. This is because, fundamentally, there are few factors that could have lead to the outcome, so we should have greater certainty about our argument regarding the issue.
However, according to my observations, many do not consider temporal factors in their balance discussion argument (it doesn’t have to be explicitly). The usual balance discussion dilemma of whether we should wait and see how players figure things out themselves could head to a much more constructive debate when we apply the reasoning of temporal considerations. If the identified problem happens in the late game, we should lean toward the “wait and see” approach. On the other hand, if the identified problem happens early in the game, we should lean toward the “fix this soon” approach.
I tweeted a poll yesterday to have an estimate of what the community think about the Widow Mine change (see above tweet), and someone was… well… See the tweet below yourself.
Clearly, Laser_SeQ is a strong believer of “wait and see”, which I have no problem with in itself. However, it is counter productive to the improvement of the game when we group every single identified problem together, and then we take an either-or side between “wait and see” and “fix it soon”. My reply tweet basically implies the importance of the topic I am discussing now.
Applying to current TvP situations
We get a clearer direction once we apply this reasoning to the current identified problems in TvP.
- Chronoboost and Oracle
- Shield battery
- Widow Mine
Chronoboost and Oracle
A proxy Oracle hits Terran base at 3:15 earliest before this patch, but it now can hit at 2:50 with the new Chronoboost change. The problem goes beyond just the difference in time per se, but it is the options Terran have. Terran could react accordingly and minimise the damage from Oracle before this patch. The Oracle’s timing was great in terms of balance, because it was not clear who was coming out ahead. Terran needed to react immediately and correctly, or else the damage could be game ending. The key was the opportunity to react to Protoss’ build, and this is lacking in the current situation. Terran could not prepare in time as the Oracle hits earlier, and this has serious snowball effect. Some have argued that Terran should adapt the builds to prepare for Oracle (i.e., wait and see approach), but realistically what could Terran have done differently?
Going back to my earlier this discussion, this problem happens very early in the game, we could easily identify what could have done differently to change the outcome. I dare to say that we have figured out this early stage of the game enough that we have high confidence in forming a fair conclusion of the situation. One could not suggest Terran to just assume Protoss would use this build and hard counter it, because this basically suggests Protoss have a clear advantage in opening options in the match up (i.e., imbalance). Starcraft is heavily designed around counters and options, and the current situation as it stands does not align well with this philosophy. Therefore, I believe this should fall under “fix it soon”. Then, it is a matter of how to fix it. All in all, I believe the proposal of increasing the build time of Oracle is the most reasonable fix based on current understanding.
I would love the current Stalkers if I play Protoss. The patch increases Stalkers’ damage and decreases their attack speed proportionally. It appears that they are more or less the same on paper (same dps), but Stalkers have become stronger in most situations now. Let say, the current Stalkers deal 2 damage per shot with 2 seconds attack cool down, and the old Stalkers deal 1 damage per shot with 1 second attack cool down. They both deal 4 damage in 4 seconds (let’s ignore advance math), but the new Stalkers deal 4 damage in 3 seconds while the old Stalkers deal 3 damage in 3 seconds. If a Stalker battles a unit of 4 hp, the new Stalkers would kill it in 1 second quicker. Clearly, the actual number in game is different, but the implications are huge. The obvious one is the enemy unit deals less damage since it dies quicker (1 second quicker in my simplified number example). The new Stalkers also prevent healing opportunity as the unit would have been sniped with the initial attack, and hence, the attack cooldown is an irrelevant trade-off in various situations. Their improved damage also allows Protoss to have a stronger map control, because Stalkers’ potential with snipe and kite is too high in the hands of good players. Beastyqt mentioned how the new Stalkers affect other aspect of the match up in the video below (timestamped). The only negative about the new Stalkers compared to the old ones is the overkill situation in later part of the game when you target fire specific units (e.g Liberator; Zurgery told me this potential weakness, thanks).
The buff in Stalkers’ sniping ability is the exact reason behind the buff anyway, because Blizzard find that Stalkers were overshadowed by Adept. I liked the justification behind this change when it was announced. Ironically, however, Stalkers have now sort of replaced Adept instead, so it seems fair to say the outcome does not work out as well as we hoped for. Then, does this fall on the “wait and see” or “fix it soon”? I lean toward “fix it soon” for Stalkers. Stalker after all is a basic unit, and the effect at the high level where players could maximise its potential with good micro could be game changing in a negative manner. This is one change that we should pay much attention the opinions of progamers. I like the fundamental idea, so I think a minor fix in the numbers would be ideal.
There is this unspoken “Shield Battery rule” when someone makes a suggestion about builds or harassment options specifically. Does it work against Shield Battery? I have to be self-critical about my idea on using Ghosts as a light harassment option in the early game, because it later became apparent to me that it is likely not going to pass the Shield Battery test. This rule basically implies how influential Shield Battery is.
As of now, my opinion is that it is difficult to evaluate Shield Battery, so I lean toward the “wait and see” for this. I can understand the frustration of how ineffective harassment is against Protoss, but there is much confound with other related factors. For instance, the new Stalkers are much better in anti-harassment, so the observed outcome of “bad harassment results” could not be fully attributed to the Shield Battery. Further, Widow Mine, which is the formerly best early game harassment option Terran have, has taken a huge hit in this patch, so Shield Battery should be of lower priority in balance consideration.
Widow Mine nerf is the most discussed Terran change. I had written about my opinion about Widow Mine before, so I would not elaborate too much here. In short, I believe the change is not justified according to Blizzard’s statement, so the change should be reverted (feel free to disagree).
In the early game, Widow Mine drop has lost most of its value. While its ability to kill workers is valuable, most of its value comes from its potential to kill workers. It is the Widow Mine’s potential to kill workers if it is not cleared that forces Protoss to react, and this adds strategic depth to the early game of TvP. Even without this nerf, a standard Widow Mine drop in the early game would have been weaker with the new Stalkers by default. Strictly judging from the early game interaction, I lean heavily toward “fix it soon”.
In the mid-late game, Stalkers have overtaken the main army role now, and Widow Mine is not a good unit against Stalkers. Of course, Widow Mine is still the best answer against Zealot, but I am not sure how much this change makes a difference. Protoss usually have Observers when they engage, so the Widow Mines are often one-time used anyway. This may indirectly promote builds with very delayed Robotic Facility, for example, Zealots and Archons. Strictly judging from the mid-late game interaction, I lean toward “wait and see”.
Off topic: I actually mess up the Twitter poll I showed above. The “no change was needed” and “the nerf is too much” are overlapped. If one believes no change was needed, then s/he should logically think the nerf is too much. Let’s ignore the potential sampling biases (e.g., self-selection bias) for this poll for a moment and look at the broad picture of what the community think. It appears that there are equal number of people believe the nerf is just right and the nerf is too much. But like I said, the “no change was needed” people would basically pick “the nerf is too much” if I did not provide the first option, so it appears that most people are leaning against this nerf. I am not going to conclude much from this, but it definitely shows how controversial this change is.