Legacy of the Void entered Beta just over a week ago, bringing with it many changes to StarCraft II, including core design changes to the economy. In StarCraft, everything flows from the economy: all strategies, all builds, and all approaches to the game, and any change to the economy will have extensive effects. Therefore, they must be examined closely. The economy in StarCraft II should be challenged, and examined critically – not to debase the work done by Blizzard, but to understand and examine the effects it has on strategic diversity in SCII. I believe that taking the economy changes at face value does a disservice to the Beta process.
This article focuses on how mining works in SCII. It begins by examining mineral mining as it exists in Wings of Liberty and Heart of the Swarm. It then examines mineral mining as it currently exists in the LotV Beta. It will conclude with the introduction of an alternative mining model which seeks to achieve similar goals in LotV with a different approach. The purpose of this article is to create discussion and encourage critical thought. Everyone who loves StarCraft II wants it to be the best game it can be, and we can’t do that without questioning everything. Through questioning comes understanding.
The implications of the 12 worker start will not be examined too closely in this article. While this change is significant, it deserves a separate discussion. Additionally, the alternative economic model and the core concepts therein to be presented in this article are incompatible with a 12 worker start. The proposed alternative economic model at the end will still address the design decision to speed up the early game without requiring such a large worker start or put an end to the ultra-early game strategies currently seen in HotS and previously in WoL such as early pools or proxies.
Intelligence in StarCraft II
One of the hallmarks of SCII is the vastly improved Artificial Intelligence in comparison to its predecessor, StarCraft: Brood War. The Improved AI does a lot for StarCraft II and brings it into the modern game space. Long gone are the days of poor Dragoon AI and units being unable to move up ramps or through choke points. Instead, armies move efficiently, creating new interactions, especially with regards to splash damage and positioning in fights.
Worker AI is also greatly improved in relation to BW. In BW, workers would rarely pair on a mineral node and instead would bounce around from one mineral node to another looking for an open spot to collect minerals.
By comparison, SCII has greatly improved worker AI, and there are additional consequences associated with it. In particular, the AI allows workers to pair on mineral lines. Each mineral node allows only one worker to be mining at a time; however the AI is intelligent enough that it checks to see if the currently mining worker is almost done with their harvesting cycle. If the new worker knows that the other worker is about to be done mining it will wait at the mineral patch for its turn to mine. These AI actions can be summarized as the “check, wait, harvest” cycle. Due to the time of a harvest cycle, the speed at which workers work, and the time it takes to travel between the nexus and a mineral node, two workers are able to synchronize their mining patterns on a single mineral patch, resulting in worker pairing.
Understanding Mining Efficiency
The result of worker pairing is that in most cases two workers will mine from a patch simultaneously (with some far patches allowing for three workers to "pair"). In this scenario, all workers from first to mine on a mineral patch through to the 16th will mine harmoniously and consistently. This harmonious mining is why it is generally accepted that players should maintain at most 16 workers on a mineral line whenever a another mineral line has fewer than 16 workers in SCII.
16 Paired Workers in SCII
With the existing AI for SCII, mineral income for each individual worker is the same from the first to the 16 worker. This number is averaged out to be 42 minerals per minute for each worker when mining either alone or paired with a partner on a single mineral node. When a third worker is introduced, the “check, wait, harvest” cycle a worker goes through when approaching a mineral line is interrupted. Specifically, the interruption occurs between the “check” and “wait” portion of the cycle. With the third worker, any worker returning to mine the mineral node will see that the third worker currently harvesting will not be done within the next one second (of a 2.762 second harvest time) and seek out another mineral node to harvest (within a range of 10).
The third worker changes the efficiency of mining. This does not mean the third worker does not mine. The third worker on each patch will always seek out an empty mineral node as per the "check, wait, harvest" cycle. Due to harvest times and travel time between the nexus and mineral line, one of the eight patches will always be available to mine up until there are 24 workers on a single mineral line.
As a result of the introduction of a third worker, the mineral income (per minute) of a base begins to scale down. Until the 16th worker, the mineral income per minute of a base is linear at 42 minerals a minute per worker. Starting with the 17th worker, this linear progression begins to stop rising as sharply and eventually plateaus as shown below. Interestingly, workers number 17 and 18 are 99% as efficient as the workers before them. But for the purpose of this discussion, we will consider anything less than 100% efficiency as a “drop in efficiency”.
One base mining curve in SCII
The Impact of the Current Mineral Efficiency Model in SCII
Now that the concept of mineral mining efficiency has been explained, we can move on to understanding its impact on SCII. A regularly proclaimed concern in the community regarding SCII is the existence of passive, three-base play or other slow-to-expand turtle strategies. Mining efficiency has a large impact on the optimal three base income.
Workers in SCII only begin to mine less efficiently (relative to their counterparts on any single mineral line) when a 17th worker mines alongside with the original 16. The result of this is the conventional game play decision to keep only 16 workers on a mineral line whenever another mineral line (with less than 16 workers) is available.
16 workers, regardless of number of bases mine the same amount in the current economy in SCII
The first gameplay scenario we will examine is a "one base against two base" scenario in which both players have 16 workers available to them. The only advantage gained by the player with an expansion is the ability to create more workers and eventually mine more minerals. While both players are at the same worker count, they both have the same mineral income per minute (since each worker is equally efficient). Splitting the workers between the bases makes no difference; only having more workers (for example, 8 workers on the additional mineral line available) will allow the two base player to have a higher mineral income than the one base player. The following image illustrates this point:
In this example we can see how only after the 17th worker does splitting workers between bases begin to matter
We can examine the “three base cap” using efficiency. Let us consider a player (Bob) with three bases and 75 workers (leaving 125 supply for army). Bob is mining from all six gas geysers with a total of 18 workers. This leaves Bob with 57 workers to mine minerals on three bases. Bob splits these last 57 workers on all bases, putting 19 workers on each mineral line to work. Each of Bob’s bases brings in 745 minerals per minute, resulting in a total income of 2,235 minerals a minute.
Bob’s opponent Chris is on four bases with the same number of workers. On the first three bases, Chris has all six gas geysers mining and 16 workers on each mineral line. This leaves him 9 workers to allocate as he sees fit on the fourth base. Chris decides to take the gas geysers and put the last 3 workers on minerals. Chris’s mineral income is 2,142 minerals per minute. In this scenario, while his gas income is higher, Chris is actually mining fewer minerals than Bob.
A chart showing Bob and Chris's relative income
If Chris chose not to take the gas geysers and put the 9 probes on minerals his income would be 2,394 minerals a minute. This is only a 7% increase in mineral income at the detriment of not taking two additional gas geysers (215 gas a minute from two geysers). Fully saturating the fourth base means an additional 25% income for Chris compared to Bob. However, this requires an additional 13 workers (resulting in a total of 88). These extra 13 workers represent 13 possible army supply that is used to gain an economic mineral advantage.
Therefore, there is no inherent advantage to expanding beyond three bases when two players have a similar worker supply. To gain an advantage, the player with a fourth base needs to make far more workers than his or her opponent and sacrifice army supply in order to do so. In this sense, there is some veracity in the claim that there is a three base mining cap in SCII when both players have a similar number of workers in game. The reward for expanding is fairly small while the risk is much larger. In scenarios where a player does not have map control, there is zero incentive to expand beyond three bases if they do not intend to make over 80 workers in a game.
The mining cap is therefore set by interactions at the worker level, which will be discussed in greater detail further into the article.
Breaking the Cap: Introducing Inefficiencies into the Mining Curve
Earlier in the article, we described how in BW harvesters tend to bounce between mineral nodes when there are more workers present than there are mineral nodes in any given mineral line (with few exceptions). In SCII, this would translate to 9 or more workers in present in a mineral line. Worker bouncing is the antithesis to worker pairing described earlier in this analysis. In the absence of worker pairing, the efficiency cap of 16 workers in SCII begins to drop off at the 9th worker instead of the 17th.
As we have already established, due to the high efficiency of workers in StarCraft II, there is no tangible benefit to expanding beyond three mining bases in order to obtain a mineral advantage over your opponent with the same worker count. Any base beyond three mining bases exists only to provide additional gas to the player with four bases. To leave less than 19 workers on each mineral line on the first three bases only serves to lose you money if you do not intend to take the gas on the fourth base. The reason for this three base cap has been identified as the high efficiency of the first 16 workers.
Taking into consideration the above analysis, to break this three base cap we must introduce inefficiencies into the mining curve. The mining curve is a representation of the growth of an economy in SCII in relation to the number of workers a player has. In the current economic model, this is mostly linear, as an increase in workers will result in a predictable increase in mineral income so long as there is one available mineral node for about two workers. As discovered in our analysis of the mining cap, when two players have a similar number of workers, the income curve plateaus for both players regardless of base count.
Any disruption to the 16 workers per base linear income curve will break the three base cap in the current economic model. There are two ways to break the cap in HotS. The first way to break the cap is by disrupting the number of minerals available to either player during the course of a game. This is the current approach for LotV. The second way to break the cap is to introduce inefficiencies at the worker level. This is done by removing worker pairing thereby making each worker after the eighth on a mineral line return fewer minerals per trip on average. This introduces inefficiencies in the mining curve much earlier than simply disrupting the number of minerals available and has other implications which we will examine closely soon.
Inefficiencies through Mineral Disruption
The first way to break the mining cap as described above is to disrupt the mining curve by limiting the amount of minerals available to players during the game. This is the approach Blizzard chose to pursue in LotV. The first attempt made by Blizzard was to limit the amount of minerals available to a player at any given time (for example 1200 minerals on 8 nodes). The intention here was to force players to take an earlier fourth base due to the fact that the main base mined out fairly quickly compared to the current economic model seen in HotS.
However, the issue with this first attempt was that it served only to shorten the amount of time a player can spend on the mining curve plateau due to the smaller total amount of minerals. The actual mining cap we discovered earlier remains due to the linear progression of the mining curve and its associated income plateau. The approach was viewed negatively by the community following Blizzcon for exactly this reason. While it shortens the time spent on the plateau with three bases, the core mining curve and associated plateau was otherwise not disrupted.
A LotV Base Mining At Full Efficiency
The second attempt to disrupt the mining curve is the current mineral model in LotV. The mining curve is disrupted earlier on than the previous attempt by cutting half the mineral nodes in any given base by half. In LotV, there are four mineral nodes with 1500 minerals, and four with 750 mineral patches. I will refer to this as the Half Patch approach. The actual mining efficiency of each worker and worker pairing is not impacted by this change, and up until the half patches are fully mined out, the mining curve in LotV matches that of HotS. Just under seven minutes after taking any given base in LotV, the half patches will be fully mined out. Mining out of the half patches creates a timer which introduces inefficiencies into the mining curve and prevents the current HotS plateau.
Examining the Half-Patch Disruption Approach
Unfortunately, there are two flaws with this approach which should be noted here. The first flaw is that the mining rate and efficiency of workers in LotV does not change compared to HotS. The second is the timer that half patches put on players in LotV.
By not changing the mining rate and efficiency of workers in SCII, the theoretical plateau and mining cap remains, though it has shifted in a manner significantly different from the “three base cap”. Recall that we previously identified that the mining curve of SCII, due to worker pairing, caps out (plateaus) at three mining bases if we assume that both players have the same or similar worker count mining minerals. This cap exists because there are just about two workers for every available mineral node: 48-57 workers for 24 mineral nodes. This particular interaction between workers and minerals does not change in LotV. The only change is that to have access to 24 mineral nodes a player needs at a minimum four bases instead of three after the seven minute mark in LotV while having between 49 and 57 workers ready to mine minerals. Having access to 24 mineral nodes is therefore disrupted earlier than in the current HotS model, but does not actually change the cap.
In LotV, players are on a much quicker clock than they are in HotS to gain access to additional mineral nodes in order to maintain their current mining curve, and being unable to secure a third base in a timely manner disrupts the mineral curve before the three base cap is even a concern. In terms of gameplay, players lose a certain element of strategic diversity. While the denial of a third base has always been an option, prior to LotV Beta, the only way for a player to disrupt their opponent’s mining curve was to harass workers or deny an additional base for a very long time. The amount of time required to disturb the mining curve of your opponent in LotV by denying an expansion is shortened by 50% in comparison to HotS.
A LotV Base with the Half Patches Mined Out
While both players may be mining out the half patches in their main, the aggressor is in a much more advantageous position than the defender when one considers the impact that map control has on the ability to secure additional mineral nodes. In short, the half patch approach does more than impact the three base cap. It places a timer on all players that is effectively half the length it was in HotS to maintain their mining curve. If a player cannot maintain their mining curve while on one base their income drops by 50%.
Another concern associated with the half patch approach is how it limits some of the strategic diversity currently in StarCraft due to its impact on the mining curve. Players are not able to choose to play defensively for an extended period of time while teching prior to obtaining an expansion or building a large and powerful army while slowly expanding. I think it is important to understand that defensive play should remain viable in StarCraft. While it might not be entertaining to watch players sitting on just one, two, or three bases and not moving out, the strategy should remain viable, the strategic option should remain open in the name of strategic diversity. More importantly, if possible, the defending player should be encouraged to try some form of counter harassment while defending; this alone would make "turtle" games more interesting.
In StarCraft there is always a defender and an aggressor, and these roles are often defined by decisions related to economic investment, tech, and map control. Some strategies focus on obtaining particular army compositions, upgrades, or other tech breakpoints in order to shift from the defensive role to the offensive role. Depending on the strategy chosen, defensive play can last a few minutes or most of the game.
The half patch approach limits the time a player can afford to be defensive while reaching a particular breakpoint. While the exact influence this has had on the game will not be fully understood until LotV becomes well developed in its current form, one thing is certain: limiting the number of available minerals and disrupting the mineral curve of defensive strategies as sharply as half patches limits some of the strategic diversity in StarCraft II. While half patches may introduce new strategies, it also removes others. The goal of LotV should be to increase strategic diversity as a whole by adding more options while removing less.
This brings into question: is disrupting the number of minerals available an effective way of disrupting the mining plateau in SCII?
The True Culprit: Being Critical of the “Three Base Cap”
We must be critical of what we like to call the “three base mining cap”. While it is true that we have proven a three base mining cap exists in the example of Chris and Bob, we need to examine this more closely.
The three base mining cap in the fictional scenario of Chris and Bob is more accurately called a 24 node cap. This is due to the fact that three bases have 24 nodes, and access to 32 nodes (four bases) does not change the way the 24 node cap operates if both players have between 48 and 57 workers intended for mineral mining to put on these nodes. As demonstrated earlier, if Chris chooses to make enough workers to gain the full income from his fourth base, he will see a mining curve and plateau that is higher than Bob’s.
In this example, we can clearly see that the mining curve and plateau only grow based on the number of workers available as a ratio to the number of mineral nodes also available.
In a worker paired economy, equal workers on a similar mineral count will result in the same level of income
So long as the number of workers mining minerals for either player approximates a 2:1 ratio of workers to available nodes and they both have a similar worker numbers, the cap in SCII cannot be broken. When we think of the curve and associated plateau as the result of the ratio between available workers and available mining nodes, it quickly becomes clear that the mining plateau in StarCraft II is the result of the worker AI and worker pairing.
Breaking the Cap: Introducing Inefficiencies at the Worker Level
To truly break the mining cap in SCII, we need to introduce inefficiencies in mining at the worker level by eliminating workers pairing on mineral lines. When you remove worker pairing, workers become less efficient beginning with the 9th worker, as opposed to the 17th, and a non-linear mining curve is introduced to the game, as income remains consistent until a base completely mines out.
The current worker pairing economy in HotS and LotV has an optimal harvester to mineral node ratio of 2:1, while an economy without worker pairing has an optimal harvester to mineral node ratio of 1:1. What does this mean in a real game? If both players have the same number of workers mining minerals, the player who is able to approximate a 1:1 ratio of harvester to mineral node will have a higher mineral income than his opponent. Just how much more of an income is a function of total minerals being mined, total number of workers and exactly how the workers are split. How does this impact the “three base” or “24 mineral node cap”?
More expansions make more money in an economic system without worker pairing
We previously established that a 2:1 ratio results in an optimal available mineral node count of 24 for macro games in which both players want to maximize both income (66-75 workers) and army (125-134 army supply). In the current HotS mineral model, this translates to three bases, and in LotV it translates to three to four bases depending on quickly a player can secure these expansions. The base count for optimal mineral income without worker pairing from a purely theoretical standpoint using the same numbers as in the above example doubles to six bases containing eight mineral nodes with 1500 minerals each.
Breaking the Worker Pair: A Simple Example
When we discussed how worker pairing works, we discovered the “check, wait, harvest” cycle. We determined that two workers have almost perfectly synchronized cycles due to worker AI and the travel time between mineral nodes and town halls for all three races. Without the ability to change harvester AI, we cannot manually adjust the “check” or “wait” parts of the cycle. Instead we examined various ways to adjust the harvest cycle through the map editor.
The first approach was to change the amount of time a worker spends at a mineral patch mining and how many minerals they collect while harvesting. While we tried a few different mods and extensions created by others in the past, we decided to try something simple first. We created an extension mod for HotS in which workers spent twice as long at a mineral patch (about 5.5 seconds as opposed to 2.7 seconds) and returned twice as many minerals (10 per trip as opposed to 5). I will refer to this as the Double Mining Model. In this model, bases use the mineral model of HotS, eight mineral patches with 1500 minerals each and 6 starting workers. The resulting mining curve for 16 workers is shown in the image below.
Mineral Income Curve for Double Mining
In this very simple model the first 8 workers mine at a rate of almost 60 minerals a minute compared to the 42 minerals a minute of the first 16 workers in HotS. The 9th worker only returns 57 minerals a minute and the 16th worker returns 48 minerals a minute. This encourages players to Maynarde their workers from the main to the natural as soon as possible. In a situation where two players have 16 workers, the player who keeps all 16 on one mineral line will have an income of 775 minerals a minute while the player who expands to two bases and splits his workers to have 8 on each mineral line will have an income of 953 minerals a minute. That is an increase of 20% for taking a natural base.
In the above example, we can see how players can be greatly rewarded for expanding as opposed to turtling. Each expansion will increase the income of the expanding player relative to their opponent even if they have the same number of workers. The 1:1 ratio effectively removes 24 mineral node cap and it becomes a 48 node cap, which is almost impossible to reach.
Keep in mind when you are thinking about the new 8 worker max efficiency model, having 16 workers on 8 patches will still provide more income than 8 workers. It just doesn't provide DOUBLE the income for double the workers as in HotS.
Examining the Double Mining Approach
Similar to how we closely examined the approach implemented by Blizzard, we need to examine the double mining approach as well. Since we are able to impact the mining curve through worker inefficiencies, we no longer need to change mineral patches in order to impact the mining curve. This means we can return to having 8 mineral patches with 1500 minerals each and 6 starting workers.
The increased income from the initial 6 workers speeds up the early game by a few supply (for example 8 pylon/depot/overlord, and earlier production facilities) without completely eliminating the extreme early game. Overlords and scouting workers have more time to reach the opponent’s base before the early build is fully developed compared to the current LotV economy. The earlier income also makes proxy builds slightly weaker (especially in mirror matchups) without effectively eliminating them as currently seems to be the case in LotV with a 12 worker start.
Perhaps the best outcome of the double mining approach is the fact that instead of punishing players for not expanding, it rewards players for expanding. This increases strategic diversity in the game of StarCraft overall. Players are no longer on a harsh, in-game timer to obtain expansions.
Expanding should be a strategic choice, not a requirement
As mentioned in the half-patch section, strategic options should not be eliminated but instead expanded upon. In a no worker pairing model, the choice to expand is influenced by the player’s chosen strategy and the actions of their opponent. Choosing to play an ultra-defensive mech style similar to HotS is once more an option. In addition, while we've managed to retain ultra-defensive strategies by returning to a Full Base Mineral Model, we have also simultaneously provided more options to the player who faces ultra-defensive strategies.
Unlike in HotS, the no worker pairing model provides better income to players who are able to achieve a 1:1 worker to mineral node ratio. Due to greater mineral income from spreading workers across bases, players are rewarded greatly for being able to expand and saturate new bases. Recall how in HotS players would have had to sacrifice army supply to have an economic advantage against ultra-defensive opponents. In the double mining model, players do not have to sacrifice army supply to obtain a larger economy. Instead they only need to out expand their opponent and spread their workers across the map.
An ultra-defensive HotS strategy that does not harass its opponent mining on 24 mineral nodes will have to trade against an opponent’s equal supply army who has double the mineral income since they have full map control and up to 6 bases (the new 48 mineral node cap). The six base player has more of the map to base trade against, a stronger core army than in HotS and 25% more mineral income as well as double the available gas geysers. Ultra defensive strategies therefore, while still viable, will need to consider the power of their opponents economy and production, and will probably have to do more than tech to 3/3 and 200/200 army supply to win the game.
This increases strategic diversity. Instead of limiting the number of strategies a player can import from HotS or WoL, we instead provide players with more options for counter play. The player who is expanding aggressively needs to spread their workers, and needs to control much more of the map as they do so. The closer to a 1:1 worker to mineral node ratio a player has the greater their income is due to worker efficiency making each worker more impactful. By spreading workers out across their various bases, the expanding player provides more opportunities for their opponent to harass, increasing action across the map.
While expanding rewards a player with more income, staying on fewer bases makes harass easier to defend. In the HotS mining model, losing 8 workers at a base with 16 workers results in 50% less income for that base until a player is able to resaturate. In the double mining model losing 8 workers of 16 is less damaging due to the lower efficiency of the second eight compared to the first eight workers. Instead losing 50% of your income, you lose only 40% of your total income. While this might not be the biggest difference, keeping 60% of your income after losing half a mineral line could increase the chance of a comeback.
While there are a lot of positives to double mining as a result of the reduction of worker pairing, there are also a few negatives as well. The first is the higher income overall of a saturated base in the double mining vs the current economic model. Bases in the double mining model result in an additional 15% mineral income compared to their counterparts in the current economic model at the 16 worker count and only normalize at the 24 worker count on a single mineral line. This has far reaching implications for balance with regards to unit costs, the mineral to gas ratio, and overall pace of the game.
The overall mining curve is much sharper and could be far too severe an income rise for one base. Ideally, you would want to find an economic model which closely approximates the current mining model in SCII (2:1 ratio mining curve) at both the 16 and 24 worker breakpoints, while still providing a higher income at the 8 worker count (1:1 model provides more income workers 1 - 8 compared to 2:1 model). This would preserve income ratios for players who are playing more defensively, encouraging rapidly expanding players to invest a few more resources into gas mining as they spread out across the map to turn their economic mineral advantage back into a tech advantage. A similar income on three base means we can evaluate the impact of a fourth or fifth base by comparing the new model to an existing model with five years of gameplay and balance behind it.
Mineral Income Curve for Double Mining Might be Too High
Another large problem is associated with the increase of mining time. When a worker takes twice as long to complete their harvesting action, stopping this action prematurely results in a huge hit to the economy. Instead of missing out on 5 minerals from interrupting a 2.7 second harvest cycle in the current economic model a player who pulls their workers will lose 10 minerals from interrupting a 5.4 second harvest cycle. I want to thank Artosis for pointing this out as a huge downside to the double mining model. He made reference to the impact that pulling workers in PvP against a one gate proxy would have and how large an advantage the proxying Protoss player would gain. Similarly, he made reference to kind of advantage one could gain in BW ZvZ when forcing the early workers to pull off the mineral line. This mining model makes worker pulling extremely costly for the player who is defending and puts the aggressive player in a large advantage. Exactly how detrimental this would be in a real game scenario is hard to predict.
Zerg in particular may become problematic with such a high mineral income since they are the race with a 300 mineral cost hatchery which also doubles as a production facility. The larva inject mechanic is the only concerning race specific mechanic in this new economic model. Inject may or may not result in an extremely quickly growing economy which could get out of hand. More than just an issue for XvZ matchups, the ability to turn the economic advantage into a severely large army advantage could be problematic for a defender. This requires exploration in game and can only be fully understood with many games played using a new economic model. So the issue of unit balance, and race balance, while important to mention, is not important when examining any economic model that removes worker pairing. Numbers can change later, the core design of the economy and its impact on strategic diversity is more important than discussions on unit or racial balance.
In short, the double mining model may be a touch too extreme. The increased early mineral income may be too sharp, the increased overall mineral income and higher mining curve may be too extreme, and forcing workers to pull off a mineral line could be far too powerful. The positives however associated with increases strategic diversity and options for players justify an exploration of removing worker pairing in SCII and bringing the perfect mining efficiency ratio down from 2:1 to 1:1. We decided to explore a variation of the double mining model and found what may be a perfect compromise. The alternative non-pairing worker model still achieves the same goals as double mining, but softens the associated negative blows.
The Double Harvest Economic Model
In a number of economic model trials we tested and analysed, we found that the best way to approach no worker pairing is likely the Double Harvest model. We borrowed the basic concept for the Double Harvest model from BlackLilium and Uvantak. In the model presented by these two individuals workers complete three harvest actions instead of one in every trip to the mineral line. We found that similar to the Double Mining model, the mining curve may have been too high, and the return of 15 minerals instead of 5 may be too punishing due to the potentially high number of lost minerals on worker death in scenarios of harassment.
The model we would like to see as a trial in the LotV Beta is the Double Harvest model. The Double Harvest model is a simple and elegant solution to the negatives seen in the double mining model while still rewarding expansion based gameplay (as opposed to punishing players who do not expand). The Double Harvest model removes worker pairing by abusing the existing AI of workers in SCII. The “check, wait, harvest” cycle remains the same as it currently exists, with workers arriving at a mineral patch, checking to see if it is free, waiting, and then harvesting 5 minerals. What the Double Harvest model does however is it forces workers to complete two harvest cycles before returning to the Town Hall building to return 10 minerals instead of 5.
A Single Worker Harvest Trip in the Double Harvest Model
This eliminates worker pairing by forcing harvesters to mine from more than one mineral node before returning home with 10 minerals as the number of harvesters on a mineral line goes up. For example, when a probe gets to a mineral patch, they will wait for the first worker to complete the first harvest cycle and then discover that it is actually not free, forcing them to move on. This removes worker pairing and due to 10 mineral returns increases the initial income.
A full mining base of 16 workers in the Double Harvest Model
The mining curve in the Double Harvest Model is not quite as steep as in the Double Mining model for the first eight workers but is still high enough to retain the faster pace of a quicker opening build order. The overall mining curve is also very close to that of the current SCII model as the worker accounts rise and reach the 16 and 24 worker breakpoints on each base. As you will notice in the mineral curve comparison image below, Double Harvest mines about 5% more than the existing model when players have 16 workers on a base.
The Double Harvest mining curve compared to the HotS, worker pairing mining curve on one base with 16 workers
The Double Harvest model does more than just impact the early game mineral income curve. It also results in players being greatly rewarded for expansion based gameplay by removing the 2:1 cap found in the current SCII economy. Starting with the natural expansion, players who manage their worker spread will see a minimum of 24% more income than their one base opponent assuming an even worker count. By removing worker pairing, and eliminating the 2:1 worker cap, even just a natural expansion makes a huge difference in mineral income.
Players with a natural will have far more income than those who do not at even just 16 workers
Shown in the image above, the income of 2 bases with split workers on minerals in Double Harvest is also far higher than in HotS. Players who believe their opponent is doing a one base all in have the option to cut workers early, and invest the greater economy on same worker count into defense. In a Half-Patch economy where the 2:1 worker ratio remains the expanding player's income is only greater once the natural begins to saturate more fully, meaning they must not only invest in a natural expansion but also workers to see a higher income.
In a Double Harvest economy players are given the choice to cut workers when defending aggressive actions with greater income on similar worker counts thanks to the removal of the 2:1 income ratio. The timer placed on the one base player is no longer set by Half-Patches which will mine out if an attack fails. The timer is instead controlled by the player with the expansion. Every minute the natural base is mining, the greater the mineral based defense becomes and the less time the attacker has to equalize the game with either harass or an attack.
Overall however, as worker counts rise and players obtain 16 workers their income rates begin to stabilise and these income rates become close to those seen in HotS. Maintaining a similar feel to the economy as players approach a HotS like mid game is extremely important. Blizzard stated in the LotV beta 1.0 notes that: "The main goal here was to make a change that would keep the feel of resourcing rates similar to Heart of the Swarm ...". The Double Harvest model approximates this on one, two and three bases with 16 workers on each mineral line. The image below shows income on two bases with 32 workers comparing Double Harvest and the current Worker Pairing model.
Two bases with 32 workers in Double Harvest compared to Worker Pairing (Standard)
But how does the Double Harvest model look when we take two players with 48 workers on even greater base counts? As we established previously due to lack of worker pairing, players with more mineral nodes available to them will have a higher mineral income than their opponent. The image below shows mineral income per minute for a player who spreads their workers with three, four, or five bases. On all base counts, the player has 48 workers mining minerals which represents 16 workers on three mineral lines. The four base player has 7% more income than the three base player. The five base player has 20% more income than the three base player. As you can see, the mining cap has been removed and base scaling has increased significantly when similar worker counts.
Relative income on 3, 4 and 5 bases with Double Harvesting and HotS and 48 mineral workers
By removing worker pairing, and unlocking the 2:1 ratio, the Double Harvesting model encourages players to expand through a reward based system, as opposed to a punishment based system. The more bases one player has over another on similar worker counts results in higher income. The option to out expand your opponent in order to obtain an income advantage truly returns. Players can obtain higher income without sacrificing army supply simply by expanding more and splitting their workers evenly between all of the bases. This increased army supply can then be invested into harass units, core army units, or used alongside increased income to trade (even inefficiently) with the opponents army and forcing them to starve out. The more money and available supply a player has the more options become available to them, and the greater strategic diversity.
In the image below, you will see the difference in mineral income per minute between four different scenarios. The first scenario is HotS (purple) where players have achieved the 2:1 cap with 48 workers and 24 secured at least 24 mineral nodes. No matter how many bases, unless a player makes more than 48 workers the mineral income will not change. The last three scenarios are Double Harvesting on various base counts with 48 workers. The 3 base income in Double Harvesting (Blue) approximates that of HotS. The 4 base player (red) gains a slight advantage over the 3 base player in mineral income as well as additional gas geysers being available. In HotS many would take a 4th base only for the gas, now we can see however that you also gain a mineral advantage. The Five base player (green) has a significant income advantage to the three base player in both the HotS and Double Harvest models.
HotS Income alongside the Double Harvesting Income on 3, 4 and 5 bases with 48 workers
The biggest impact of the Double Harvest model (other than removing worker pairing) may come from the way workers would now interact with harass-based play. Pulling early workers is far less dangerous in Double Harvest when compared to both Double Mining and Brood War (ZvZ early drone pulls often put the defender in a very poor position). The reason for this is that workers mine 5 minerals in a 2.7 second harvest, and then put those first 5 minerals in what is best described as a "basket". They then complete a second harvest cycle and with a full 10 mineral basket, they return the full load to the town hall building. Pulling workers instead in this model interrupts only one of the two harvest cycles as opposed to interrupting a full 5.4 seconds harvest cycle in the Double Mining model reducing the impact of pulling workers and resetting the harvest cycle in the early game.
The basket system has almost no negative interaction with harass. Losing a worker with 5 minerals does result in the permanent loss of those 5 minerals. but his is no different from the worker losing their 5 minerals when returning to the town hall in the current economic model in SCII. The visual cue that a worker is carrying 10 minerals is the mineral package all workers carry when returning home. Due to the fact that there is no visual cue for workers who are carrying 5 minerals in their basket, incentivizes a player using a single target harassment unit (banshees, oracles etc) to focus their efforts on killing workers returning mineral packages, thereby removing 10 minerals from their opponent’s income and a worker instead of just five. This is a small interaction that may not make a difference to most players, but may show a larger skill discrepancy at higher levels of play.
The biggest downside from an implementation standpoint for the Double Harvest model is how visually clear it is or is not. I do not have the skills to create a visual cue for when a worker has a 5 mineral basket. Perhaps some small blue outline could surround the workers, a slight glow, or a small mineral package model could be used to visually represent that the worker has a mined 5 as opposed to 10 minerals.
Comparison of the Three Models. The sharpness in the Double Mining model is attenuated by Double Harvest, and 16 worker economies are close to normalized between Double Harvest and HotS maintaining Mineral to Gas income ratios
The overall mineral income increase in the Double Harvest model results in bases mining out a little sooner than in HotS. This is in line with the approach Blizzard has taken in LotV in trying to steer players away from turtling for too long. While we do not want to punish players for not expanding in our model by disrupting the mining curve too soon, a small change to the length of time a player can play ultra-defensively is not inherently a bad thing and may still be something to be explored.
We also collected data for relative mine-out times in both the standard and Double Harvest models with various mineral counts on each mineral node. The increased mineral income translates into each base fully mining out about one HotS minute (40 LotV seconds) sooner than in the standard mining model with 16 workers on a mineral line. If this is still perceived as being too slow, dropping the total mineral count to 1400 minerals may be a good compromise. With 1400 minerals on each node bases will mine out in 2:10 HotS minutes sooner (1:30 LotV adjusted time sooner). We believe it might be worth trying at 1500 minerals but starting with 1400 wouldn’t be the worst. It still provides players with far more time than they currently have in LotV to plan their expansions, tech and execute effectively. Of course, any slow expansion based play may still be punished by the opponent obtaining a large economic lead, necessitating some form of harass or timing from the defensive player (or an extremely cost efficient army).
Double Harvest Mine out times vs. Standard Mining (HotS) assuming 6 worker start in the main base at 1500 and 1400 mineral count nodes
Summary and Conclusion
In this article, we examined the current economic model of StarCraft II and the myth of the three base cap. We determined that the three base cap exists as a by-product of optimal army vs worker supplies and worker AI which results in Worker Pairing. Worker pairing in SCII means that the optimal mineral income is found at a 2 worker to 1 mineral node ratio, and that as long as both players have similar worker counts, and both approach a 2:1 ratio, the number of bases a player has are irrelevant in relation to mineral income. By removing worker pairing we are able to move the optimal income ratio to 1:1, which greatly rewards players for expanding.
The current economic model of LotV does not address the 2:1 ratio. The side effect of the LotV economy is that it punishes players for not expanding and places them on a clock to obtain additional mineral nodes roughly six minutes (LotV time) after establishing a base in order to not lose mineral income. By keeping worker pairing, the theoretical 24 mineral node cap (24 nodes coincide with three bases in HotS) does not go away. Half patches alongside 12 worker start do, however, limit some of the strategic diversity we currently see in HotS. The goal in LotV should be to further increase strategic diversity by adding options while removing as few as possible. This goal is better served through the removal of worker pairing as opposed to introducing half patches and 12 worker starts.
By increasing the efficiency and overall income of the first 8 workers, we can return to a 6 worker start with build orders which develop more quickly than their HotS counterparts. This not only preserves some of the extreme early game and Protoss proxy strategies, but it also provides players with more time to scout their opponent’s build order before they fully develop themselves.
The loss of pairing results in achieving the goal of removing the 24 mineral node cap. The Double Harvest model in particular rewards expansion based play while retaining the strategic choice to expand more slowly. It also creates the requirement to slow down the opponent’s economy via harassment so as not to be outpaced, creating more action on the map. Spreading workers and properly managing one’s own economy (while also considering the economy of the opponent) becomes far more important.
Over-making workers with too few minerals to support them is far more punishing than it is in HotS. The notion of “over droning” typically felt by Zergs becomes a concern for Terrans and Protoss as well since investing too many minerals into workers on too few bases results in a much smaller army, opening a timing for the opponent. Scouting the economic choices of your opponent becomes just as important as making economic decisions for yourself. If they far outpace you in expansions, you must either slow them down or build an extremely cost efficient army.
Most importantly, players are rewarded for expanding instead of being punished for not. Without cutting much of the existing strategic options we currently see in SCII and adding far more interactions with a better scaling economy, strategic diversity can only grow.
If you would like to see all the data I collected during my research and analysis you can Download the Excel Spreadsheet Here