According to a recent post on reddit, it appears that four player maps are not liked by many because of the perceived randomness in scouting. Whether certain situations are down to luck is debatable once you entangle the elements involved, and I will discuss that in this post.

Takeaway from the thread

The post shared a tweet from Artosis that reflects how much he doesn’t like four player maps and why. I think the temporal contiguity of this tweet is rather obvious (haha).

The sentiment on the thread is generally negative towards to a four player map, and I will quote a few comments here.

“I watch a lot of SC2 and honestly feel like its just plain RNG, isn’t that bad? Why dont they just have the enemy marker like in VS AI?”

“RNG, even in this small amount, does not belong in Starcraft IMO.”

“Getting a 4-p map against a random player is the best! You don’t know where they are, you don’t know what opener they could be doing! It’s a true mystery box of cheese.”

Notably, the common issue is the “randomness” on four player maps, and this can be traced back to the perceived involvement of luck in Starcraft. Although the discussion is more directed to the scouting issue on four player map, I believe there is more to the topic beyond just scouting. I will start off with a broader discussion on the luck involved, and subsequently discuss how it is related to scouting in non-two player maps.

What is luck

I will not go into what luck is in general, but it is important to at least clarify what I think luck is in the context of games. Factors that are relevant, but are meaningless to the discussion are ignored (e.g., internet connection, hardware problem etc.).

In my opinion, luck is a “force”, which comes from game mechanics beyond the control of the players, that affects the outcome of the game in favor of one player at the expense of the other (assuming it is one versus one).

Based on this definition, if a unit attempts to scout for proxy and does not spot the building because it is just a pixel away in vision, it is not luck.

It is not perfectly defined, because rock-paper-scissors would then be considered a game without luck involvement. Whether rock-paper-scissors should be considered a game of luck is another argument, but the point is that my definition should at least be adequate for discussion in the context of Starcraft.

Luck in Starcraft

It is fair to say that players want Starcraft to be a game, which allows the players to have complete control of the outcome. That is, it is a game of perfect meritocracy. Blizzard’s recent community feedback echoed this.

We agree heavily with many players in our global community that SC2 is one of very few games where you are solely responsible for whether you win, or lose…

Blizzard have also shown that they will make necessary changes to ensure the game mechanics is in line with this goal. The minimum scan radius issue was fixed in Legacy of the Void, as it affects the consistency in unit interaction beyond the control of the players.

At the best of my knowledge, spawn location is the only thing that has an element of luck in the game.

Luck in spawn location

As far as I know, spawn location on any map is random, and that includes on a two player map. Let’s take King SeJong Station for example, whether you get spawned at the top or bottom is random, and neither you or your opponent has control over it. Does it affect the outcome? Yes.

Terran will want to spawn at the bottom, because the add-on on the Barracks that completes the wall off is less vulnerable than the it would be if the player is spawned at the top. In fact, this add-on problem has been highlighted by many all the way back in 2010. The below image (taken from here) sums up the differences according to the ramp direction, which varies according to the spawn locations on map.

Such positional issue is not limited to Terran, the larva always spawn at the 6 o’clock of the Hatchery, and the travel distance of the spawned Drones to the mineral patches varies according the mineral line direction. This again is affected by the spawn location, and hence, it is down to luck. Like the add-on problem, this has also been a long standing issue. The only conclusion is that the difference is not big enough for Blizzard to make changes which may potentially confuse players.

Also, I won’t be surprised, all else equal, human play better when they are “playing towards a certain direction”. Just imagine playing chess with the board swapped 180 degree, but the change only affects one player.

Luck in four player maps

Let’s use Frost as an example. Broadly speaking, there can be three types of spawn situations on this map: vertical, horizontal and diagonal. This actually has a notable impact on the outcome of the game, because the spawn situation favors a certain race in some match ups. All else equal, diagonal spawn situation in TvZ is more favorable to Zerg than to Terran. Long story short, Terran want to attack Zerg, and the shorter the distance between the two spawn points the more favorable it is for Terran attack. This variance is down to luck.

Let’s expand this luck factor in spawn situation a little. The actual spawn situation affects the outcome, but does the uncertainty of this factor affects the outcome in regard to luck? While there is no doubt that the actual spawn situation has luck involved, whether the uncertainty of what the spawn situation is at the start of the game has an element of luck is debatable. A build may work better for certain spawn situation, and that the decision to use the build is made before the player knows the spawn situation even if you send out the earliest scout possible. Theoretically, players should pick a build that maximises the likelihood to win after taking into account of three possible spawn situations (note: conceptually different from three possible spawn points for opponent). Whether a player decides to roll the dice and picks a build that works well in a specific spawn situation is not down to luck, but the outcome is because the random spawn situation is not in control of the players.

Take a minute to think about it, because it is not as intuitive as it seems.

The uncertain condition is the same for both players, so neither player has an edge over the other (assuming racial asymmetry does not play a part here). The decisions that players make under such condition are completely under the control of the players, and hence, there is no luck involved. However, the actual spawn situation, which is unknown to the players (yet), is not under the control of the players, so there is luck involved.

Whether the uncertainty that stems from variance outside of players’ control is considered to be luck sets the foundation for the scouting issue in four player maps (or simply non-two player maps). This is because, scouting decision under such uncertainty is completely under the control of the players, but the actual spawn situation is not.

Scouting in non-two player maps

Based on the comments on the reddit thread, the issue that people are focusing on is the difficulty to get information through scouting early on in a four player map relative to a two player map, and how this results in an outcome that is strongly influenced by luck. More specifically, it is about how a standard worker scout may reach opponent’s base at a variable time, because the worker moves from one possible spawn point to the next and so on. It is in the player’s advantage that the first spawn point the worker goes to is the actual spawn point of opponent, and the later the worker reaches opponent’s spawn point the more disadvantageous it is. This is then attributed to luck.

There is an assumption in the discussion, and that is, the habit of worker scout on two player map should be the same on four player map. Indeed, it is common for professional players to send a worker to scout on non two player maps the same way they do on two player maps. An intuitive way to get around the “luck” problem is to send more than one worker to scout, and it has been brought up in the thread. More importantly, it is deemed to hurt the economy too much to be worthy to send more than one worker to scout or adjust the timing to send out a worker to scout. The below comment taken from the thread sums it up,

“Options: Send out 2 scouts, send out scout earlier, fuck it dont scout at all. Alright fair these options are all pretty bad”

But does this assumption hold? In fact, I posted a comment to highlight this assumption, and how it affects the way we look at the issue. I expected someone to reply that it is not worthy to send more than one worker, and someone did. (Oh, I got downvoted too.)

I asked on the Facebook page of Terrancraft, how do people scout on non-two player maps? They brought up some good points, and it shows that players make precise changes to basic worker scout according to map. I will use Frost for an example, Zerg can send the Overlord to the vertical spawn location, as this can potentially spot opponent’s town hall at the natural. In line with this, the Drone can then move to the horizontal spawn to scout the other direction. Alternatively, the Drone can move to the diagonal spawn and check the middle of the map for proxy.

However, the other two races do not have Overlord, and early scout is solely reliant on worker scout. The decision to use the usual one worker scout and issue queue command to it to move to the three spawn point one after another is in complete control of the player. In other words, given the fact that the player knows the possibility of three different spawn situations, the player decides in game theory sense that investing more in scouting (i.e., send more workers or send an earlier worker) is not in his/her favor due to the mining lost. It is important to note that the decision made is not affected by luck, as the information you have to make the decision is the same every time even though the actual spawn situation is random and unknown. Now you see why I said the argument I put forward in the last section is related to scouting in a four player map.

After acknowledging that the decision to whether investing more in scouting is not down to luck, the next question is whether investing more resource in scouting than a typical one worker scout at the standard timing is worthy. This is an extremely complex question to answer, and DeepMind (AlphaGo) may just be the “chosen one” to answer this. Let say the answer is yes, what should be done? In my opinion, if you send out a worker to scout at the “standard” time, it is hard to send out an earlier scout that actually makes a difference to the problem. Thus, it seems that it is more fruitful to think along the path of sending more worker to scout instead.

The reply to my comment on reddit suggests that “sending two workers is horrendously bad for your eco”. How does one come to a conclusion that more than one is horrendously bad, but one is good? This is a good presumption to challenge, albeit it may be right. Instinctively, it appears that the confidence in such unjustified presumption is down to the fact that we have little to no doubt in what we have been seeing through out the years. A norm is not necessarily “right”.

Furthermore, have you wonder what is the difference between a worker scout on a two player map and non-two player map? On top of the information you gather on a two player map, a worker scout on a non-two player map also gathers information on the opponent’s spawn location. Then, can one say that a worker scout in a non-two player map is actually more valuable than the one in a two player map? Clearly, you may argue it is not that straightforward, as a worker scout in a non-two player map may not gather the usual information you get on a two player map because the worker may not go the “correct” spawn location early enough. This then leads back to the problem that motivated Artosis’ tweet at the beginning of this post.

Let’s continue to challenge our thinking, doesn’t the above train of thought mean that we actually value the information we can get from a two player map with a normal scout very highly? Then, is it not high enough to invest more resource to increase the certainty of getting this information? If the answer is no, why not delay worker scout on a non-two player map for more mining, as the delayed worker scout still gathers information on opponent’s spawn location. If I don’t get “lucky” with my usual worker scout to gather information I would normally get on a two player map, a delayed scout is a better choice on a four player map. If the answer is yes, why are people so reluctant to invest more in scouting on a non-two player map?

I hope at this point, you start to consider that we may need to rethink our approach to scouting on non-two player maps, and how much we should actually attribute the outcome to luck.

I will end this by making a suggestion to worker scout adjustment on a four player map. If you do not spot opponent’s spawn location at the first spot with a normal worker scout, send a second worker to the third spot while the first worker continue to move to the second spot. This basically changes the subsequent 50% of “getting lucky” to 100% with a delayed second worker scout. The rationale behind this is that a worker reaches opponent’s base early enough to deduce whether you need to make immediate reaction, and the “I scout them last and lose directly” situation that Artosis is suggesting is less likely to happen.

---

If you enjoyed this article, I’d love you to share it with one friend. You can follow me on Twitter and Facebook. If you really like my work, you can help to sustain the site by contributing via PayPal and Patreon. See you in the next article!