# Comparing the Classes of Arkham Horror; Why Survivors Need Work

## Introduction

This blog post was prompted by this meme posted in the Arkham Horror: The Card Game Facebook group:

This meme got 144 comments and 94 reactions, and I’d say that most of them were in effect agreeing with the point the meme makes; the Survivor class feels like a black sheep among the Arkham Horror LCG classes (not that there aren’t people who like Survivors).

Seriously, we’re three cycles in, starting the fourth, and there’s no Level 4-5 Survivor cards? It makes you wonder if there will EVER be high-level Survivor cards!

I’m sure you can make one that does [sic]. Here’s one: King of the Hobos! When you have no resources, take two extra actions.

I dunno, just give us something. Late in the campaign it becomes a pain to upgrade your deck because all cards are low XP. It also makes seeing “Survivor cards level 0-5” a joke and Lola basically another survivor.

This comment, on its own, got 58 replies. So let’s just say that how the Arkham Horror designers are handling the Survivor class is a hot-button topic. It’s pretty clear what I think about how Survivors are handled. But I also wanted to see to what extent the community agreed or disagreed with me.

So I created a poll (now closed) asking people to rank the Arkham Horror LCG classes from best (1) to worst (5). I’ll first present this and other data, then present my own opinion.

## Data

The first data I saw related to this was a response to my comment, showing the popularity of investigator decks on ArkhamDB. Below is a screenshot of the interactive tool (that you should check out):

(This data set does not include the investigators released with The Circle Undone, but considering how new the cycle is that may be for the best.)

Survivors supposedly occupy the low tier of ArkhamDB decks, and thus are less popular. I don’t agree with this conclusion from this data set; as others pointed out by others, lots of people may be playing survivors without posting their decks, and the amount of new decks being made doesn’t necessarily correlate well with how people feel the class performs. For instance, supposedly there are not many Mystic decks because all Mystic decks include the same key cards and thus look largely the same. (And that’s not a good thing, by the way.)

I also want to add that the fundamental question is how good a card pool is rather than how good Survivor investigators are. I don’t think that there’s anything wrong with the Survivor investigators in a vacuum; they’re all great investigators. (That includes Calvin, too; he’s not just a “challenge” investigator, he can pull his weight and more in a game when played right.) That said, the identification of an investigator as a “Guardian” or “Seeker” or “Survivor” serves no mechanical purpose. Nothin in the game pings off the class of the investigator; all investigators could be in the “Neutral” class (like Lola) and the game would be the same because of the deck building requirements. All that the class of an investigator does is indicate what the deck building requirements will be. And some investigators, such as Carolyn and Norman, really drive this point home with their deck-building requirements.

So to answer this question, I created a poll, asking people to rate the card pools of the classes. My question was simple: “Rank which class card pool you prefer, from most favorite (1) to least favorite (5).” Not everyone agreed that this was the best question to ask; there are different aspects in which classes may be “better” or “worse”. My primary Arkham Horror LCG partner and friendly local game store owner Matt Freed (who owns Mind Games, LLC, in West Valley, Utah) refused to answer the question as phrased because his rankings would differ completely depending on whether the class in question was the primary class or an off class of an investigator (I told him that I cared most about the card pool class as a primary class of an investigator). That said, the simplicity of the question (plus attempting to publicize the poll on the Internet as well as I could) managed to get me 421 responses, a decent sample size.

When looking at this data one must remember that this is not a random sample as statisticians prefer. The people who participated chose to do so and they’re all from the Internet, which means that one can raise questions about the external validity of the data. That said, I think that we can still learn a lot from the data, even if it’s not perfect.

By the way, analyzing the data I got was my first forray into analyzing rank data. I had to learn new statistical methods to meaningfully pry into the data to see what it said. I loved the methodology and the details of what I did will be presented in a later post, along with the scripts I wrote for doing this analaysis. For now, I’ll just mention that my primary reference for learning how to analyze the data was Analyzing and Modeling Rank Data, by John Marden.

Let’s start with some basic charts. Below is the “marginals” plot; simply put, it’s how frequently a class was assigned a particular ranking.

One reading of this plot is that Seekers are rated high, Survivors are rated low, Guardians are rated third, and it’s hard to tell how Mystics and Rogues rank, though it appears that Rogues are better liked than Mystics.

I however don’t like this plot since it doesn’t take advantage of the fact that the data is rank data. A plot that better accounts for the nature of the data is the “pairs” plot, which shows how many people prefer the $x$-axis class to the class represented by the bar. If the bar is above 50%, then the $x$-axis class is preferred to the bar class, while if the bar is below 50%, the bar’s class is preferred. The pairs plot (with a line marking the important 50% cutoff) is shown below:

It’s much clearer from this plot which classes people prefer. Seekers are preferred to all other classes. Guardians are preferred to all classes but Seekers. Mystics are peferred to Rogues and Survivors, and Rogues are preferred only to Survivors. The Survivors are handily in last place, not being preferred overall to any other class.

This suggests that the ranking of the classes are, from best to worst: Seeker, Guardian, Mystic, Rogue, and finally Survivor. This ranking was the ranking obtained by one estimator of the central (or “closest to consensus”) ranking obtained by finding a ranking that minimizes the sum of Kendall distances (which is closely tied to the pairs plot). Also, a statistical test confirmed that the respondents were not equally likely to give any ranking and thus have preferences. A 95% confidence interval could only conclude, though, that Seekers were preferred to Rogues and Survivors; any other ranking is plausible under that confidence interval. That means that Seekers overall rank in the community is at least three, while Rogues and Survivors cannot be ranked first in the consensus ranking. (All tests and confidence intervals were based on the pairs matrix/Kendall distance.)

These are the statistics for the community considered as a whole, but it’s possible that players fall into different “archetypes” and thus may have different class preferences. Matt Newman, the lead designer of Arkham Horror, seems to believe so according to this article he wrote. I never asked players what “type” they were, but I attempted to determine types via cluster analysis, based on spectral clustering.

Let me start by saying that I’m not convinced there are meaningful clusters of players in this data. All the metrics for finding clusters were bad. But if you insist that there must be clusters of player types in the data, read on.

If there are clusters, my best guess is that there are five clusters of players. Based on reading the (optional) responses of players in each cluster, I’ve labeled (in a very subjective manner) cluster 1 as the “power” players (this is the cluster of Matt and I), cluster 2 as the “versatilityi/efficiency” cluster, cluster 3 as the “bling” cluster, cluster 4 as the “kill monsters” cluster, and cluster 5 as the cluster that likes “theme” in the game (although honestly this cluster was the hardest to define and often looked like the cluster of the most confused participants; the comment rate from this cluster was the lowest.)

31% of players were in the “power” cluster, 20% in the “versatility/efficiency” cluster, 19% in the “bling” cluster, 16% in the “kill monsters” cluster, and 14% in the “theme” cluster. The class preference in each cluster was, in order:

1. Seeker, Mystic, Guardian, Rogue, Survivor
2. Seeker, Survivor, Guardian, Mystic, Rogue
3. Rogue, Seeker, Guardian, Mystic, Survivor
4. Guardian, Survivor, Rogue, Seeker, Mystic
5. Mystic, Survivor, Seeker, Rogue, Guardian

The confidence intervals (which are starting to lose their validity in the cluster analysis due to sample size and some preference pairings never being seen in the cluster) suggest that the “power” players dislike Rogues and Survivors; any ranking that puts Rogues and Survivors in the two worst rankings could be this group’s “central” ranking. The second group is more difficult to infer; Guardians and Survivors are supposedly better than Mystics and Rogues, while Seekers are better than all other classes. The “bling” cluster loves Rogues more than any other class and prefers Seekers to Survivors. The “kill monsters” cluster prefers Guardians and Survivors to Mystics and Seekers and Rogues more than Mystics. Finally, the “theme” cluster prefers Mystics and Survivors to all other classes.

## What’s Wrong with Survivors?

I would say that, taken together, this data suggests that the Survivor class is, indeed, problematic, and players on the whole are not a fan of how it’s handled. (That said, one could make a case that this is true for the Rogue class too.) Here’s my argument regarding what’s wrong with the Survivor card pool.

Let me start by saying there are great Survivor cards. Lucky! may be the best card in the game. Peter Sylvestre is a great ally (at both levels), and I even really like Level 3 Rabbit’s Foot. Survivor cards are among the top 20 most popular (faction) cards in ArkhamDB decks.

While no one can credibly argue that Survivors have bad cards, I think we can credibly argue that the Survivor card pool is weak and eventually becomes a drag to play in a campaign.

The two most recent Survivors I played were “Ashcan” Pete and William Yorick. My “Ashcan” deck was a deck that heavily utilized Yaotl, Cornered, and the desperate cards. I was playing this deck in The Forgotten Age cycle. At first I really enjoyed the deck and how it worked. I don’t think it was the most powerful deck but it was fun to play. Eventually, though, I fell out of love with the deck and I don’t think I will ever try it again. Eventually I could not upgrade the deck without replacing what I saw as “core” cards. This problem occured mid-campaign, too.

The same problem happened with my William Yorick deck (again in Forgotten Age). Granted, Yorick has access to the Guardian card pool, which provided another outlet for spending my experience without removing core cards. The Guardian cards were an important experience point outlet. Eventually, though, I ran into the same problem: lots of experience points and nowhere to spend them (but at least it was late in the campaign when I hit this problem).

Thus I have basically one complaint with the Survivor card pool: there are no high level Survivor cards. It turns out this is by design; Matt Newman told the hosts of the Drawn to the Flame podcast in episode 22 that he liked keeping the level of Survivor cards capped at 3 since it fit with the Survivor theme of “not being ready.” Thus we are starting our fourth Arkham Horror cycle and there are no high-level Survivor cards.

What’s the consequences of this? Well, in my FLGS, “Survivor cards level 0-5” is a long-running joke since there are no high-level Survivor cards. Now Matt (the FLGS owner) said that, as an off-class, Survivor cards are one of the best pools to pull from, and I agree with him. Consider the table below:

Class Total Cards High-Level Cards (at least 3 XP)
Seeker 81 18
Guardian 79 14
Mystic 80 17
Rogue 79 12
Survivor 80 11

69 Survivor cards are accessible to off-class Survivors, more than any other class. Furthermore, these are just about all the best Survivor cards.

Now let’s take a step back. Remember when Joe Diamond was announced to be a Seeker? That was a shock to many in the community, who were pretty sure that Joe Diamond would be a Guardian. Furthermore, making Joe Diamond a Seeker was a big deal that had major implications for his deck building. A primary-class Seeker/off-class Guardian will look very different from a primary-class Guardian/off-cass Seeker, both in deck and style of play.

Let’s take William Yorick. What would change if William Yorick went from primary-class Survivor/off-class Guardian to primary-class Guardian/off-class Survivor? Well, Yorick would lose access to the 11 Survivor cards that are level 3 and gain access to the 14 Guardian cards that are levels 3, 4, and 5. And to be completely honest, I wouldn’t miss any of the Survivor cards I lost; I didn’t use them in the Yorick deck I built and honestly none of these cards feel like great cards we’d enthusiastically put in decks. I’d say that the level 3 Survivor cards are generally cards that get placed in decks that have experience points to burn; that XP has to be spent on something. So by making this switch, my Yorick deck would, almost unambiguously, become better.

Thus my first point: high-level cards help distinguish class capabilities. It is because these powerful cards are not accessible to other investigators that these classes are distinctive. Seekers, Guardians, Rogues, and Mystics all have cards that make that class memorable and help separate that class from the rest. The Survivor card pool does not have this since off-class Survivors are basically just as good as primary-class Survivors. Heck, even Lola has full access to the Survivor pool! She may as well be one!

I think that if Matt Newman were to read this his response would be “Survivors lack of high-level cards is what makes them distinctive. It’s their lack of preparedness that makes them thematically work.” First, I think we’ve seen that the lack of high-level cards makes the class worse from a gameplay perspective. Second, think about who we would call “survivalists” today, such as tribal people, prehistoric humans, or guys that go off into the woods and cut themselves off from civilization. There are actually many aspects of these people that appear almost superhuman. They’re impoverished but often skilled in everything necessary to survival. Most people from regular society would die if immediately forced into the dire situations survivalists deal with on a regular basis. Survivalism is not about being unprepared or low-skilled. It’s about being well-rounded and internalizing all strengths, thus not dependent on the tools available to you. Survivalists are actually well-prepared! And high-level cards can be designed to fit this ethos.

And thus my second criticism of the low-XP policy: it restricts player growth. Upgrading your deck not only is a way to get new toys; it shows how the encounters with the mythos caused the investigator to grow and improve. By preventing access to high-level cards, the investigator’s growth is restricted. There is access to a lot of low-level cards, but putting these cards into a deck pushes out other cards to such an extent that the deck at the beginning and end of the campaign look extremely different. This was the case with my aforementioned Ashcan deck; I could not upgrade it without drastically altering it. I could not keep the same deck archetype while at the same time staying a Survivor. The deck would have to transform in character in order to upgrade; it wasn’t really possible for the deck to just get better at what it already does.

Now there are the exile that can help players burn XP. But I hate those cards! Not only are they very narrow cards, I don’t like burning XP when I play a card. (That said, I like the upcoming Survivor ally Guiding Spirit and wouldn’t have a problem putting it into a deck even early in the campaign.) I think most players don’t want to use their experience points on Exile cards either, so I wish there were other places I could spend my experience points in the Survivor class than the exile cards.

## Closing Thoughts

I’ve spent this article picking on Survivor cards but while there is a lot of evidence suggesting this class is in need of the most work there’s also evidence that people don’t like the overall design of the Mystic and Rogue classes, either. I think people’s main complaint with the Mystic class is that deck building with Mystics feels stale; there are some key cards that every Mystic deck includes and thus they all start to look the same. I think this is a valid point, and one good way to fix this would be to make a Mystic permanent granting another Arcane slot. That would make more Mystic players willing to look beyond the Shrivelling/Rite of Seeking/Mists of R’lyeh staples.

As for Rogues, I don’t see why Rogues get hate. The cluster analysis suggests there’s a class of player that loves Rogues. I think that Rogue-hate stems from a belief that Rogues are not good enough at investigating/fighting, or a general lack of interest in evading enemies (which Rogues should do well at).

Seekers and Guardians are great classes and don’t need much work. If anything, those classes are too good. But no complaints from me.

But I stand by my conclusion that Survivors need work, and that what they need are high-level cards. I think it is possible to give Survivors high-level cards while keeping with the ethos of the class. In keeping with the “Survivors’ strengths are innate and well-rounded,” I think high-level Survivor cards could consist of events, skill cards, and non-item and non-ally assets. For instance, perhaps a Level 5 permanent called “Survivalism” that allows a survivor to spend two resources to boost any skill, or a high-level Dark Horse that gives Survivors +2 to all their skills when they have no resources or assets. (I’m spitballing here, guys; I’m sure these card ideas suck.)

I bet there are probably die-hard Survivor fans who will bristle at my criticisms of the class. To them I have to ask: do you like the fact that you don’t have the option to buy high-level Survivor cards? I certainly don’t, and I hope that “Survivor cards level 0-5” will no longer be the joke it currently is.

## 2 thoughts on “Comparing the Classes of Arkham Horror; Why Survivors Need Work”

1. I played as Yorick and Ashcan Pete. With my Yorick deck I spent 10 xp on two red gloved men and they felt like the best cards ever since I could replay them all the time, steam rolling the late game. In my Ashcan Pete deck I also spent a lot of xp on neutrals and rebought fthe flare every game (because solo du which has your deck swarming with so many allys) and it felt amazing as well. What I find is that in the other classes I don’t want to spend my xp on neutrals and I think that’s sad. Great article! My fav class is Seeker because the are overall amazing (maybe too good indeed)

Like