Input and Output Randomness in Kingdom Death (Patreon)

Fore note: I wanted to write more about this and tidy this piece up, but I got my second Covid-19 vaccination yesterday and it came out swinging like Mohammad Ali. So if there are unfinished thoughts or bits that could have been expanded on further, please forgive me. This is actually a super interesting part of game design and it's something that the video game industry is well ahead of the board game industry in understanding and constructing around.

So before we get off into the realms of the Sunstalker; I have been wanting to write a piece about the way that randomness and chance are used in board games, both in general and then in specific for Monster. It's an interesting area of game design and KDM is an excellent example because it holds pretty much the entire spectrum of randomness, from dice to cards, from input to output and from good to bad. But before we dig into what I mean by all of that, lets just walk through randomness in board games and explain the concepts.

So randomness is an aspect of games where the outcome of a choice or event is obscured from the player to some aspect. You do not know what the outcome will be in advance and it is by design intended that way. The classic examples of this are throwing a dice to determine an outcome or drawing a card, but they can also take the form of other player decisions, having decisions that must be made without knowing the outcome(s), drawing lots and so on.

Randomness in gaming holds a unique place in that it exists to create spaces where the outcome of a decision is not fixed and this is one of the major leverage points adopted by the genre of games known as thematic (previously called Ameri-trash) and it is something that is avoided a lot in Euro games. If you put your piece on a space in a game and you know what you are getting out of it, then that would be a zero randomness situation, where as if you put your  piece on the space and have to roll a dice or draw a card then that is providing varying degrees of randomness. Which is all very straightforward and something that we all should instinctively know.

The advantages of having randomness being a part of a game, especially a solo or cooperative game, are that it increases variation in the repeated experience and that it stops things being easily solvable. If all of the set up and outcomes in a game a predetermined with a single solution then you arguably no longer have a game, you have a puzzle, once it is solved there is no reason to return to it – and that is one of the core things that causes people to take issue with the genre known as 'walking simulators'. However, a game can be replayable without this kind of randomness if it embraces a randomized start up, the board game Newton is very close to this when you solo play it, you have almost complete control over your path through the game, but the starting set up does have some randomness in it. In short, the mastery of a game like that is over the mechanics rather than the environment.

You can even have forms of randomness that seek to assist in flattening out the skill curve, where the potential outcomes are better for a player who is doing worse. This could be done in a board game by having a different deck, dice results pool or similar that is available for a player dependant on their position on the victory point track. What this does is give players with vastly disparate levels of skill and experience in the game a closer shot at 'anyone winning'; but it costs the potential for players to return to your game over and over because becoming skilled in the game's mechanics are not rewarded.

Now, for the rest of this we are going to look at the different types of randomness that are used in Kingdom Death: Monster, why that particular mechanic is used, the benefits of it and if there are alternatives that could have been applied (and if they had been, what the cost would be). KDM is very much a descendant of the dungeon crawling genre and it belongs to the oldest family, which is the Warhammer Quest family. It has certainly done enough to branch off and present its own element of the genre, but you can trace its lineage back down to Warhammer Quest, not Descent. Descent holds the other initial main branch of the dungeon crawling genre and it has most recently experienced its own taxinomical split with the advent of the low randomness, euro-style game Gloomhaven (and the less well known Perdition's Mouth).

Warhammer Quest as a game embraces the full suite of 1990s randomness, with environments being generated by cards, journeys being filled by events from dice rolled on tables, even foes, quests and combat is randomized in a whole bunch of ways. In Warhammer Quest (WHQ); the strongest characters tended to be the ones with the best ways of handling this randomness, Wizards in particular have a lot of definitive spells that do something known when you activate them (and some random ones); but the strongest potential character in the entire game was the Chaos Warrior, because everything about them was randomized – they could end up utterly dominant and powerful, or they could turn into a Chaos Spawn and be eliminated from the campaign during levelling up.

A lot of core randomisation mechanics migrated across from WHQ to KDM because of Adam's age; he grew up with this style of 1990s game and it completely informed his design ethos. Adam's single greatest contribution to this genre was the advent of the Artificial Intelligence and Hit Location decks; both of which represent some of the best forms of randomisation we have seen in games. Thanks to the various items which interact with these decks we have a huge amount of potential skill development and learning growth that players can undergo (also it provided items for the hipsters to rally against in the Cat's Eye Circlet and Rawhide Headband). But that's not the only kind of Randomness the game uses, and there are severe costs that it has paid for its adoption of the 1990s design that WHQ represents. We're going to talk now about input and output randomness.

Input Randomness is a very interesting form of randomness because it is a kind that the decision is placed around. For example, when you have a Wisdom Potion in play you are experiencing input randomness. The Hit Location deck has been randomised, you have no control over the order the deck was shuffled into, but you do get to see what the next hit location is and you can plan accordingly. The Cat's Eye Circlet family of support items and the Rawhide Headband also all perform forms of Input Randomness and you can see just how powerful that this is. In short, the randomness happens and then the player makes the decision based on that randomness.

The act of activating the Rawhide Headband gives the players multiple pieces of key information and also some control over the outcome of the randomisation. Not only do the players get to decide the order of the next two AI cards, but they also get to see exactly what that action will be and make appropriate adjustments. Perhaps looking to remove the card through a wound (another super interesting mechanic we'll talk about in a bit) or position the right survivor to be the one who bares the brunt of that next card. In fact, what's notable about the AI deck is that we are missing a gear that performs the same function as the Wisdom Potion does for the HL deck. Getting to see the card, but not be able to alter the deck's order.  I am hopeful that the future expansions do provide this missing link, because the experience of playing with the passive ability of the Wisdom Potion is very different from the ones you get when playing with the CEC family.

Output Randomness on the other hand is a form of randomness that decides the outcome of a decision. In contrast to seeing what the randomisation has provided and deciding what to do as a consequence; in Output Randomness you instead make a decision based on whatever information you have and then the outcome comes afterwards. An example of this would be activating the Choreia innovation, you know that there is a chance a male will get the Harvestman fighting art, or that the female involved will eat him and gain the Death Touch Secret Fighting Art (or most of the time nothing will happen). So you can make some decisions (like having a female who you want to gain Death Touch and a male you don't mind losing), but the final result is out of your hands.

This output randomness also has a range of how wild it is, something like the aforementioned innovations, where you roll on a table of known outcomes, is relatively low output randomness because you choose to interact with the randomness; you can avoid it if you don't like any of the results, also you know the weighting on the percentages and they are not a flat distribution (same % chance for each outcome). In general you'll find that most players are quite content with the settlement phase's design of endeavour randomness because of this optional, known weighted outcomes design. In fact the whole of the settlement phase endeavour action selection has this, it's the most euro-style portion of the game due to its being at heart a worker placement game. It is this lack of control and the frustrations that it can create which sit at the heart of much of what can cause a player to drop a game. KDM does a great job in making a lot of input randomness look initially like output randomness, especially when you consider monster behaviours, but it never manages to shed the trappings of output randomness fully and some of the most maligned areas of the game lean almost completely into output randomness. It is very hard to get both of these two things right and for the most part APG got the Input Randomness right (there's some debate over the HL/AI control because of how limited the options are and how powerful they are); but their work on the Output randomness can leave a lot to be desired.

The Showdown in contrast varies a lot on how much input vs. output randomness there is; when you start out, without any knowledge or experience then the output randomness is very high. Because not only do you not have any tools that allow you to discover or control what the monster is going to do both when it hits and is hit, but also you have zero experience of what the monster's typical behaviours are. One of the interesting parts of the AI deck is that once you have passed through it a single time, you can know every single attack that the monster is capable of on the second (and later) pass through – the monster becomes more and more predictable the longer the showdown lasts.

More than that, due to the exceptional design of some of the monsters, you can even end up coming to know roughly how a monster acts and reacts without even having access to any deck scouting/control options. This is part of why an experienced KDM player can make the White Lion look tame even if they do not have the CEC or a Rawhide Headband – they have a far further information horizon than the new player does, the experience of slaying dozens and dozens of iterations of the monster previously add up and inform the decisions being made. That's because the White Lion's behaviour is very simple and clear and you can control a lot of it through variables that are not random, like positioning and use of encourage.

For the most part it likes to target the nearest survivor or punish knocked down survivors – and many monsters in fact boil down to this, with experience you can draw up a set of rules for how to deal with the monster and lean on the control/scouting elements a lot less. However, because the price for failure in KDM is so high (loss of survivors, loss of the year's new resources) going into fights blind is something that players tend to do only if they have a good buffer to soak up failures. I will, as a rule, always advise that someone goes into a fight against a new monster, or a new level of a monster – either over prepared, with gear that might be too strong, or with utterly disposable survivors. Either aim for a walk in the park, or be content with the worst possible outcome the less information you have about the monster you are going in against.

In respect of output randomness in the showdown, the most overt version is the dice rolling involved in attacking; we can control this to a certain degree by making sure that we have good accuracy and strength/luck when attacking, but because the system uses the WHQ/Chainmail/D&D design of having the maximum number always be a success and the minimum number always be a failure, there is a certain degree of flattening occurring – especially when you compare this to say how players attack in Gloomhaven.

Gear construction is also quite beloved because it represents a fascinating form of input randomness. You get the resources randomly from the monster showdown, but that is the end of the randomisation (Gorm potions aside); once you have those resources back at the settlement then you get to decide what to do with them. Spend them all, spend some of them, save some, convert them into other types of resources. It's all in the hands of the players to optimise as best they can.

But, crucially, certain gear cards remain in the hands of what is essentially output randomness, many of you have probably experienced the pain of trying to get an Eye of Cat, Spiral Horn or worst of all the Hardened Ribs of the Dragon King. When you have a specific item you need, and the only way to get it is through a specific resource, then the game can devolve into crit-farming the same monster for years and years. Which is why the existence of an item like the Wisdom Potion, which is built with generic resources and provides a weaker alternatives of the CEC is such a good piece of design. You are given an alternative pathway to pursue – with multiple different steps to get there. You can endeavour more Gorm innovations (even getting them directly via Gorm brains) to get more dice to roll when crafting and you can also use past potions to gain rerolls. All of which give you more control and options to get to a space where you can have an item which is a facsimile of what you were aiming for.

It's also why Rawhide and Leather are so absurdly powerful in the game, both of them can be crafted with very no dependence on a monster specific resource to complete the set. You're not waiting for a handed skull, spiral horn, hardened ribs, shadow tentacles or scarab shell to drop. Instead you can make use of any old hide, scrap and bone to get these outfits online. That in itself isn't a bad thing, but the issue that raises its head is that these armor sets are superior to the monster versions, it is harder to make a complete set of Gorment, White Lion and Screaming Armor than it is a set of leather, but leather is superior to two of those three sets and Rawhide is generally half the ost. So the question is; why bother with them? Why bother with all that work and those random draws if you end up with something that does less than the generic options.

It is this lack of weighting on the amount of randomness involved in getting the parts of an armor set that stands near the heart of what has happened to our armor meta. No matter how good the Dragon Armor set actually is; if you never see the hardened ribs, you can't get a full set – and full sets are worth. So. Much. More. Than partial or mixed sets, because the individual pieces of various sets are incredibly difficult to synergise.

Finally; we can't finish without talking about the two most extreme forms of Output Randomness that the game presents. These are of course almost exactly the same mechanism, just expressed a little differently in each, and that is the Hunt Event Table and the Settlement Event Cards.

Both of these are completely flat sets of randomness, with every single event weighted to occur the same amount as the rest and they also occur after the player has made all possible decisions. This is pretty much the furthest along the randomness spectrum you can get and this can be a fine thing if the results are pretty tight in their relative power/benefit level. However, as we are all very keenly aware, that is not the case. An encounter like the Mask Salesman or Saliva Pools is completely different to one like the Mudslide or Harvester or Lights in the Sky vs. Plague; and this combination of flat probability with highly varied effects creates an interface that players can feel frustrated and helpless.

Interestingly, I in recent times experienced a game from the designer Peer Sylvester, which is the closest I've seen to the Hunt Phase out of any game since the original version of it came out in Warhammer Quest (I've mentioned this before a few times, but the Hunt Phase is a literal direct translation of the Warhammer Quest journey phase, with a few extra bells and whistles). This is The Lost Expedition, an incredibly hard game with a whole bunch of unique events occuring while you journey. Seeing this implementation of the mechanics and how every event gives players hard choices, how the game gives some control over the order of events (making it feel like scouting) and I honestly hope that someone at the APG design team sees this game because it could turn a maligned and overlooked portion of the game experience into something engaging, challenging and exciting. I'd love to write more about this, so please do comment below if you'd like to see a pitch on the design concept.

The Settlement Event Phase is possibly even more problematic than the Hunt Phase, at least in the hunt phase most of the 'rules' have been sussed out and understood – avoid noisy, don't have everyone be insane, bring along a pick axe, Sickle and whip(s); be aware that fragile gear can delete itself, bring along at least one useless item per grid etc etc. But the Settlement Event cards have few rules for mitigation – don't wear heavy gear if you can't reroll a dice roll, get ammonia asap and make a 'murder bait' survivor. It's also more frustrating than the hunt phase due to the smaller size of options. You have typically around the twenties in different card to draw and that means the really punishing ones turn up far more than issues like the ones mentioned above. The main thing that the game wants to do here is give players more control over what happens, or at least turns what happens into input randomness. A location that provided 'seer' like powers, or similar would be great. We've seen hints that the scouting system is looking to help with this, and if it does it will probably become a 'must have' element of the game, but the single most desirable thing is more event cards.

So; there we have a bit of a discussion on the Input and Output randomness inherent in game design and specific examples in Kingdom Death. There is no 'right' way to do randomisation in a game. Dice rolls can be replaced by cards, cards can give multiple options or have multiple uses, and so on. But for a game such as this, randomisation is at the heart of what makes it replayable and engaging, you'd never want to completely remove that because then the game would be more solvable than it is already. It's a tricky tightrope to walk, and APG have managed to really knock it out of the park where the showdown's spectrum of randomness is concerned, but there are areas where output randomness has too high a set of stakes linked to it and could be improved.