Reputations and Reactions in TTRPGs
I have, over the last few weeks, had occasion to enter into a few conversations about the complexities of instantiating reputations and reactions in TTRPGs. I am particularly grateful to Nate at The Grinning Rat and Lawrence at Glorified Notepad for putting up with my sometimes poorly articulated and only partially thought through contributions in comments on their Substacks. This post has also benefitted from discussions with my regular group of players and GMs and I am grateful for their insights. As always, what follows is neither objective nor prescriptive but only an account of my subjective tastes and preferred methods for implenting them in play. Your mileage will vary.
“Reputation, reputation, reputation! O, I have lost my reputation! I have lost the immortal part of myself, and what remains is bestial.” Cassio in Othello, Act II, Scene 3
A reputation is a shifting, insubstantial thing for two reasons. First, it may or may not reflect some actual truth about the character. Second, whether true or not, one man’s desirable reputation is another man’s evil rumour. In practice, what this means is that, in settings where sentient creatures will disagree about what constitutes ‘the good’, a reputation usually cannot be phrased in concrete moral or ethical terms or in a way that presumes the ‘rightness’ or ‘wrongness’ of it when considered by others according to their own inclinations, beliefs and values.
To take two crude examples. A reputation for being ‘lecherous’ is not very useful. It is true that it describes a pattern of behaviour associated with the individual - irrespective of its accuracy. But it tells us nothing about the very thing that we should care about when assessing PC reputations - what other people think about that supposed behaviour. An ‘upright’, ‘proper’ and ‘stolid’ yeoman farmer’s wife might think of a lecherous individual in quite different terms from how a pimp or a prostitute views the same person. To the former, the lecher will probably be viewed negatively as an immoral libertine and potential sex pest while, to the latter, they may be regarded as a good customer. In the same way, a reputation as ‘a competent thief’ has no real utility because a merchant will see someone with that reputation as a menace, while the boss of the local street gang will surely have a very different opinion.
The solution to this problem of subjectivity is perhaps to classify reputations not as ‘behaviours’ or ‘inclinations’ but as simple facts. I am told that the game Yggdrasill manages reputations in this way, and no doubt there are others. Rather than say that a character is ‘lecherous’, we would wish to say that they are known or believed to have engaged in acts that indicate that quality such as ‘seduced his brother’s wife’ or ‘worked his way through every woman (and man?) in the most renowned local brothel’. The same applies to our ‘good thief’ example. To say that Bill Whitlock is purported to have used a fishing hook and line to pluck the fabulously expensive, diamond-encrusted wig from the head of the Duchess of Albemarle as she rode past is to state a supposed fact that others will respond to appropriately based on their own preconceptions, pre-occupations and outlook on the world.
From this it follows that we need to know (or be able rapidly to decide) what NPCs think about these ‘facts’ that have become attached to the character. An early trend in the hobby was for games to employ a reasonably granular and schematic alignment mechanic - such as AD&D 1e - that could be used to determine the attitude that an intelligent agent might adopt towards a character based on their reputation. However, this can only work if one makes assumptions about the place on the alignment chart occupied by the actions that generate the reputation and, as argued above, game design that asumes a reputation is indicative of an objective or universal moral quality is sometimes less than ideal.
This is going to seem terribly loosey-goosey and will perhaps strike people as not very helpful, but I don’t think there is a way that one can mechanise NPC responses to character reputations. The near limitless variety of potential NPC attitudes and assumptions to a character do not admit of reduction to charts and tables or, perhaps more accurately, any reliance on such abstracted methods is indistinguishable from that which computer RPGs already manage to do. Personally, I’m not that interested in copying what a machine can do, especially if that machine offers only a pale imitation of the available breadth of potential in-game experiences.
This does not mean that we are committed to stopping the game while we, as GMs, spend several minutes thinking deeply about the backstory of some incidental NPC to deduce their moral imperatives when engaging with a PC. We use heuristics all the time as a purely practical way precisely to avoid this situation. As a general rule most honest farmers will regard most people with a reputation for being a nocturnal burglar negatively. But, at the same time, surely we do not wish to be constrained by these mental short-cuts by having them reduced to a pro forma series of tables? Of course, one may say (rightly) that a GM is never compelled to employ any such mechanical aid. But I do think there is something to the notion that we all need to practice that which we aspire to do well. When running TTRPGs I try to present a world that has the illusion of depth and, in those worlds, it should (must?) be possible for me to say that - for whatever reason - this particular son of the soil is positively disposed towards a character with a reputation for stealing the wigs of the nobility or why this particular goodwife thinks the lecher a pitiable, almost comical, wretch. Unless I practice this habit - free from the support undoubtedly offered by charts and tables - I’m probably not going to get better at it and will rely instead on a mechanic that ‘solves’ this ‘problem’ but at what I view as a cost.
Those of you who have been paying attention will have noticed that the discussion has begun to slide towards the second element of this post: reactions. This is inevitable because the ‘reaction’ of an NPC will often be influenced by the reputation of the character. The broader point, however, is that all NPCs will always have a reaction to a character, even when they have no knowledge of who that character is beyond context clues, and even when that reaction is ‘meh’.
Much of the history of reaction mechanics stems from the fact that, in the earliest days of the hobby, the link to wargames was sufficiently robust and influential that it was logical and expedient to ‘schematize’ social encounters. The result was a series of tables and charts similar to that reproduced below from AD&D 1e where a percentile roll, modified to a degree by stats and circumstances, determined the reaction of ‘intelligent creatures’ from ‘violently hostile’ to ‘enthusiastically friendly’. Of course, from the very first, individual tables abandoned this mechanic in favour of role-playing such encounters and reaction tables themselves have undergone a great deal of refinement in the decades since. But the basic underpinning architecture remains the same - a tabular breakdown of how NPCs interact with PCs, prescribing (or suggesting) responses according to a greater or lesser degree of randomness.
In the course of my discussions with others about reaction tables, two points have been raised in their favour. The first is that they provide relief from ‘decision paralysis’; the second is that they inject a level of desirable unpredictability. Like almost every other aspect of the hobby, these claims are defensible and any objection to them is just, like… an opinion, dude… However, I am going to make an attempt here to construct a small - ‘waffer thin’ - case for why these two justifications are not entirely plausible to me.
Decision paralysis commonly occurs when the GM has not previously given any thought to some element of the game - in this context, the attitude and behaviour of an NPC - and now, faced with the need to make a choice, they have no basis for choosing between any particular course of action. This is surely more commonplace in ‘sandboxy’ games, where the odds are sharply increased that the players will wander off and encounter something or someone that no-one at the table had the faintest idea existed - until it did. So in a post-apocalyptic ‘wanderers in the wasteland’ game, the PCs happen upon a small compound of fellow survivors, on the walls of which is a look-out. Until that moment, the GM had no idea that this was on the agenda - perhaps he rolled on a random location table (but that opens a whole other clam…). In any case, it certainly seems as if, in these circumstances, delegating the decision to a reaction table to determine how the look-out responds to the arrival of the characters is a labour saving device. But is it?
Despite what may be regarded as a claim about generating a starting point for a (social?) encounter, what a reaction table commonly does is to provide GMs with an end point of the NPC’s attitude or demeanour without offering any guidance at all as to how or why that point was arrived at. Let us suppose that, after applying all of the requisite mechanical adjustments to the reaction table in our post-apocalypse example, the result was ‘enthusiastically friendly’. Cool. Now the GM has a basis for describing how the NPC speaks and behaves. Decision made, paralysis averted. But why is the look-out so well-disposed to the PCs? Or, for that matter, why would they be ‘hostile’ or any other given state? In other words, while the GM now has an end point, he still has all of his work ahead of him in deciding the reasons for that outcome. This is going to require some decisions, and so my question has to be, how much bandwidth has really been saved here?
Of course, one may object that it is not incumbent on the GM to provide their players with a reason for any particular NPC reaction and, therefore, there is no necessary imperative for the GM to know the why of things in these circumstances either. That’s true-ish, I suppose and, at many tables, the players’ actions in response to the NPC(s) will in themselves generate session content - some of which doesn’t demand much exploration of context. If the PCs and the compound dwellers are just going to try and exterminate each other, that dynamic doesn’t benefit at all from knowing (say) that the NPCs’ hostility stems from the fact that they have recently been attacked by the nearby gang of cannibal survivors. But if your reaction table came up with the result of ‘uncertain… prone toward negative’, then I would suggest that this lack of concern for the why of things is going to run out of road very rapidly, and we’re back to the need for expenditure of cognitive effort. This is not, therefore, a critique of the use of reaction tables per sé; it is an objection to the claim that their use necessarily reduces the GM’s mental workload.
The second apologia I have heard for the use of reaction tables is that they generate randomness that is stylistically preferable for some, often because it functions as a spur to spontaneous creativity. The issue I have with this contention is not the desirability of randomness or even the ways in which unexpected and seemingly non-sequitous results can generate creative responses. They may not be my default personal preferences, but I can certainly see their validity and appeal. My question is more fundamental. Why do we outsource randomness in this context to dice? More precisely, why do we employ the pseudo-randomness of dice rolls that are adjusted or nudged by mechanical or circumstantial considerations in the way that sophisticated reaction tables often are?
Let us take a doubly hypothetical example - doubly so because it will employ a notional in-game encounter and employ a non-system specific reaction mechanic. On the streets of Edwardian London, the character has reason to interact with a police constable and the GM - not having any advance notice of this encounter - wants to decide how that NPC will react to the character. The (imagined) reaction mechanic requires the GM to take account of a number of factors. What is the PC’s appearance, demeanour or attitude? What is their goal in the encounter? How do they approach attaining that goal? Does the PC have any relevant statistics, abilities or skills? All of these elements are then fed into the result of a dice roll (percentile, Xd6, it doesn’t matter) as adjustments to make a given result more or less likely on a reaction table.
What are the dice doing here? I would suggest that they are acting as a surrogate for all the things that the GM - quite reasonably - simply doesn’t know about the police officer because, up until the moment of the engagement, he hadn’t given any thought to the question at all. But with a quasi-random dice roll compared to a reaction chart resulting in the constable being ‘Distrustful and suspicious’ (say), the GM now has a guide to playing the scene. All well and good. But why do we need the dice to tell us this? Could we not, as GMs, ‘randomly’ decide that the policeman is going to regard the character with distrust? Or earnest solicitude? Or mistake the character for a wanted man and arrest him? Clearly, we should take into account the in-game circumstances and the character’s stated actions and words. But, after all that, what is the difference between a modified dice roll deciding the outcome and the opaque mental process by which the GM opts for a particular result? In one case, tumbling maths rocks decide the matter; in the other, the GM (with due regard to all the circumstances) just… makes it up and arrives at a decision consistent with the fiction with no greater justification than the former method.
Does this latter course offer anything in the way of concrete advantages? Perhaps. I might be tempted to argue that, in spending a few seconds considering the ‘what’ of an NPC reaction a GM could also be giving some consideration (perhaps even sub-consciously) to the ‘why’ of it. However, given the previous discussion about the need for thinking about the ‘why’ even when using a reaction table, this is, at best, a marginal gain. It can also avoid the obvious drawback of getting an outlier result on a reaction table of the kind that their proponents will sometimes argue should just be disregarded. That’s fine, but does pose the pointed question about why one would use such a device in the first place if one is just going to ignore it when inconvenient?
Actually, this does sort of feed into what I think is perhaps an advantage for GMs to use their own judgment to determine reactions. It allows a GM to take into account not just all the mechanical and other circumstances of an encounter - something that a necessarily limited set of modifiers cannot do by definition - but also the feel or tone of the game and the session dynamics in the moment. If your group of PCs are trying to enter the great walled city without drawing attention to themselves, but this effort is just a prelude to what everyone at the table wants to be the focus of the action (what they do once inside), then making use of a prescriptive reaction table when the characters encounter the city watch is setting the table up for a possible derailment of those aims and expectations. Unless, of course, you just ignore any tabular result that clashes with the tenor of play... Again, one could say that, in such cases, one doesn’t make use of the reaction table and I would agree. But a GM free to form his own view of how the encounter proceeds has a much more fluid and expansive array of options bubbling and crackling away in his synapses and which can be much more finely calibrated to the moment.