First:

I'd do away with the karma terminology. The rain, after all, falls upon the "good" and "evil" alike - and so too does the fallout! And, that's why "no good deed goes unpunished" - speaking of which

The Fallout 3 practice of sending mercs after player characters, whose karma has strayed too far from neutral, could be upgraded quite meaningfully. If a faction, which the player doesn't overtly or directly attack, is aware of a player's attacks on factions it is either loosely allied with or meagrely friendly towards, then it would make sense if that faction sent mercenaries to deal with the player character instead of getting directly involved. This would be especially meaningful in the context of inter-faction relationships which are to evolve into full-blown alliances later in the game. This isn't going to be pulled off effectively if intelligence (e.g. a contract on the corpse of a careless or arrogant merc) can't be used by the player to influence the state of more delicate inter-faction relationships. This also has to be contextualised within the outlook of the faction in question. If they think that the other faction has done something to "deserve" the player's hostility, it may make sense for them to turn a blind eye depending on their inter-faction policies and whether they would perceive and pursue an opportunity to deepen their relationship with the faction in question. On the other hand, a tyrannical or solipsistic faction may not care if the player character has a justification and may consider any action not handed down by the faction administration as punishable. These kind of responsive dynamics only really work when there is the option for the player to pursue a cessation of covert hostilities, either by negotiating an agreement with the faction administration or by completely wiping out a faction which can, generally, be achieved if all of the faction's administration can be taken out before the faction has a chance to promote enough replacements for a clear chain of command to remain in place. Once a faction no longer has a clear chain of command, only three things can happen to that faction:

1. If the faction has sufficient concentration of members in one communication network to continue functioning without schism, the faction policies and outlook will change (either favourably or unfavourably and that is a coin-flip). In this case, alliances will also shift to reflect the change in the faction's "attitude"


2. If insufficient faction members in communication with one another remain, the faction will disperse as other "best-fit" factions take on its function for its members.


3. If there is more than one concentration of members as per item 1 (above) but with insufficient communication between the concentrations, the faction will split - usually into extremists and moderates and this is further facilitated by the fact that when people feel that they no longer fit on one "side" of a social grouping, they will generally migrate to the side which offers the better fit. Sometimes, even, this will happen when doing so divides their own family very bitterly - which is just to say how compelling this social mechanism really is.

In item 1 (above) the coin-flip is obviated if the player can be offered the option of gathering intel on faction leaders and aspirants to faction leadership. By, for example, taking the extremist aspirants out of the picture prior to successfully decapitating the faction, the remaining moderate aspirants (to leadership) may not feel so friendly about what has happened but they will still, nonetheless, carry their moderate stance into their new roles in the faction's replacement leadership. Likewise, one would expect the same dynamic to work with extremism. Moreover, it would seem to be an effective stratagem to "guide" a weaker faction into extremism in order to destabilise a broader, more powerful alliance of factions. As in the real world, killing need not be the only way to decapitate a faction. I think it would be a very effective dynamic, especially for players who relish a very difficult challenge, if targeted faction leaders could be led into what their faction perceives as dishonour, then those leaders will be expected to fall on their swords, whether literally or figuratively. This could be achieved with a system of passive radiant quests i.e. radiant quests not announced to the player but emerging from gameplay as a means to ensure a variety of opportunities are randomly generated which fit the nature and character of the faction in its current state (which would have to change when its leadership changes to make sense). These random opportunities need not be obvious alternatives to simply killing off targets of interest because the point of the exercise is to raise the bar on cerebral difficulty as opposed to point and click action.

Second:

I'd fix the broken crime stats:

1. Dead people can't own property so gear ownership needs to be set to be automatically cleared of ownership on the death of the owner (e.g. Covenant) or the crime stats make no sense.


2. Crime stats are only really meaningful if limited by jurisdiction. A theft committed against one of the BOS does not count as a theft against the Minutemen or the Institute as neither faction have any authority within the organisational or social jurisdiction of the BOS. It's important to keep in mind, on this point, that in lawless territories, jurisdiction is defined by faction membership, not borders. So, crime stats in a Fallout world need to be arranged and organised by faction if they are to be meaningful.


3. Friendly fire is manslaughter, not murder. If friendly fire killings were listed as such, they could be used meaningfully with respect to faction standings. Now, maybe a player can use friendly fire as cover for a murder but complaining because the game can't read the player's mind is a bit like asking why Napoleon didn't call in an air strike at Waterloo.

Third:

Faction Standings:

1. Crime stats need to include, (in braces beside each figure or in a separate column) the number of times the type of crime was brought to the attention of the faction in question. After all, if you take out a dozen raiders and none escape and there are no raider allies present as witnesses, how can the action affect standing with the raiders?


2. Bear in mind that, from the most extreme perspectives of survivalism (e.g. Eugenics), murder is a "good" deed because it weeds out the "weak" so a survivalist faction (like the raiders) might be more inclined to befriend someone who frequently kills off their members if the group is extreme enough and true to its point of view.


3. In order for all this to be meaningful, the player has to be able to distinguish how many of each type of crime is committed against each faction irrespective of how much or how little the factions in question are aware of the player's crimes against them. This is very important because it is the best plausible way to simulate the simplified output arising from the complex balancing relationships between conscience, allegiance and rationalization.

Finally, I wouldn't bother second-guessing things beyond first tier inferences, as pure logic is pure subjectivity. If you've any doubts about this, you'll want to read some of Immanuel Kant's work to understand the factual basis for a statement like this. Owing to the increasingly subjective nature of increasingly tiered logic and the tendency of human beings to attempt anolyses which go beyond the performance limitations of logic, no player will ever agree, in totality, with any judgement system you devise. Everyone has a unique and constantly developing "code of conduct" or "way of doing things" based on a unique and ongoing experience of life and this "way of making room for others" is characterised by the individual's priorities and made less accurate by commonly overextended theorising (i.e. proposing the next hypothesis before confirming the current or even previous hypothesis). Due to the imponderable scope of resulting interpretations, as they apply to any individual's social experience, there is no second-guessing the totality of "a person's ethics" even if, in life, we can generally "avoid doing things which really tick other people off" by "extending them the same consideration we would like them to extend to us". And, yes, I really did use six completely different phrases to refer to the same thing Confucius, was able to sum all this up in one word. When confronted by one of his students and challenged to do so, his response was simply, "Shoo!". Accidental and, perhaps, sentimental onomatopoeia aside, the closest translation in English is either "reciprocity" or "empathy", depending on which of the many English dialects you speak.

On this point, while judgements are perceived as an essential element to some people (one temperament in particular), being seen as "passing judgement" is utterly offensive to people in one temperament I'm aware of (possibly more than one - I'd need some more data to be certain on this question). And, by "offended" I don't mean the politically contrived, "Oh, I'm just so offended because you used a word with a Germanic root instead of a Latin root to refer to something private and what would the people think if I didn't say I was offended?" kind of fake offence. I mean real, unadvlterated, sincere, honest, truthful, actually heart-rending "I feel deep traumatic emotional pain related to being singled out and persecuted by what you just said/did" kind of offence. By contrast with all this high melodrama (both contrived and real), a major advantage of the faction-based crime stats is that it can communicate, in considerable detail, the amount of harm done to any given faction without being potentially offensive in this way. Moreover, I think there just might be one or two of us who'd like to keep score of faction-specific actions (body count, assaults, thefts, etc) when a faction does something that really ticks us off - and a faction-specific set of crime stats would really achieve this effectively while providing a set of figures with tremendous potential to deepen the relationships existing between the various factions and the player character.

What would be truly superb, if this system were implemented, would be the option for the player to specify, at any time, their character's attitude towards each faction in very simple terms such as "Friendly", "Neutral" and "Hostile". Then, in the case of factions where the player indicates that the player character is "hostile" towards them, the language of the crime stats is altered to reflect the language used, in character, to describe those actions. So, in this case of player character hostility towards a given faction, "murders" would be replaced by "executions", "assaults" would be replaced by "arrests", "thefts" would be replaced by "confiscations", etc. to actually reflect the player character's attitude and consequent state of mind towards the faction in question. The faction in question doesn't know this attitude because that would involve mind reading which I don't think would fit the genre without raising some kind of psychological objection and consequent breaking of immersion. But the faction in question could still have its own attitude to the player character adjusted according to those crimes it knows about and would have reason to care about.