Please, don't flame me! ;-)
Though there is a specific context for this post, I prefer not talking about it. For once, context seems to matter less! ;-)
Flame wars (FWs) are those personal confrontations which happen so frequently online. FWs are seen as the bane of the online world. I don't find them particularly appealing myself. Some FWs have been at the centre of the collapse of some online communities. FWs may even be related to some people's fears of communicating online (or offline!).
There's a wealth of literature on FWs. This post is mainly based on my experience on (literally hundreds of) mailing-lists, forums, discussion boards, and blogs since 1993. I did read some of the research on FWs but this post is more about my own thinking.
Though it will probably sound more general than it should be, it's based on something similar to an ethnography of online communication. As such, I don't think so much on direct causalities but on different patterns, linking FWs with other dimensions of the culture of online groups.
Ostensibly, FWs come from breakdowns in communication. Moments in which communication ceases to work properly. Note that the notion that communication is a direct transmission of a signal is a very schematic model and that I tend to prefer models which take into account diverse goals of diverse participants as well as inter-subjectivity. Authors that have influenced my thinking about those models include Gadamer, Hymes, Jakobson, Goffman, Sperber, and Molino. (Luckily, all of these authors are easy to find by their last names! Unfortunately, all of these names refer to male speakers of European languages...)
Communication breakdowns (CBs) happen in a variety of contexts and seem to be related to a large variety of factors. Differences in communication norms are quite common, even in contexts which seem to be fairly homogeneous in terms of "communities of communication" (or "speech communities"). According to some, there are speech communities in which gender differences imply such discrepancies in communication norms, causing the "You Just Don't Understand!" principle. Quite often, a communication event will break down when the goals and expectations of different participants clash on the very possibility of communicating ("We just can't be having this conversation!"). In my experience, rarely does CB happen when people simply disagree on a specific topic. There are many online groups in which it is quite common to take disagreement "the wrong way," and get angry because of what appears to be much of a challenge. Though such a perspective on disagreement may contribute to communication breakdowns, my observation is that disagreement alone doesn't cause CB. Though the term "misunderstanding" («malentendu», «quiproquo») may seem to apply to any CB, it could also be used more specifically to refer to the (very frequent) cases in which discrepancies in the way specific utterances are understood. The whole "this is not what I meant by my use of the word 'banana' in this post on electrical conductivity!" and other (funny to the outsider) examples of miscommunication.
In my experience, CBs are more the norm than the exception, in many contexts. Especially in verbal-intensive contexts like discussions among colleagues or fans of different teams. Quite clearly to me, online communication is also verbal-intensive and a talkative (garrulous?) guy like me takes to online communication like a fish to water.
Come to think of it, it's really an extraordinary event (literally!) when two people fully understand each other, in a conversation. I mean, when each of them really groks what the other is saying. On average, people probably get compatible understandings of the communication content, but the kind of "merging of horizons" characterising true inter-subjectivity is quite uncommon, I think. Notice that I'm not talking about people agreeing with each other. As you probably notice, people often misunderstand each other more when they strive to make sure that they agree on everything. In fact, such a "conflict avoidance" attitude toward communication is quite common in certain speech communities while it's ridiculed by members of other speech communities (some people probably can think of examples... :-D). Some communication scientists probably disagree with me on this matter (especially if they apply a strict Shannon-Weaver view of communication or if they hold McLuhan's view too dearly). But, in the speech communities to which I belong most directly, disagreement is highly valued. ;-)
If miscommunication is so common, it's difficult to think of CB as the "root cause" of FWs. As so many people have been saying, since the explosion in online communication in the early 1990s, written language can be especially inefficient at transmitting "tone" and other important features of a person's communicative intention. Online communication is mostly written but attempts to fulfill some of the same goals as oral communication. Instant Messaging (IM) and other systems of synchronous, typed communication constitute an excellent set of examples for the oral-like character of online communication. They also constitute a domain in which communication norms may differ greatly. Usually based on comparative age (most IMers are relatively young, which may cause a "generation gap") and not, as far as I know, based on gender (i.e., younger women and younger men seem to hold fairly similar norms of communication in IM contexts). More interesting to me than the tired tirade about the "poor quality" of IM language is the fact that IMers appear quite efficient at transmitting more than just information through a rather limited medium.
So, now, how do FWs get started? Is it just that older people don't know how to communicate efficiently? Don't younger people have FWs? Aren't FWs caused by (other) people's inability to understand simple concepts? ;-)
To me, FWs happen mostly in difficulties in recuperating from CBs. When a CB happens in face-to-face communication, there are well-known (and somewhat efficient) methods of preventing an outright confrontation. In some speech communities, much of those methods centre on "saving face." At least, if we are to agree with Brown and Levinson. Whatever the method, preventing confrontation is often easy enough a task that we don't even notice it. Even in offline written communication, many speech communities have well-established norms (including genre-specific textual structures) which make confrontation-avoidance an easier task than it can be online. To me, it wouldn't be unfair to say that part of the issue with FWs is that specific strategies to defuse conflict are not shared very widely. Some would probably say that this lack of standardisation came with the democratisation of writing (in Euro-America, a larger proportion of the population writes regularly than was the case in the era of scribes). Not sure about that. Given the insistence of some to maintain online the rules of "étiquette" which were deemed appropriate for epistolary writing in the tradition they know best, I simply assume that there are people who think online writing had a negative impact when people forgot the "absolutely minimal" rules of étiquette.
What happens online is quite complex, in my humble opinion. Part of the failure to recover from CB may relate to the negotiation of identity. Without going so much into labeling theory, there's something to be said about the importance of the perception by others in the construction of an online persona. Since online communication is often set in the context of relatively amorphous social networks, negotiation of identity is particularly important in those cases. Typical of Durkheimian anomie, many online networks refrain from giving specific roles to most of the individual members of the network (although some individuals may have institutionalised roles in some networks). One might even say that the raison d'être for many an online community is in fact this identity negotiation. There might be no direct relationship between an online persona and social identity in (offline) daily life, but the freedom of negotiating one's identity is part of the allure of several online groups, especially those targeted towards younger people.
In a context of constant identity negotiation, face-saving (and recovering from face threatening acts) may seem scary, especially when relative anonymity isn't preserved. To those who "live online" ("netizens") losing face in online communication can be very detrimental indeed. "Netizens" do hide behind nicknames and avatars but when these are linked to a netizen's primary online identity, the stakes of face management are quite high. Given the association between online communication and speech communities which give prominence to face (and even prestige) as well as the notion of communication as information transmission, it is unsurprising to see such a pattern.
In my personal experience as a netizen, FWs are quite easy to avoid when everyone remains relatively detached from the communication event. The norms with which I tend to live (online or offline) have a lot to do with a strategy of "not taking things too personal." Sure, I can get hurt on occasion, especially when I think I hurt someone else. But, on average, I assume that the reasons people get angry has little to do with my sense of self. Not that I have no responsibility in CBs and other FW-related events. But I sincerely believe (and would be somewhat unwilling to be proven wrong) that taking something as a personal attack is the most efficient method to getting involved in a FW. As I want toavoid FWs as much as possible, my strategy can be measured for efficiency. No idea what the usual average is for most people but given the very large number of online discussions in which I have participated in the last fourteen years, I feel that I have been involved in relatively few FWs. Maybe I'm just lucky. Maybe I'm just oblivious to the FWs I cause. Maybe I'm just naïve. But I live happily, online and offline.