In my recently published paper that I linked to yesterday, I attribute the fall of Kerista primarily to their adoption of value of purity. I want to clarify a few points that I perhaps did not fully explain in the paper.
Every community is a complicated mix of personalities, institutions, customs, habits, and is embedded in a complex social and physical environment. Kerista was no exception. The most glaringly obvious feature of Kerista was that by the 1980s, Jud, the founder of the commune, had become a significant destructive force. Every few months, Jud would confront a target, usually a man from the Purple Submarine BFIC (to which he and I both belonged), and just hammer them until he left, or everyone was exhausted from the confrontation. It's clear that Jud was, by nature, kind of a dick. All right, a pretty big dick. If Jud hadn't been a big dick, the commune would definitely have lasted longer.
On the other hand, most leaders are dicks. A leader, by definition, is someone who is able to get followers to do what they wouldn't have done had they not been following that leader, i.e. to do what they naturally wouldn't have done. Everyone likes to think that followers and leaders alike are motivated by the mutual interest in the success of the organization. In this view, the leader just serves as kind of a "focus" for that mutual interest. People are not so much obeying the leader as obeying their own self-interest, which is furthered by the interest of the organization. (I.e. if the company does well, we all get paid and have nice stuff.) The leader just makes mutually beneficial behavior possible, and suppresses only internal cheating and free riding, which undermine mutual benefit. (See my perseveration on the drisoner's dilemma.)
But I have spent enough time as a follower and manager to know that the most effective leaders are those who go beyond this model, who can effectively use negative incentives (e.g. the desire to not get yelled at) to get more performance from even those among their followers who are acting on their own accord for the mutual benefit of the organization. An effective manager will find ways to get a good employees to work more and more intensely than they would were they motivated only by the expectation of mutual benefit. This feature of leadership is what makes it enormously difficult to be a leader. It's not enough to just yell at your employees or followers all the time; relying almost exclusively on negative incentives is worse than never using negative incentives at all. Followers can't just fear the leader; they have to respect and admire them too. And not only is there a fine line between too much and too little negative incentive, there's a dialectical relationship between the leader and the followers: they both affect each other, and how incentives play out in the organization.
This "crack the whip" leadership is deeply problematic at the analytical level. On the one hand, the leader's use of negative incentives is morally indefensible: it consists of getting followers to do things that really do not sufficiently benefit the followers, at least not materially. On the other hand, because motivation by negative reinforcement is possible, it happens, and organizations and institutions with effective leaders. i.e. those who can effectively use negative incentives, will out-compete those with less effective leaders. Regardless of any moral qualms about negative incentives, we would have to make deep changes in human psychology to eliminate them, and attempting to make deep changes in personal and social psychology, however morally justified those changes might be, usually results in catastrophe.
In one sense, Kerista fell because of the entrepreneur's dilemma. The entrepreneur's dilemma is well known to anyone who has worked in a lot of small businesses that intend to grow into large businesses (and has probably been discussed at length in the academic and popular literature): the leadership qualities that make an effective leader of a small business are very different from those that make an effective leader of a large business. To grow from a small business to a large business requires either a leader who can encompass both leadership styles, a leader who can relinquish control when the business becomes large enough, or an environment that facilitates ousting the small business leader. If none of those things happen, the small business will inevitably collapse when it gets too large. (Some businesses just stay small; however, if the leader wants it to grow large, he or she will keep trying to grow it, and become frustrated and irrational when it fails to grow, which will cause its collapse.)
Jud was manifestly effective at building a 20-30 person commune and keeping it together for about 20 years. When the commune became large enough and sufficiently economically successful, the qualities that made Jud an effective leader at the small scale made him a counter-productive leader at the larger level. Since Jud had considerable legitimacy as a leader, the culture of the commune could not easily replace his leadership. There's also the economic issue: the commune started really disbanding, I think, just when the computer business started to decline. People will tolerate a lot when they're working hard and making money; take away that economic success, and they get more critical of their social environment.
There are also other factors. Anecdotal evidence suggests that most people in the commune did not actually like polyfidelity, and did not like communal living. This dislike is difficult for me to understand because I really did like those aspects of Kerista. I don't think I'm particularly special; I think, perhaps, that there are many people like me, who would find communal polyfidelity itself rewarding and satisfying. But all of the people in Kerista (myself included) ended up in economically individualistic monogamous relationships even after Jud was ousted. No organization, culture, or institution can last long if it is at odds with the fundamental desires of its members. Without Jud more or less making people practice communal polyfidelity, the conflict between the members' desire for economic individualism and preferential dyadic relationships doomed the ideal of communal polyfidelity, and the commune justly disbanded.
All of these factors definitely contributed to the fall of Kerista. They may even be more important factors than the one I focused on in my paper: the toxic value of purity. However, I think my analysis of this value in the commune is important, for a number of reasons I'll discuss later.