Mary McAuley’s Soviet Politics: 1917-1991 is one of those rare works that marries concision with intellectual heft. Though only 123 pages in length, every page sparkles with insight. Though not a tour de force in the traditional sense, it manages to say something noteworthy about nearly every aspect of Soviet political history.
Many of the passages have wider application than the Soviet experience. McAuley’s theory for why Communist regimes are prone to develop personality cults around their leaders is particularly interesting:
This brings us to the question of leadership. In both the Soviet Union and other Communist Party regimes we notice a tendency not merely for a single leader to emerge (despite attempts at collective leadership) but for a severe personalization of leadership, including the creation of a cult around the leader. (This only happened retrospectively with Lenin, which is something very different.) Stalin, Mao, Ceaușescu, Kim Il-Sung, even the aging Honecker and Brezhnev, with more or less success, enjoyed a cult. The phenomenon cannot be explained in terms of a country’s tradition or culture since it occurs in societies without any such tradition. Its roots, we suggest, lie in the concept and practice of the Vanguard party. The party’s right to rule, it was claimed, lay in its ability to find the best, true, way forward for society. As we saw, however, its leaders disagreed on what this was, thereby threatening the party’s very basis of authority, and, as they sought desperately to reach agreement, the political tactics they employed, curbing both discussion and elections, affected relationships within the party, and eventually the system of leadership itself.
The system of appointments, began during the civil war, continued. It was not just one of appointing Bolsheviks to state offices, but also of using appointments rather than elections to posts within the party. Instead of a party organization electing a secretary, a higher party body would make an appointment to the post. The result was the emergence of patronage networks associated with individual political leaders. If a party’s rank and file is not involved in deciding policy positions, and in electing its leading officials, the nature of its relationship to the leadership changes. The ban on factions meant that first the Left, then the Right, Opposition could not, or would not, organize a following at grassroots level. Unable to identify, in any meaningful way, with a particular political stance, members could do little else but identify with a particular individual (who became a patron, appointing his loyal followers to office) or follow the leadership uncritically. In turn political leaders needed a loyal following, prepared to vote without question when the occasion arose. The reward they could offer for loyalty was promotion.
The 1920s, then, saw the emergence of groups within the party identified with particular individual, the development of what are called patron-client relations. At the top of the Party the holding of an office which gave an individual the power to make appointments influenced his ability to get the necessary backing for policy decisions in key bodies—the Central Committee and the Party Congress. An individual who was amassing this power more successfully than anyone else was Stalin, in charge of the Party Secretariat….
However, A system of patron-client relations provides an insecure basis for power. Followers will desert the leader who cannot provide rewards or guarantee their safety, or if another can offer more. Somehow the leader must convince everyone that he, and only he, is the rightful leader. What can he do? First, he must weaken or remove any rivals. Second, he must obtain a personal commitment to himself, the individual, and substitute a belief in his infallibility for mercenary loyalty. When his followers die on the battlefield, or in a labor camp, with his name on their lips the leader has authority indeed. But was this not meant to lie with the party? Given, however, the inability of a collective leadership defined one correct way forward, the task had devolved onto the leader, who in turn jealously guarded his position. But if the leader has a task of discovering the true, the only way forward, he must be superhuman. If identifying the scientific laws that govern the creation in progress a socialism depends upon one man, he must be an extraordinary individual indeed.
McAuley’s explanation has two parts, one describing the problem in terms of patron-client relations, the second in terms of party legitimacy. McAuley’s description of the patron-client side of the relationship could be augmented by the “signaling” function of dictator worship. Public sycophancy is a natural outcome of dictatorship; the more important the autocrat is to one’s personal safety and advancement, the stronger the incentive to signal one’s loyalty to him as ostentatiously as possible. There is a Red Queen dynamic to these displays. No one wants to the first man to stop clapping. 
That might explains how dictatorship works, but it has more difficulty explaining why dictatorship is. The reasons for personalized autocracy will differ, I think, from system to system, but for ‘vanguard’ party states — and I am tempted to include in this category many parties that never identified as Marxist—McAuley’s explanation has something going for it. But I have caveats. McAuley’s conclusion that the conveyor belt of dictatorship is driven by party members themselves, who must turn to client-fealty because a vanguard party cannot brook ideological faction, seems a tad off. It is not uncommon—or at least it is as common as personality cults themselves—for individuals to be associated with certain ideological programs and for party members to support them for these reasons. This was true for Trotsky, Bukharan, Stalin, et. al. in the 1920s, and for Deng Xiaoping, Hu Yaobang, Chen Yun et. al. in the 1980s. This style of governance is just as ‘usual’ a product of Leninist systems as rule by god-dictators.
The move to personality cults, be they the totalizing cults of Stalin or Mao, or the milder cult that now lifts up Xi Jinping, seems to coincide with attempts to re-engineer existing arrangements. Deng Xiaoping and Chen Yun could play their see-saw game partially because the horrors of the Mao cult were still quite fresh, but also because Deng’s reforms were (though transformative in their total impact) gradual and piecemeal in nature. To this it might be added that the Cultural Revolution had mangled existing party institutions so badly that “vested interests” that might stop reform had been destroyed.  In this environment, internal debate did not threaten the party’s claim to legitimacy.
Contrast this with Stalin’s decision to launch the Soviet Union on his crash-course program of centralized industrialization. His new policy stood in stark contradiction to the party line the CPSU adhered to in the years previous. The New Economic Policy had created powerful interests that were hostile to his new path. These interests had burrowed deep inside the CPSU itself. Mao’s story is similar. He began his rule with an attempt to reshape Chinese society from its roots; during the Cultural Revolution he sought to tear apart the CPC itself.
While the chasm that divides Xi Jinping from the consensus politics of his predecessors is not nearly as large as that which faced Stalin or Mao, Xi was also selected to clean house. He sees himself as a helmsman charting China’s journey into a “new era.” A new era implies a break with what had come before. It means a confrontation with the interests that have benefited most from the old path. This explains, in part, why he believes a personality cult is necessary for his purposes.
When a vanguard party commits itself to drastic changes in policy, dissent and debate are destructive. This is not true in eras of collective leadership, where faith is put in the party state as a process. The leadership will light upon what is best for their country by dint of their theoretical acumen and collective experience. But this presupposes a fairly narrow range of policy options. When great ruptures are made with past practice and corrupt interests inside the party itself must be confronted, old divides suddenly yawn too wide. The party cannot be a trusted to be a vanguard, for it is the party itself that needs reform. But who then can be trusted? Only that supremely (super-humanly?) far sighted leader—far sighted because he understands that cadres in his own party have become part of the problem that must be solved.
 Mary McAuley, Soviet Politics: 1917-1991 (Oxford: Oxford University Press, 1992), 46-47.
 The allusion is to a passage from the Gulag Archipelago. The story is reproduced at “Some Chilling Public Speaking History,” Manner of Speaking (12 May 2010).
 For more on that theme, see Andrew G. Walder, “Bending the Arc of Chinese History: The Cultural Revolution’s Paradoxical Legacy,” The China Quarterly 227 (September 2016): 613–31.
I think I might expand the idea in at least one aspect.
Revolutionaries (including military junta) who replace the countries elite do not have a base of support to easily fall back on. Sure the population may support them, but a general population is a difficult base to mobilize toward specific actions (as Mao found out in the Cultural Revolution).
So if you are a Argentinian Junta of many, or Pinochet, you may have people who don't like you, but you probably have a base of support within the established elite, and are not as threatened from all corners.
Whereas all communists and many of the ex-colonial junta (Gadhafi or Idi Amin for instance) either replaced the elite or the elite weren't particularly strong (having been suppressed by the colonial powers). Thus a higher level of instability, thus paranoia, and thus the move toward a "Great Man" method of leadership.
Not sure it holds across examples, but it seems to work with a comparison of the US Revolution versus the French Revolution at least in a facile.
There's a question of scale here, isn't there? Within a fairly small network it's possible to maintain personalized patron-client relationships. As the network scales up it becomes harder to assess individual clients' loyalty, or for individual clients to signal their loyalty. To a lesser extent it also becomes more challenging for the patron to deliver material benefit back to the client; there's a need for intermediary layers in the network.
Developing a package of cultist tropes is a solution to this problem; if you want to signal your loyalty, you do so by buying into the cult, and costly-signalling theory suggests this will work better the wackier the content and the more humiliating the kowtow required. Similarly, if you want to assess someone's membership in the patron-client relationship, you can take a shortcut by observing their performance of the cult tropes.
This is especially important for intermediaries as they may not themselves have regular access to the original patron, and their own importance to the patron scales with the size of their own sub-network. Cultism permits intermediaries to test the potential loyalty of recruits to the patronage network before they provide services or receive patronage, and therefore both helps the intermediary increase their own importance to the root patron and permits the network to grow faster, by transforming individual clients into recruiting intermediaries.