As with economic cycles, so there are political cycles. Democracy is neither the apex of a pyramid nor a value on a polemic, but one among several forms of government which follow each other in potentially predictable cycles (perhaps referencing or using as a springboard Vanhanens index of democratisation, whose variables include education, occupational diversification, and degree of broadness of distribution of agricultural land). Any inherent moral value assigned to any of these forms of government depends solely and uniquely upon the relative value assigned to the individual as opposed to the state, counterpointed against societal perception of the relative equality of any given individuals ability to choose for themself.
Once a self-contained democratic unit* has exceeded a certain size, probably around ~6400 or so (the approximate level where the masses cease to have individual identities),** it becomes unstable for much the same reasons as western-style communism became unstable, except that the relevant currency is self-identity rather than products of labour. (Linked but tangential speculation: within current known patterns of general human psychology, viable communistic communities would have a similar population limit.) Following up the self-serving bias (and especially ego-defensive forces) (Carlston & Shovar, Barron, Gergen), all persons tend to seek reference groups against which to judge feedback and with which societal norms are shared. While persons with lower self-esteem will tend to rely more heavily on the feedback from their respective reference group as a way to identify themselves against the anonymity of the masses, persons with high self-esteem, requiring less in the way of self defense, will tend to place a relatively greater positive value upon their own perception and analysis than upon that of the masses (tying in with the self-discovery that the masses are not always right). A very, very few people among these last manage to develop and maintain a truly unique viewpoint. Far more, consciously or unconsciously, end up accepting to a greater or lesser extent the authority of existing leaders and/or idea and/or mass media in defining what is right, and thereby in defining what constitutes their own image.***
Once the masses have ceased to have individual identities and begin to divide into and be perceived as reference group identities (depending upon the particular demographic(s) and identity(ies) of its participants), the democracy may shift in one of three directions:
1. In the presence of strong leadership or an established leadership class with or without a dominant mass medium but in the absence of a strong sense of nationalism, the shift will be toward oligarchy/dictatorship (either overtly or drastically, such as through a military coup) and an extremely stable caste system.
2. In the additional presence of a strong sense of nationalism**** (possibly including religious identity) and/or external threat, the shift will be toward fascism and overt or covert societal eugenics. (I tie in under eugenic programmes such practices as sterilisation, encouraging high birth rates among or providing other familial incentives to a specific demographic or racial group, selective disenfranchisement based upon demographic or racial group, preimplantation genetic diagnosis and genetic abortion, in addition to those far better known from Nazi Germany: since each of these has the active potential or actual effect of selecting in favour of one demographic or against others solely on the basis of demographic or racial group.)
3. In the absence of all of the above, the shift will be toward anarchy.
The first and third seem relatively self-explanatory. The second requires closer examination.
Societal eugenics may be inextricably intertwined with democratic federation or confederation where such democracy exists in the co-presence of a strong sense of nationalism. Every government which has overtly or covertly encouraged societal eugenics programmes, without exception, has been democratically elected in the context of strong concurrent nationalism. Please note that I do not say that a democratically elected government is necessarily pro-eugenics: only that those which are or have been pro societal eugenics have been democratically elected. Interestingly, to the best of my knowledge no non-democratically elected government has ever focused upon societal eugenics (although they have undertaken relatively isolated projects: consider the nature of the Egyptian and Hawaiian royal families); rather, these tend toward evolving extremely stable caste systems which strongly resist all subsequent attempts to remove them.
Democracy demands an inherent faith in ones fellow human being, every last one of them, that they are inherently neither less nor more capable of choosing what is best for the whole than oneself. Where that faith is lacking, democracy indeed may implicitly demand a pro-eugenics viewpoint: for it, uniquely of all governmental structures, attempts to place power in the hands of the people generally -- and thus it implicitly requires that the people generally be capable of wielding such power and dealing with the results of those choices responsibly.
Factors affecting the rate of cycling
Under systems granting de facto relative autonomy (perhaps due to lag of communication), its effect is to keep reducing the size of self-contained democratic units to below the instability level. Asymptotic limit: lack of readily accessible frontiers.
Self-contained democratic unit size is necessarily constrained by the limits of communicative technology. Regardless of formal definitions, a de facto political unit size can be no greater than allows for an effective response to stimulus time. What defines an effective response to stimulus varies based upon the urgency of the required response, the frequency and manner of existing two-way communication, and the nature and degree of the stimulus. Asymptotic limit: effectively instantaneous communication.
Near autonomous internal political units serve to diffuse and delay the destabilising effect of population pressure. Asymptotic limit: treat as individual countries or see Co-dominant national groups below.
Factors influencing the direction of cycling
Co-dominant national groups
Such governments are rarely stable (although they can work and work well: q.v. pre-invasion Lebanon). More often, however, they split into two or more separate political units; or else tensions build eventually to erupt in rebellion or civil war: which in turn results in a single self-contained political unit with one dominant national group and all other national groups weakened or eradicated. Depending upon the specific manner of resolution of such conflict, the persistent perception of threat and/or serious economic downswing may be invoked.
Ethnic or state religion
In the presence of a threat to a state or near-statal religion (whether openly acknowledged or not), the political cycle will tend to swing toward fascism. Although the Judeo-Christo-Islamic monotheisms seem to be somewhat more inclined to this pattern (and particularly the evangelical versions), other forms of religion are not immune (q.v. Hindu nationalism in India).
Where the leader or leadership class is a figurehead, the political cycle will tend to swing toward oligarchy/dictatorship (sometimes initially masquerading as a meritocracy) even in the presence of mild to moderate nationalism: but will in general tend toward greater stability than most. These also tend to be the types of societies which spawn the sharpest political humour.
Persistent perception of threat
In the presence of nationalism, the political cycle will tend to swing toward fascism with a strong statal military-economic component. In the combined absence of a strong sense of nationalism and an adequate military, anarchy may result instead.
Serious economic downswing
A sharp and persistent drop in the economy of the self-contained democratic unit will tend to raise the levels of racism and/or nationalism, thus increasing the likelihood of a swing toward fascism.
* Although no country is truly self-contained in this day and age, I use the term self-contained demographic unit loosely to indicate an autonomous state capable of determining its own policies without outside intervention.
** On average, sociological surveys suggest that most persons seem most comfortable with (knows) between 75-150 other people. Mesopotamian villages have been estimated as holding 150-200 people (Oates), Hutterites split their communities in two when they reach 150 people (the average is just over a hundred) (Mange & Mange), research specialties in the sciences only very rarely number more than 200 people (Price & Beaver), academic network sizes (number of people whose publications are paid close attention to) tend to average out at 200 (Becher), etc. Most interestingly in the context of optimising social unit cohesion and efficiency, basic military units throughout the ages tend to average out at ~150 people (range 100-200). Forge speculates that while ties of kinship and affinity suffice for social cohesion in units smaller than 150, larger units require hierarchical structuring. Allowing some leeway for duplicate knowings, the maximal size recommended in early democratic theorising (blanking -- I think this was Tocqueville along with a few others and based in a Greek precedent but I cannot track it, but the numbers ranged from ~6000-7500) would also be the maximum for individual representation of spontaneously socially cohesive units such that the representatives themselves also formed a spontaneously socially cohesive unit: ie. the maximal number within which everybody knows everybody or is likely to have a friend in common.
*** As is being currently researched in the leading edge discipline of neuromarketing, the final effect of subtle influences may be far greater when it remains unrecognised or actively denied. Generally, I argue that those with low self-esteem (or, I would suggest, a truncated, amputated, or otherwise inadequately developed self, ie where significant social aspects of the self, instead of forming in interaction with other human beings directly, have instead been substituted from a young age by the constant input of various mass media) are far more likely to directly incorporate authority (of the reference group, of a charismatic leader or idea, of a presented image) into their own self-image. Note how this factor affects the proposed possible evolution of democracy, especially in the lowest parts of economic cycles. (Another relevant correlation ties increases both in nationalism and racist movements with economic falls.)
Additionally, there is a question of whether or not there can be any meaningful distinction between the individual and their society (reference groups) (Lacan, Althusser, Hall, and in some interpretations Marx: indeed the entire concept of communism evolves naturally from optimising a blurred self-society. Giddens, in contrast, pieces together a newly re-created self in much the same way as Eliots Wasteland pieces together a unique poem -- but can it stand in complete isolation?).
See also the various discussions of self-image and self-esteem (in the context of humanist psychology) in the communication, cultural & media studies database linked through the Through the Looking Glass banner.
**** I find it useful here to reference Virgils three desirable national qualities: labour, fatum (mission/destiny: the nations raison dtre), and pietas (piety juxtaposed with patriotism in the context of family: a respect for the contract of an enduring social order and always, always, duty).