The standard theory of the state — in Weber's definition the institution(s) that exercise a monopoly on the legitimate use of violent coercion — has been that the state shapes its citizens. Even when the state is legitimized by more-or-less popular means (as in nominally "democratic" republics), the citizens legitimize the state to coercively shape the citizenry. We want the state to create good citizens. Democratic republics don't change this fundamental role of the state; they merely change how and to whom this fundamental role is given.
"Anarchism" (with scare quotes, very loosely defined, leaning far more towards Rothbard than Bakunin) takes almost completely the opposite view: no one should use coercion to shape anyone; the only legitimate use of coercion is in self-defense. However, coercion is not so easily dismissed, and "self-defense" (not to mention "coercion") is not so easily defined. It a depressingly obvious trope that so-called Libertarians just define whatever coercion they like (or coercion in the service of what they like) as "self-defense". Fundamentally, the only kinds of "anarchists" I've seen are those hopelessly naive, confused, deluded, or actively mendacious about their own political philosophy, or those who just label whatever kind of society they like as "anarchistic".
What is curious, though, is that there are situations where I personally want to be coerced, at least hypothetically, a notion that seems wildly counter-intuitive: if I want something, why should I want to be coerced into doing it? Why can't I just do it because I want to? The answer is signalling. I know what I want, but you cannot know directly, by observation, what I want, what's in my mind. (And, of course, vice versa.) The best you can do is draw conclusions about what is in my mind based on the evidence of my actions. But talk is cheap: even the most vicious serial killer can talk about the sanctity of life. If I want to convince you of my good intentions, I need to send you an expensive signal, one that convinces you that if I had bad intentions, the expense of the signal outweighs the satisfaction of my bad intentions. But expensive signals are expensive: what I'd like to do is send a signal that is cheap if I have good intentions, but expensive if I have bad intentions. Coercion works nicely: if I have good intentions, I do as I please, cheaply; if I fulfill bad intentions, I am subject to expensive penalties: fines, prison, ostracism, disgrace, etc.
As Hal Draper persuasively describes in Marx's Theory of Revolution*, Marx and Engels developed a theory of the state. In their theory, the conflict between individual interests and communitarian interests is the fundamental contradiction that eventually gives rise to the bourgeois state. All societies, even those before the invention and hegemony of the bourgeois state, use coercion to promulgate social norms, norms that encourage cooperation and discourage destructive kinds of competition and individual or small-group conflict. In pre-state societies, the coercion is diffused throughout the community, and largely internalized, because the benefits of cooperation in a subsistence economy are readily apparent. However, with the division of labor and increased economic productivity, the coercion necessary to maintain a more complex, interdependent society becomes concentrated in an identifiable organization, the "state", rather than being diffused through society.
*I'm about halfway through volume 1 of 3; expect more insights as I continue to read. My interpretation of Marx and Engels' theory is largely from Draper; where I mention Marx and Engels below, I am relying heavily on Draper's interpretation.
As the coercive apparatus necessary to maintain social cooperation becomes concentrated, the concentrated institution, like any other institution, naturally and inevitably develops its own interests, and uses coercion both materially and ideologically to promotes it's own interests, rather than the general interest of society. Furthermore, as the division of labor creates class differentiation, the economic ruling class, the bourgeoisie, captures considerable control over the formal state apparatus, and the bourgeoisie uses that power to maintain its rule, and maintain its domination over the working class.
It's important to note that Marx and Engels do not consider the development of the state or its capture by the bourgeoisie as some sort of conspiracy, and especially not an alien power imposed from outside society. The state in general (which precedes capitalism) and the bourgeoisie's capture of the state are the resolution of contradictions from within society, and are the result of real human beings trying to solve real problems, and keep a complex society from collapsing.
Economically, capitalism depends on diverting as much labor as practically possible to the accumulation of capital, which necessarily entails that capitalism must divert labor from consumption; given the size of the working class, even a small increase in per capita working class consumption entails a relatively large decrease in capital accumulation, it should be unsurprising that the focus of bourgeois ideological production entails (contradictorily, but ideological contradiction is much easier to swallow than material contradiction) both the valorization of work and the condemnation of poverty. Workers are good because they work, but they are bad because they are poor, and thus deserve poverty. And, furthermore, society must be organized in compliance with law, i.e. an ideal concept of justice, rather than organized by interest: the interest of the working class is fundamentally hostile to capitalism.
Except perhaps as a counter-apologetic, whether this historical course of events was good or bad (it was bad) is really immaterial. It is what happened. The question is, where do we go from here? How do we start to think about the role of coercion emerging from the concept of the bourgeois state?