Alexander Downes sees patterns:
The reasons for consistent failure are straightforward. Regime change often produces violence because it inevitably privileges some individuals or groups and alienates others. Intervening forces seek to install their preferred leadership but usually have little knowledge of the politics of the target country or of the backlash their preference is likely to engender. Moreover, interveners often lack the will or commitment to remain indefinitely in the face of violent resistance, which encourages opponents to keep fighting. Regime change generally fails to promote democracy because installing pliable dictators is in the intervener’s interest and because many target states lack the necessary preconditions for democracy.
Is it true that U.S. leaders typically don’t realize that intervention for regime change can increase the risk of civil war, that today’s puppet may not last long and subsequent ones may be less pliable, and that democracy may not flourish as a result? I doubt it. I suspect that when these considerations do not dissuade policymakers from attempting regime change, they typically go ahead for some combination of three other reasons: they don’t care, they think there are other benefits that regime change will most effectively realize, and they think the alternatives are even worse.