European democracies are experiencing declining levels of popular support for institutions and processes of representative democracy, while support for the participatory - and populist - model of democracy is on the rise. The academic debate on populism and direct democracy mainly revolves around policy output failures: e.g. the role of the socio-cultural `left-behind' vs. the economic disaffected. More recently, researchers have turned to political representation, and therefore to `input' failures in representative democracies. Is support for direct democracy a thermostatic response to a) failures to represent the vox populi; b) technocratic overreach? And does this vary by policy issue salience/politicization? We use the European Election Study 2019 to compare the effects of various explanatory models in the literature on democratic attitudes and support, and compare them to indirect measures of political dissatisfaction, pillars of our argument. We then run a large survey experiment via YouGov, where we refine the measurement of political dissatisfaction by using the sociotropic ideological congruence construct, and where we directly test the thermostatic explanation of declining support for liberal representative democracy by experimentally exposing respondents to a technocratic discourse. We find that support for popular decision-making is a thermostatic reaction to representation failures and technocratic discourse, but only in the case of salient and `easy' policy issues.