Much political commentary is highly misleading. Though media evaluation of policy agendas and the conduct of politicians tends to be better, analysis of policy effects and the role of parties is very poor, even in the ‘quality’ press. When assessing policy records, commentators tend to mistake noise for the effects of policy interventions, research suggesting that performance metrics are mostly outside the control of politicians.

Why would this be? If societies are to function optimally, should they not use information which conforms tightly with reality? As cultural evolution theory emphasizes, the most efficient societies tend to outcompete alternatives, good information being crucial to this.  

In the case of political commentary, better information is available, academic research being more reliable than media analysis. The reasons why this phenomenon occurs are fascinating, demonstrating the limits of rationality. Misleading commentary boosts confidence in liberal democracy; it may be better not to challenge it.


Political commentators make assumptions about performance metrics and party differences. On performance metrics, most analysts presume that one can evaluate the effects of distinct policies. For example, comment pages are filled with assessment of the effects of government policy in areas such as crime and healthcare. Typically, these analyses will be confident, passing judgement on the efficacy of distinct policies, even in quality newspapers. In the UK, there is lively debate about the government’s levelling-up agenda – the Johnson government is trying to improve conditions in the north of England - many commentators concluding that it has been fruitless. All countries have their equivalent, analysts dissecting the effects of political interventions.

But there are good reasons for doubting such judgements. A recent study is damning. Writing in the prestigious American Political Science Review and using quasi-experimental methods, Dynes and Holbein demonstrate that voters struggle to hold governments accountable. Performance metrics are outside the control of politicians, Democratic and Republican state administrations performing equally well in areas such as the economy, education and crime. 

Admittedly, this is just one study, others coming to different conclusions. But if Dynes and Holbein are somewhat correct – and their study is at the cutting-edge of this subject - commentators are mainly evaluating noise. Certain metrics appear even more impenetrable. The performance of ministers, a staple of political commentary, is difficult for outsiders to judge, requiring acquaintance with processes which are hidden from public view.

Relatedly, the assumption that different parties achieve distinct policy outcomes is problematic. Research on partisanship demonstrates this. Certain studies emphasize partisan differences, but most find that the ideology of governing parties has indiscernible to small influence on policy, one meta-analysis establishing that the ‘average correlation between the party composition of government and policy outputs is not significantly different from zero’. Factors such as economic conditions, globalization and public opinion are more important, forcing the hands of governments.

This is counterintuitive, but becomes plausible if one reflects upon governing records. Since the Second World War, Western countries have had conservative governments which have expanded the state and liberal/socialist governments which retrenched it. These governments followed the spirit of the age, reflecting structural factors. In academia, one can attend seminars on the causes of inequality without hearing mention of parties; their role is secondary, at best.

Caveats are important. Certain metrics are more informative than others. Some acts can be judged in their own terms, such as the Johnson government’s disregard for lockdown policy or Russia’s invasion of Ukraine. And legislative agendas are far more transparent; when commentators criticize government records of passing legislation, they stand on more solid ground. Details will differ. Had the Labour Party won the 2010 UK election, a Brown government would almost certainly have implemented austerity, yet may have done so less aggressively and targeted different areas.

But one thing is clear; academic findings are very different from the assumptions of politicians and media, often radically so. Admittedly, there are good commentators, some being careful to emphasize the limits of party influence. And as I have emphasized, certain topics lend themselves to accurate commentary. But overall, we are left with a media which distorts reality. Much commentary palpably misrepresents the subject, assessing imaginary metrics and overestimating party influence. The title of this essay is provocative (!), but I stand by it!

Why is so much political commentary misleading?


In an ideal world, political commentary would better correspond to reality. What is the advantage of being misled? Aside from offending those who care about truth, this may entail conflict with social need. As theories of cultural evolution postulate, societal institutions have evolved to be efficient, reflecting competition with alternatives. Accurate information has manifold advantages, enabling humans to maximize use of resources and avoid dangers; this is true in evolutionary and modern environments.

Some commentaries reflect demand for accurate information. Newspapers such as the Wall Street Journal and Financial Times, predicated on the needs of businesses, have better analysis of politics, sometimes emphasizing the limits of parties. But this analysis is far from perfect, certain writers committing the usual mistakes.

Though incongruence between political commentary and reality has several causes, some of which I emphasize below, human agency bias underpins this phenomenon. As a species, we relish the illusion that we are in control, this providing psychological assurance. We like to think that political challenges have solutions; issues such as global warming and crime are alarming, entailing preference for narratives which emphasize human control. Relatedly, we like to think that we understand the effects of political interventions. If we are unable to measure these accurately, our ability to exercise control is threatened.

Aside from this natural sympathy for explanations which emphasize agency, people with high emotional involvement in politics have special incentives to stress agency. This category includes those who are actively and/or professionally engaged in politics. Aside from an attraction effect, those who think that politics makes a difference being attracted to such roles, politicians and activists spend inordinate time in politics. Naturally, they like to think that their actions make a difference.

Yet the high emotional involvement category is broader, including those with tribal perspectives on politics. Tribalists emphasize differences between ingroups and outgroups, using emotive language; they have always existed, but their ranks have recently swelled. Crucially, tribalists obtain emotional payoffs from emphasis on party influence, resembling sports fans. The greater one’s investment in politics, the greater the dopamine hit.

This encourages skewed analysis of relevant metrics. Famously, Democratic and Republican evaluations of the US economy sharply rise and fall, depending upon the occupant of the White House. As the saying goes, partisanship is a helluva drug. Consequently, activists emphasize partisan differences, heightening emotional stakes. And such behaviour is addictive. Depressingly, humans enjoy stigmatizing outgroups.

Systemic influences consolidate these pressures. Parties compete for votes, creating incentives for inflated promises. Alternative influences on policy outcomes, such as economic conditions and electoral systems, never run for office!

Such factors create a culture in which misleading commentary thrives, the media being a primary conduit. Many journalists may not be tribalists, but they work in a sector which responds to public demand. And accounts which emphasize agency are more engaging and easier to read. Resultingly, inaccurate commentary becomes institutionalized, writers being trained in it and editors demanding it.  The rise of social media has consolidated this, links with public demand becoming more direct.

Belief in misleading political narratives has costs. If people had more accurate beliefs about politics, they could allocate time more effectively. For example, some might conclude that work on charitable causes is better directed than political activism. Furthermore, the tone of our politics might improve; most hysterical discourse in contemporary politics depends on the verdict that the effects of opposition policies are vicious. Certain parts of tribalism may rest on bona fide ­judgements – as I have noted, we may deplore a prime minster who breaks lockdown regulations – but more realistic assessments of other metrics would probably reduce the temperature of politics.

But these pressures seem insufficient. Most of the time, people do not need to know that key metrics are spurious and party influence is limited, just as people do not need to know how televisions or cars function. Indeed, belief in politics may help maintain confidence in democracy. If people were more aware of the limits of politics, disengagement and cynicism would probably increase. These attitudes are already a problem, particularly among the less educated, and are associated with a series of negative outcomes. The cases of post-communist countries, burnt out by disengagement and cynicism, are a warning.                        


We should not expect things to change soon. This is a case in which the benefits of misinformation outweigh the costs, functional pressures underpinning it. Across the West, this culture pervades the media. Some who should know better are drawn into this. Certain academics minimize partisan influence in their research, yet emphasize it in their public utterances; one suspects that this reflects peer pressure. For those who care about accuracy, the phenomenon is grating.

But is this phenomenon very corrosive? A thought experiment will be informative. If I had the power to disabuse people of misleading beliefs about politics, would I do so? I do not think that I would. Despite problems associated with this phenomenon, tribalism threatening liberal democracy, the costs appear too high. Were people to learn that politics has much less potential than is commonly imagined, cynicism would burgeon. This attitude is already pervasive, threatening the foundations of democracies; our systems do not need more.

Some will baulk at this defence of misleading analysis. The rationalist community emphasizes the importance of truthful discourse, arguing that this maximizes individual and social wellbeing. This evokes the new atheist argument that religion is worth opposing whatever its social utility.

The case of climate science is more concrete. In this field, there is debate about how research on climate change should be presented to the public. Many worry that sceptics will seize on ambiguity; as with all fields, climate science has plenty of this.

But climate science is distinct from the case of political science, public debate being more consistent with the findings of researchers. Moreover, climate change may be the gravest challenge facing humanity. Contrastingly, political commentary has lower stakes and is more reflective of popular prejudices. And if this is what people want, why disabuse them?      


New Comment
5 comments, sorted by Click to highlight new comments since: Today at 7:43 PM

If people were more aware of the limits of politics, disengagement and cynicism would probably increase. These attitudes are already a problem, particularly among the less educated, and are associated with a series of negative outcomes.

this is a particularly interesting statement to me. on the one hand, the bulk of your post is about illuminating the limits of politics, and you mention academics and such admitting to these limits. hence, “awareness of the limits of politics” is supposed to be a highly-educated view. then you illustrate the downstream effects of disengagement and cynicism, but cite these as problems among the lesser-educated — the opposite end of the education spectrum.

so which is it? is awareness of the limits of politics a good thing only when a person is “highly educated enough” to convince themselves that carrying on the illusion, and not disengaging, is a good thing? if this is true, you should be able to convince the other person of this view, and then feel safe in revealing the rest of the truth. much of this post has the vibe of “we can’t trust these fools with the truth, thus we’re going to withhold it and thereby ensure that they remain fools”… there’s a lot of hubris to that idea.

If you read the science section of the NY Times it's also highly misleading. It's full of stories about how one research team found something surprising and the reader is conned into believing that the research result is very meaningful.

Most newspapers are in the entertainment business and care about having large audiences that pay them money or watch their ads. Most of their commentary is shitty. 

Were people to learn that politics has much less potential than is commonly imagined, cynicism would burgeon.

Politics doesn't have less potential. FDR managed to create massive change. The US would likely look quite different without him. It's just that the latest presidents mostly implement deep state agendas that are pretty independent of the party that's in office. 

Dominic Cummings writes on his blog a lot about the actual potential that does exist and could be fulfilled if you had a governing coalition that would be willing to do so. It just about boring topics such as civil service appointments. 

Moreover, climate change may be the gravest challenge facing humanity.

Given AI risk (and other X-risks) that seems to be a very dubious claim that's likely misleading in the typical way political commentary is misleading. 

Can this all summarize as "the null hypothesis is often dominant, but it doesn't help any political position".   Thus, the motivations for any investigation start with the need to show that it matters, and only secondarily how it matters.  

I think that's true to some extent, but structural variables are often found to be significant influences upon policy outcomes and to a greater degree than parties. 

New to LessWrong?