Scott Sumner  

Ram on overconfidence

PRINT
Noah Smith's Unpersuasive Case... Externality...

A commenter named Ram left a brilliant comment over at TheMoneyIllusion. Here it is (emphasis added):

I'd contend that the main problem in America is that the public, including its highly educated members, is social-scientifically ignorant. Most people I talk to about policy do not even realize that there is anything non-trivial about policy analysis. They want the government to make sure that four phases of rigorously designed RCTs be performed before drugs are made available to the public, for fear of unintended consequences of intervening on a complex system like the human body, yet they think they understand the consequences of highly complex interventions on human societies by introspection alone. Not only do they think they understand the consequences of alternative policy choices, but they're so confident that their understanding is right and that its truth is so obvious that the only explanation for disagreement is evil intentions. When I point out that on virtually every policy issue, at least somewhat compelling arguments for many conflicting points of view have been made by relevant experts, people usually react in disbelief or denial, or immediately retreat to questioning the motives of these experts ("of course they say that, they're on the payroll of Big Business" or whatever). These patterns of speech and behavior are uniformly distributed across the political spectrum, even if intelligence and knowledge of well-established facts is not. Even many experts in particular areas of social science evince no awareness of the lack of expert consensus on almost anything in their field, and give the impression of unanimity to an unknowing public.

My guess is that if you were to convince a supposedly non-utilitarian person that their (e.g.) deontological prescriptions might have terrible consequences, then they would revisit them. Anti-consequentialism is easy to maintain so long as you believe the consequences of your proposals are desirable, but most would fold if convinced otherwise. The real problem is convincing anyone, which involves first convincing them they don't already know the answers, which involves getting them to disassociate with their political allies enough to think critically, which involves upending a defining feature of their identity.

Imagine how different the world would look if, in order to secure the most votes, politicians had to say "you know, I don't really know the best way to deal with this problem. My educated guess, based on consultations with experts, is that we should try X, but the truth is no one knows if it will be a change for the better. Consequently we will undertake a small scale experiment, designed to maximize our understanding of the consequence of X. Should it succeed we will scale up and perform follow up studies. Should it fail we will scale it back." Seems to me we will never get there till the public becomes cognizant of and concerned about its own policy ignorance. The details of moral philosophy will be easy to reason over once we've managed that much.

Happy Thanksgiving


Comments and Sharing


CATEGORIES: Economic Philosophy




COMMENTS (28 to date)
E. Harding writes:

I don't see what's brilliant about it, especially the quoted section. I thought it was OK, but unimpressive, and by no means brilliant. This is typical complaining about contradictory preferences among common folk.

maynardGkeynes writes:

Calibration question: Did you think the Unabomber's manifesto was (a)"brilliant" or (b) a rehash of sophomoric profundities. If (a), at least you are consistent.

Bedarz Iliaci writes:

If people are wrong about

they understand the consequences of highly complex interventions on human societies by introspection alone

then either consequences, especially long-term, can not be predicted, at least by non-experts.
Now, it means that either
a) we drop worrying about long-term consequences i.e. say good bye to consequential-ism
b) or we trust that experts in social sciences do know the consequences, but that raises the question
i) Have the experts in social sciences shown that they have the requisite knowledge and have a proven track record in predicting?
ii) we trust the political and social views of these experts in social sciences. That is, we agree with the experts about what the desired consequences are.

Kevin Dick writes:

On the bit about experts seeming to be unaware of controversy, Tyler Cowen posted a link to a paper years ago showing that expert overconfidence is actually adaptive under a fairly reasonable game theoretic specification.

I blogged about it at the time--http://possibleinsight.com/2009/09/24/must-read-paper-on-overconfidence/

Procrustes writes:

I think it's a pretty good comment

After 30 years working in government policy making in Australia and the UK, I have two reflections relevant to the above.

(1) fewer than 1 in 10 of the policy issues I've worked on were thoroughly based on strong evidence (certaintly by the final decision stage) but ...

(2) I would be very worried if any of my soft left-leaning but well educated and well meaning friends got anywhere near the policy making process.

Merkwurdiglieb writes:

In other words: Politics in not about Policy.

Why one might think it should be is the strange thing.

ThomasH writes:

Good comment, but the US political system is supposed to produce just the small, incremental changes that you say and I agree that we need. To get that, we need legislators who are willing to compromise and present alternatives. Do some think the proposed minimum wage hike too high or will it lease to too much unemployment for the income it transfers to low wage workers? So propose a smaller amount or an increase in the EITC or a wage subsidy. Is the "mandate" in ACA insufficient to get a big enough risk pool, it could be made stronger or we rely more on subsidies.

The problem I see is that every issue is approached as "yes" "no" rather than "more" less."

Matt Moore writes:

Construing this as a cultural issue is not helpful. Politicians and voters are responding to institutional incentives. A change based on better understanding is unlikely and would in any case be fragile under the existing system.

Rather than lamenting the lack of sophistication in policy making, we should be considering what institutional changes might improve the problem.

Tiago writes:

Ram put it brilliantly.

Amelanchier writes:

Commits the classic error of thinking that deontologists don't care about consequences, when in fact what is distinctive about consequentialist ethics is that 1) it cares only about consequences, nothing else, and 2) it focuses on maximizing a particular consequence.

Ram writes:

Dr. Sumner: Thanks.

E. Harding: Sorry, I'll try harder next time. I wrote this from my phone during an irritating family discussion about foreign policy. Also, note that I wasn't complaining about contradictory preferences, I was complaining about ignorance and overconfidence on policy matters.

maynardGkeynes: Sick burn, bro. I wouldn't claim these observations to be original or insightful. Indeed, I don't think I need to even make an argument to convince any of you that they are correct. I just think it is useful to be reminded of them time and again, and that if their truth was (much) more widely appreciated, we might have a more constructive discourse and desirable political developments. Also, crucial context is that I was responding to Dr. Sumner's contention that "[t]he main problem in America is a lack of utilitarian thinking", which I agree is a problem, but not nearly as important as the one I described.

Bedarz Iliaci: I don't think we can drop consequentialism. Obviously it matters for our choices what consequences they bring about, and perhaps that is really all that matters for our choices. If, as I contend, we generally don't know with reliability, precision, and great accuracy, what the consequences of alternative policies are, the proper response is not to stop caring about consequences, but to be more humble, inquiring, and open-minded in our efforts to learn from our experiences, and willing to change course in response.

Kevin Dick: Interesting pointer. Thanks.

Procrustes: Thanks. Unfortunately, I think nearly everyone with their hand on a lever, even if well-meaning and well-educated, is afflicted by this mix of ignorance and overconfidence. I am, too, though I try hard to keep my perspective in check. It is hard to imagine, probably because it is such a pipe dream, how different things would be if those people behaved entirely like disinterested scientists or engineers, interested simply in solving a problem, whatever that may involve, and cognizant not only of what they know, but also the many things they do not.

Merkwurdiglieb: If you think policy is important, and politics is the only means by which to influence policy, then presumably you think politics *should* be about policy, even if, as Robin Hanson notes, it is mostly not. This is a useful perspective for understanding much political speech and behavior, but I also think it is something people can, and have, overcome from time to time, and it seems hard to imagine sustained progress in policymaking without a widespread overcoming of this tendency. I have no illusions about being able to make this happen myself, but perhaps all of us who really care about policy, and are conscious of the limits of what we (meaning all of us) really know about how the world works, can work on changing one mind at a time.

ThomasH: Great point. Policy is made on a continuum and not dichotomously. Note, though, that I do not favor incremental change for its own sake. I just think large scale changes are (1) very difficult to learn from, and (2) very difficult to undo. To me, more important than the changes being incremental is that they be designed to optimize what information can be extracted from them, so that we can make increasingly informed decisions about how to proceed. Otherwise it seems we are just swinging from one policy to configuration to another, without any guarantee of progress.

Matt Moore: There is likely an institutional incentive--I agree. But I have been able to change minds on this score (not many, and not always persistently) without modifying any institutions, so I don't think it is hopeless to start there. I do think politicians are influenced by voter mindsets, even if politics is frequently not just "the will of the people". Additionally, whatever institutional changes would encourage greater consciousness of the limits of our social scientific understanding, greater efforts to remedy it, and more humility, open-mindedness, etc., will not come about until people are at least receptive to such changes.

Tiago: Thanks. As others have pointed out, Robin Hanson and others have made these points better, and more analytically, over the years.

Thomas Sewell writes:

Unfortunately, those who take the time to read Econlog, or Robin, etc... are the ones you're more likely to be able to convince with evidence. If you're reading the comment above, you're less likely to need to hear it.

If you're nodding your head with the above, is that because it's true, or because you're suffering from overconfidence in your own understanding of the issue? Where's the evidence point? :)

Seriously though, it appears those willing to reserve judgement until the facts are in, experiments have been tried, etc... haven't been numerous enough to influence politics very much over time. For Socrates-types, the more you learn about an issue, the more you realize how complicated it is.

For much of our current "elite thinkers" in academia, the government and the media, the process seems to be more of a cognitive shortcut where for the vast majority of things you aren't an expert in, you just rely on adopting the opinions of your social circle and get rewarded. If you express doubt or ask questions designed to reveal weaknesses in the popular opinion, you are punished.

How can we make it popular to base opinions on actual evidence? Can it somehow be high status to ask tough questions calling into question if the Emperor has clothes on?

Jeff writes:

Byran Caplan has explained why this is so repeatedly. It is not individually rational for most people to inform themselves about policy because doing so does nothing to change the policies adopted. Politics is tribal, and you take positions on policies to signal your acceptability to others in the tribes you want to be part of.

While economics tells us a lot about how the world works, it does not explain everything. Economists, for example, generally don't think much about status seeking. Unlike production of goods and services, status is relative and thus a zero-sum game. We care about status because it matters a lot in at least one important arena we compete in, mating. Behaviors that are usually thought of as destructive, such as back-stabbing, sabotage and bullying, sometimes actually make sense in the mating competition.

So I'm with Bryan on this. Political posturing has more to do with status and signaling than with policy.

maynardGkeynes writes:

@ RAM -- Apologies -- unfair, inane, and tasteless comment on my part. If moderator cares to delete, please do. Please ignore/forgive.

[FYI: Once someone responds to a comment, it's not easy to remove it purely as a matter of moderation. Readers are clearly paying attention already. So, your original comment is in this case not being deleted. But rest assured, your gracious self-excoriation is equally well appreciated and well read.--Econlib Ed/Moderator.]

Ram writes:

Amelanchier: (1) What is distinctive about a deontological approach to morality is that it contends that some acts are right/wrong irrespective of their consequences. The point I was making is that this notion is easier to swallow when one believes that the consequences of what one takes to be right are reasonably benign. If convinced that the consequences are sufficiently awful, most would revisit the issue. This suggests that consequences matter a lot, even to non-consequentialists. As a result, I would argue that getting a consensus that we should care *only* about consequences, instead of *mostly* about consequences, is less important than getting a consensus about what these consequences actually are. That was the point I was making in response to Dr. Sumner's post. (2) Maximizing in the abstract is pretty empty, except that it rules out incommensurability of values. Is consequentialism incompatible with irreducible pluralism in your view?

Thomas Sewell: Fair point. I try to influence my social circle similarly, but it is usually more poorly received than in forums like this, so consider this an opportunity to vent about that fact. Also, I think that you're onto something that I didn't articulate well. If status in the political sphere was distributed according to epistemic self-awareness, then we may get somewhere.

Jeff: I agree that signaling is a big part of what is going on, though probably not the only part. Still, I'm interested in getting people to overcome this habit. Policymaking is extremely important. Arguably it is as important as anything we do as humans. If we want to improve it, we need politics to be exclusively about policy, even if it is not at present.

maynardGkeynes: No worries--no offense taken. People have said a lot worse about my political views. Many seem to think that if you respond to a controversial political issue by saying "I don't know", or even worse "no one really knows beyond a reasonable doubt", that you're morally depraved, or else have ulterior motives. I may be naive, but what I've learned most from the social sciences is how complex the social/political/economic realm really is, and how hard definitive answers are to come by.

Russ Roberts writes:

Scott,

I'm with you on this one. I think it's brilliant because it does much more than just say people are poorly informed or make systematic errors. It reminds us of one very important kind of error, overconfidence, and how to respond to it. The next time someone makes a claim for some policy position asking them for the evidence behind their position will often jar a person into realizing they're living in a cherry-picked post-hoc world of narrative fallacy and confirmation bias.

Ali Bertarian writes:

If I accept the truth of Ram's thoughts, then why would I accept the communal democracy model that is the foundation of our government? There is no property or people that our neighbors may not constitutionally control from the voting booth. (The only thing that may not be altered in the Constitution is the relative number of representatives that each state in the Senate may have.) Governmment at local, state, and federal levels now spend about 40% of everything this nation produces, so that means that 40% of everything, from the complex to the simple, is controlled by the ignorant. What good, either materially or morally, can come about from that?

Jim Glass writes:

Daniel Kahneman says that the most obvious mass cognitive failure is that while 100% of people know very little or nothing about 90+% of everything, everybody has an opinion about everything. Often strongly held.

Imagine how empty the political talk cable channels would be if it were not so.

Scott Sumner writes:

Ram, I wish I could have you do all my comment replies, the quality would increase.

Russ, Yes, I like the contrast with medical testing procedures. People are very insistent that the body is a a highly complex mechanism, and there might be all sorts of unforeseen side effects from tinkering with it. But not society?

Ali, So it's the worst system, except for all the others? Seriously, the "wisdom of crowds" hypothesis suggests that individually foolish voters can be collectively wise.

Jim Glass writes:

If I accept the truth of Ram's thoughts, then why would I accept the communal democracy model that is the foundation of our government?

You shouldn't, as it isn't.

I've seen a good number of professional historians comment about how shocked they were to discover how anti-democratic the Founders were, and how solid their reasons for being so were. Calling someone a "democrat" in those early days was actually a slur that could get one in a duel.

The Bill of Rights was popularly insisted upon in no small part explicitly to block majority rule in several directions. The bicameral legislature, Senate, electoral college, independent Supreme Court, and many other provisions of the Constitution were deliberate obstacles to democratic majority rule. One of the driving political forces behind the adoption of the constitution was the need to stop states from pursuing populist majority policies such as making debts unenforceable or wiping them out through inflation of state currencies.

The USA was founded not as a democratic state but as a liberal state. Many people today take "liberal democracy" as a two-word name for one thing, but that's a big mistake. Liberal states protect the rights of the individual. Britain evolved as a liberal state long before it became a democratic one - and that carried over into the USA federal constitution. (Democracy as we think of it evolved in the USA at the state level of politics over subsequent decades.)

OTOH, majority rule democracies without liberal institutions commonly elect, in free and fair elections, the likes of Putin, Chavez, Mugabe, Hitler, Mussolini ... Majority-rule illiberal parliamentary regimes produce such guys on a regular basis as they mobilize political majorities to eat political minorities.

For a reminder on the differences between "liberal" and "democratic" I might suggest Dani Rodrik & Sharun Mukand's paper, The Political Economy of Liberal Democracy, in short at VoxEU, of in ungated full text (pdf)

Ram writes:

Russ Roberts: Thanks. This really is common knowledge, but is also commonly ignored. Emphasizing the unknown has a way of bringing down the temperature in the room. The hard problem is how to make awareness of it last beyond the immediate conversation.

Ali Bertarian: I would extend these observations to the political process itself. Changes to the political process may have any number of unintended consequences, and may fail to have any number of intended consequences. Social science can help guide us in reforming political institutions, but whether such reforms will bring about more good than harm is almost always far from certain, and requires careful study.

Jim Glass: Nicely put. Moreover, many people's opinions are so tied up with their personal identity that any criticism of their opinions, however constructive, gets interpreted as a personal attack.

Scott Sumner: Thanks. I probably have much more free time than you, and yet I'm finding this extremely time consuming. I wonder how you manage to reply to everyone so consistently.

Jim Glass: One could evaluate the Bill of Rights similarly. There is certainly some wisdom in these founding documents that have survived for so long as the bedrock of one of the most successful countries in history, but presumably there is lots we know now that the Founders did not, and perhaps this knowledge could be used to revisit some of these ideas. The challenge, as with any other policies, is to determine whether such reforms would do more harm than good, and this is something that no political system reliably ensures.

Ali Bertarian writes:

Scott Sumner wrote, "Ali, So it's the worst system, except for all the others? Seriously, the 'wisdom of crowds' hypothesis suggests that individually foolish voters can be collectively wise."

There is a better governmental system: A legal foundation in which individual rights (and hence responsibilities) are guaranteed. We have no such non-revocable rights today that are not subject to constitutional amendment, or to Supreme Court dictates.

If you need to have brain surgery, do you want it performed by someone whose actions inside your brain are determined according to the "wisdom of the crowd" in the voting booth, or by a real doctor of your choosing? Now think of all the areas of life in which the government is involved, because everyone thinks he is an expert on how to run everyone else's life and business, and the culture encourages him to think so. Seriously, there is a good reason that you had to honestly use the word "hypothesis."

Jim Glass wrote, "The Bill of Rights was popularly insisted upon in no small part explicitly to block majority rule in several directions."

You made a fine semantic distinction from what I intended. I wrote only about the government, not of the culture of the people who form it. The Founders may have thought they put into place a democracy of people who at that time were liberal people, but that is not the case today. The people of today are not liberal, they are communalists, bred so by government employees in government schools and a leftist information media. Our children are property of the government, in government schools. We came within the vote of a single person on the Supreme Court to having our right to political speech revoked in Citizens United. The next Democrat to be President can change that.

The reason that 40% of everything the US produces is taken by the government is that the people are no longer liberal. For those who revere individual rights, the Constitution is not suitable as a guarantee of those rights in a culture of communalists.

Wim Nusselder writes:

Maybe rigourously testing interventions in complex systems in advance is not a good idea anyway, because it creates false trust?
Maybe intervention from outside in complex systems is not a good idea anyway, because trusting and going along with (and gently redirecting) endogenous change (and self-mending) is wiser?
Maybe intervention from outside in systems of sentient beings with agency is immoral, because they have an inalienable right to be convinced rather than pushed around or even nudged?

I nevertheless doubt the wisdom of perspectives that limit themselves to individualism and political programmes that focus too narrowly on individual agency & agency.
We, human beings, not only KNOW almost nothing about >90% of everything, >90% of our BEHAVIOUR is guided by social and intellectual patterns not consciously chosen, with which we go along because our consciousness and attention are limited.
That implies that those complex social systems that policies intervene in are constituted for >90% by our collective behaviour that we haven't consciously & individually chosen.
Societies are for >90% organisms rather than collections of individuals.

See my "Economics of want and greed" for further development of that ideas.

Ali Bertarian writes:

Wim Nusselder wrote: "We, human beings, not only KNOW almost nothing about >90% of everything, >90% of our BEHAVIOUR is guided by social and intellectual patterns not consciously chosen, with which we go along because our consciousness and attention are limited."

It is a non sequitur to conclude anything regarding individualism from your comment. There is no person on this planet who could possibly build completely even a simple pencil (see "I, Pencil"). That fact implies nothing about whether those people involved in either the purchase or production of that pencil should act voluntarily, or at the direction of the state.

mark e writes:

My take is two powerful forces are at work here, one ancient and one very modern.

Confirmation bias and the unlimited supply of information.

If any bias you already have can always be confirmed thru a simple Google search, because of course nowadays there is always some study somewhere that agrees with it, how can people ever be dissuaded from their closely held (and often irrational) beliefs?

Anonymous writes:

I like David Friedman's perspective on the 'worst system except for all the others' description of democracy: that if even the best way of running a government still works pretty badly, that's an argument for having governments in charge of as little as possible.

Wim Nusselder writes:

Dear Ali,

Thanks for your response.
Nothing wrong with individualism, if only it is not the only perspective that you take.
As what I wrote should have made clear: I use individualism as a description of a perspective and not as a description of a phenomenon.

Daublin writes:

Well done, Ram! Spreading this idea would do more for peace than all the Nobel Peace prizes put together. People need to innoculate themselves against the toxic intellectual atmosphere created by democratic elections, especially the American federal elections.

As a brief way to do it, I like to use a variation of your "Imagine how different the world would look" in the last paragraph. It goes like this.

Imagine how different the world would look, if a politician running for office said, "it's really important that you accurately understand the implications of these policy decisions. It's okay, I can wait--let's talk this through slowly. "

It's just inconceivable. Politicians want your vote. They don't even pretend that they want to educate you and make you smarter.

Comments for this entry have been closed
Return to top