Identifying bias. A Bayesian analysis of suspicious agreement between beliefs and values.

Here is a new paper of mine (12 pages) on suspicious agreement between belief and values. The idea is that if your empirical beliefs systematically support your values, then that is evidence that you arrived at those beliefs through a biased belief-forming process. This is especially so if those beliefs concern propositions which aren’t probabilistically correlated with each other, I argue.

I have previously written several blog posts on these kinds of arguments (here and here; see also mine and ClearerThinking’s political bias test) but here the analysis is more thorough.


ClearerThinking’s Fact-Checking 2.0

Cross-posted from Huffington Post. See also The End of Bullshit at the Hands of Critical Rationalism.

Debating season is in full swing, and as per usual the presidential candidates are playing fast and loose with the truth. Fact-checking sites such as PolitiFact and have had plenty of easy targets in the debates so far. For instance, in the CNN Republican debate on September 16, Fiorina made several dubious claims about the Planned Parenthood video, as did Cruz about the Iran agreement. Similarly, in the CNN Democratic debate on October 13, Sanders falsely claimed that the U.S. has “more wealth and income inequality than any other country”, whereas Chafee fudged the data on his Rhode Island record. No doubt we are going to see more of that in the rest of the presidential campaign. The fact-checkers won’t need to worry about finding easy targets.

Research shows that fact-checking actually does make a difference. Incredible as it may seem, the candidates would probably have been even more careless with the truth if it weren’t for the fact-checkers. To some extent, fact-checkers are a deterrent to politicians inclined to stretch the truth.

At the same time, the fact that falsehoods and misrepresentations of the truth are still so common shows that this deterrence effect is not particularly strong. This raises the question how we can make it stronger. Is there a way to improve on PolitiFact‘s and‘s model – Fact-Checking 2.0, if you will?

Spencer Greenberg of ClearerThinking and I have developed a tool which we hope could play that role. Greenberg has created an application to embed videos of recorded debates and then add subtitles to them. In these subtitles, I point out falsehoods and misrepresentations of the truth at the moment when the candidates make them. For instance, when Fiorina says about the Planned Parenthood video that there is “a fully formed fetus on the table, its heart beating, its legs kicking, while someone says we have to keep it alive to harvest its brain”, I write in the subtitles:


We think that reading that a candidate’s statement is false just as it is made could have quite a striking effect. It could trigger more visceral feelings among the viewers than standard fact-checking, which is published in separate articles. To over and over again read in the subtitles that what you’re being told simply isn’t true should outrage anyone who finds truth-telling an important quality.

Continue reading

Knowledge must trump ideology and populism

Translation of opinion piece published on DN Debatt, 31 May 2015

Politics must become evidence-based. Political decisions must be made on the basis of evidence, rather than on populism or ideologically motivated opinions. To step up the pressure on politicians and make sure that is actually implemented we now start the Network for Evidence-Based Policy, which is independent of the political parties, writes 18 academics and activists.

Important decisions must be made on good grounds. This obvious principle is enshrined, for example, in the Swedish Health Care Act, which states that healthcare professionals should do their work “in accordance with science and proven experience”. For, failure to do so may lead to mistakes, which in turn could lead to suffering and death.

However, even though the mistakes of a single doctor or nurse could have very harmful consequences indeed, they pale in comparison to the effects of political mistakes, which may affect hundreds of thousands of patients. It is therefore of utmost importance that political decisions related to health care are made on good grounds.

Continue reading

Gingrich vs Paul as an example of argument-checking subtitles of the presidential debates

There are now less than nine months left till the first American presidential primaries. The primary debates, as well as the debates between the Republican and the Democrat candidate, will as usual be fact-checked by various websites such as and It is hoped that the threat of being thus fact-checked makes politicians less likely to lie (there is some evidence confirming this hypothesis, though more research is needed).

As I have previously argued, there are, however, many other ways to deceive the audience besides making false or misleading claims. For instance, politicians often make logical fallacies, fail to provide argument or evidence for their claims, fail to address the question, and so on. Such errors should also be pointed out. In short, we need not argument-checking in addition to fact-checking.

Another problem with today’s fact-checks is that they’re usually published on fact-checking sites or news media which aren’t read by the majority of those who view the debates. It’s true that fact-checking is on the rise, but still it seems it could get a much wider audience, and have a much larger impact, if it was published in a more appealing format.

Specifically, the debates should be given subtitles where all fallacies, all factual errors, all failures to address the issue, and so on, are pointed out. In the short sample video below, I give an example of what it may look like. In the clip, taken from a Republican primary debate in November 2011, Ron Paul and Newt Gingrich are debating whether the Patriot Act and other legal tools intended to combat terrorism should be discontinued because they constitute an infringement on civil liberties.

Ideally, all debates leading up to the election next year should be subtitled in this way. If such subtitled videos were made available for free – e.g. on YouTube – they could become very popular, something which in turn could influence both politician and voter behavior. (See here for a general discussion on the potential significance of this kind of criticial analyses.)

Fact-checking sites are already putting a lot of work into reviewing political debates, so I don’t think the extra amount of work this would require is that much of an obstacle. Instead the real problems concern copyright. Would the networks broadcasting these debates let outside parties annotate their debates and upload them on YouTube? It seems improbable that they voluntarily would let them do this for free.

There are several possible solutions to this problem. The first is that the networks themselves argument-check them. A second possibility is that a fact-checking website, or another media outlet wanting to argument-check the debates, buy these rights from the network. A third solution is to develop a video annotation program similar to, which allows you to annotate text on any website without infringing copyright. A fouth one is to instead argument-check the transcripts of the debates (though that would be less striking, and presumably would get fewer readers).

It’s also possible to put pressure on the networks, so as to allow their debates to be argument-checked. For instance, the Commission on Presidential Debates, which arranges the debates between the Democract and the Republican candidates, could require that all presidential debates be argument-checked in this way. Another possibility is for the debating politicians themselves to demand that the debates they participate in be argument-checked. Finally, the voters could sign petitions demanding that the debates should be argument-checked.

Let us turn to the sample video. This four minute clip was uploaded on YouTube by Paul’s supporters, who clearly think he won the exchange of arguments. When we pay more attention to what is actually being said by Paul and Gingrich, it becomes clear, however, that Paul’s arguments are noticably weak. Gingrich explicitly says that he does not find surveillance intended to prevent ordinary crimes justified, something Paul blatantly ignores when he implies that Gingrich wants a policeman in every house to prevent wife-beating. Also, he gives no argument whatsoever for the crucial claim that you need not trade off liberty against security.

Gingrich fares better, but one problem is that he never explains his distinction between ordinary crimes and issues of  national security. It is clear that wife-beating belongs to the former category, whereas nuclear attacks, which Gingrich mention, belong to the latter, but where is the borderline between these two categories? Gingrich sweeps this hard question under the carpet.

It could be argued that I’m too tough on Gingrich and Paul: that it cannot be expected that they go into these details given the format and the short amount of time they have at their disposal. While there is something to that, my criticism is not only directed at Gingrich and Paul, but also at the debating format which allows, or even demands, the participants to leave big holes in their arguments, mainly because of time constraints. Thus I point out all argumentation errors, regardless of whether it would be hard to avoid them given the time constraints.

On my argument-checking criteria, Gingrich gets a 6, and Paul a 3, on a scale from 0 to 10. (Scroll down on my blog to find a number of argument-checkings and gradings of opinion pieces.)

There are four footnotes in the video, the footnote texts of which can be found below.

Paul vs Gingrich1) The claim that there has been at least 42 thwarted terrorist attacks since 9/11 is probably true.

2) It’s not easy to get independent information on the role of the Patriot Act in thwarting these terrorist attacks. This Pro Publica article from 2013 (i.e. from two years after the debate was held) claims we have little evidence on the issue.

3) McWeigh was an extreme right-wing terrorist who killed 68 people in an attack in Oklahoma City 1995.

4) The Franklin quote is:

They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.

On the other hand, John Jay wrote in the famous Federalist papers:

Among the many objects to which a wise and free people find it necessary to direct their attention, that of providing for their safety seems to be first.

One reason not to see the Founders as authorities on this issue  is that there were no nuclear weapons, or other weapons of mass destruction, in the 18th century. It could be argued that such weapons make new trade-offs between liberty and security necessary. This objection should have been discussed.

The effectiveness-alone strategy and evidence-based policy

Double-posted from Effective Altruism Forum.

Effective altruism could be said to consist of three claims. Call them Effectiveness, Altruism and EA-morality, respectively:


Effectiveness: When you’re acting altruistically, you should be effective.

Altruism: You should act altruistically a lot.

EA-morality: Global empathy, future lives are very valuable, animal lives are valuable, etc.*


Effectiveness concerns your means, whereas Altruism and EA-morality concern yourends. Now presumably it is, in general, easier to sell Effectiveness than Altruism and EA-morality, respectivelyAccepting Altruism means less money and time for yourself, whereas accepting EA-morality might mean a fundamental transformation of your moral values. Accepting Effectiveness means none of those things. Really, it seems very hard to argue against Effectiveness (though people do try to some extent, by all means).

This raises the question of whether it is wise to sell EffectivenessAltruism and EA-morality as a package deal, as it were, or whether it is better to try to sell Effectiveness alone to some people.

An obvious option is to adapt the message to the audience: to sell the package deal to those who are likely to be open to Altruism and EA-morality, but to focus on Effectiveness when dealing with people who are unlikely to accept those other two claims. You could also, of course, try to sell the Effectiveness + EA-morality-pair (which is what GiveWell is doing, I guess) or the Effectiveness + Altruism-pair if you think that your audience could buy them, given whatever circumstances they are in. In this post, I am, however, going to focus on the Effectiveness-alone alternative.

One reason why this option is attractive is that people are already carrying out a tremendous amount of work which they claim to be altruistic. For instance, most people who do political work, in an extended sense (including voters, people who participate in internet discussions, and so on), would say that this work is being carried out for altruistic reasons. They wouldn’t say that they argue for low taxes, or higher benefits, or whatever, because such policies benefit themselves. Rather, they would argue that they benefit society, are just, or some other reason which is not specifically tied to their own good.

Continue reading