Why I’m skeptical of “steelmanning”: By bending over backward to see things from the other person’s point of view, you’re implicitly dismissing other perspectives.

tl;dr. Giving other people the “benefit of the doubt” or “steelmanning” can improve the quality of a discussion. But this practice comes with its own risks, as it can lead to being uncharitable or “strawmanning” of other positions that are being opposed or caricatured by the people you are steelmanning.

We’re often encouraged to bending over backward to see things from the other person’s point of view, to be kind, to be charitable, to “steelman” (in the words of Chana Messinger, “addressing the best form of the other person’s argument, even if it’s not the one they presented”) rather than “strawman” (argue against your opponent’s weakest point).

And, in general, I agree. Charity and kindness are virtues, and it’s better to be charitable and kind than otherwise. I’m not saying we should always show charity and kindness—these aren’t the only virtues, and in life we need to make tradeoffs—but we certainly should consider charity and kindness when making our decisions of how to act.

Similarly with steelmanning. I can’t disagree with Conor Friedersdorf’s statement that “the best way to argue is to take on your opponents’ strongest arguments, not their weakest ones.” I agree, because I think that discussions (or “arguments”) should be positive-sum (where two people or groups express their disagreement with each other, this should be an opportunity for both sides to better understand both sides, thus moving forward), not zero-sum (where an argument has a winner and a lower), or, even worse, negative-sum (where an argument gets dirty and both sides end up looking bad).

This has come up on the blog before, when commenters have given me well-meaning advice that when criticizing some piece of research I should just keep my strongest arguments and not get distracted. And I always say that my goal in these discussions is not to “win” but to “explore”; the corollary to addressing the strongest arguments on the other side is to present all of my arguments and views, not just the ones that appear to be the most bulletproof.

To put it another way, steelmanning can lead to a virtuous circle: If I’m confident that the other side will steelman me, then I won’t have the incentive to only present my strongest case, and then between us, my adversary and myself can more effectively more forward.

But.

Two things.

First, there’s typically no clear rank ordering of what are the better arguments on a particular side and what are the worse. So steelmanning can at times be counterproductive; it can be a form of premature optimization.

Second, there can be more than two positions on an issue. Suppose that I’m addressing the argument of person X, who in turn is criticizing person Y. If I giving person X the benefit of the doubt, this can result in me being unfair to person Y. By steelmanning X, I’m strawmanning Y. In this case, I think the better case is to try to address the arguments as the arise.

We encountered such an example recently in the comments section. In the context of a discussion of a study of covid and the anti-parasitic drug ivermectin, commenter Eddie pointed to a post by the prolific psychiatrist and science blogger Scott Alexander. Alexander’s article had lots of interesting things in it, and I recommend you read the whole thing—it’s in the style of some of some of our posts here, but more interesting—but in our thread we ended up talking about a few specific passages, including this:

Sometimes these people even have a specific theory for why elites are covering up ivermectin, like that pharma companies want you to use more expensive patented drugs instead. This theory is extremely plausible.

I agree with commenter Joshua who responded:

It’s plausible that big pharma might promote expensive therapeutics and suppress information that cheap therapeutics work better. But it’s not “extremely plausible” that there’s a vast conspiracy among medical researchers (many of whom stated open to the view that ivermectin has some benefit) and public health officials to perpetrate the most significant fraud in human history, with complete disregard for the welfare of tens of millions, as has often been asserted with respect to ivermectin. That wouldn’t be an impossibility, but the level of implausibility there is what’s important.

The relevance to our discussion here is that Alexander was “steelmanning” and bending over backward to understand the position of the people pushing ivermectin, and the other people believing the claims, but in doing so he was swallowing whole a “strawman” about pharmaceutical companies and public health officials.

This passage by Alexander also came up:

Here this question is especially tough, because, uh, if you say anything in favor of ivermectin you will be cast out of civilization and thrown into the circle of social hell reserved for Klan members and 1/6 insurrectionists. All the health officials in the world will shout “horse dewormer!” at you and compare you to Josef Mengele. But good doctors aren’t supposed to care about such things. Your only goal is to save your patient. Nothing else matters.

This seems a bit off. What is “social hell,” exactly? Ted Cruz and the Fox News team say things in favor of ivermectin and also in favor of 1/6 insurrectionists—not much about the Klan, I guess—and they seem to be doing just fine. I guess that people who disagree with them on these issues might put them in “social hell,” but then that’s true of just about any hot-button issue, no? Express support for abortion to some people and they’ll put you in “social hell”; express opposition to other people and they’ll put you in “social hell.”

Also, yeah, Alexander is being flat-out insulting to “all the health officials in the world.” Health officials have considered ivermectin and all sorts of other things. If they weigh the evidence and decide that (a) there’s no good evidence in favor of certain treatments, and (b) the hype given to these treatments has been destructive in policy terms, then, yeah, they might get angry. If “your only goal is to save your patient,” then it can be completely rational to speak out against anti-vax people and even to use social sanctions if possible. If, as a health official, you’re a “good doctor” and “your only goal is to save your patient,” one of the steps you will do to save your patient is to convince him or her to take a vaccine when it is available, and one step along this way is to speak out against vaccine deniers and people who hype dubious cures.

And is it really true that “all the health officials in the world” will compare you to Josef Mengele? Of course not. I get it. Alexander is exaggerating for effect. That’s a rhetorical device I use all the time, so I’m certainly not gonna slam somebody else for doing it.

No, my point here is just what I said earlier, that in bending over backward to be fair to the proponents of ivermectin, Alexander has painted himself into the corner in which he is engaging in a ridiculous strawman, not just of opponents of ivermectin, but of the entire public health establishment. Without realizing that his clincher (“Your only goal is to save your patient”) contradicts his main argument.

Anyway, this is not about hanging a Gotcha on Scott Alexander, a blogger who generously donated many hours of his time to tracking down this ivermectin thing, something that the rest of us would be inclined to just leave to others. I appreciate his efforts! And I also appreciate his general policy of trying to be as understanding as possible of positions other than his own. It’s just kind of tricky—and I hadn’t though much about it before engaging in the above-linked comment thread. In some settings, bending over backward to see the other side’s view can entail accepting things that just aren’t so. And I think at that point you have to step back, look at what you’re saying, and see if it makes sense. I think this is a challenge whenever we do “anthropology”: to understand other perspectives without getting swallowed up by them.

P.S. This is all separate from my argument in praise of negativity in discourse. As can be seen from the quotes above, “steelmanning” can involve lots of negativity about third parties, while taking arguments directly (without “bending over backward,” “steelmanning”) can involve taking a less negative view of third parties.

102 thoughts on “Why I’m skeptical of “steelmanning”: By bending over backward to see things from the other person’s point of view, you’re implicitly dismissing other perspectives.

  1. “(in the words of Chana Messinger, ‘addressing the best form of the other person’s argument, even if it’s not the one they presented’) rather than “strawman” (argue against your opponent’s weakest point).”

    This is not a good approach if you have a strong incentive to figure something out. In fact, the opposite approach tends to work best. This is from Wikipedia’s article on “murderboarding:”

    “In highly risk-averse, technical endeavors where extreme efforts are taken to prevent mistakes (e.g. satellite operations), murder boards are used to aggressively review, without constraint or pleasantries, a situation’s problem, assumptions, constraints, mitigations, and the proposed solution. The board’s goal is to kill the well-prepared proposal on technical merit; holding back even the least suspicion of a problem is not tolerated. Such argumentative murder boards consist of many subject matter experts of the specific system under review and of all interfacing systems.”

    In particular, I would push back against the notion that criticizing the weakest part of the argument is a form of strawmanning. The weak parts show us that the argument might not be right, the strong parts cannot do that. The goal is to understand why some parts of the argument may be weak but still valid, or not.

    • To clarify: in steel manning, you don’t focus on the strong points of the argument. You would take the strongest overall argument for the position you are challenging. Then if you think there are weak points in that argument, you point them out.

      So for the murderboarding point: presumably the team putting forward a design is actually putting their best design forward, and those doing the reviewing are critiquing that “best” design. So if the reviewers are poking holes in a plan to go to the moon, they’ll be focusing on the plans for a sophisticated rocket ship, not the latest ACME product purchased by Wile E. Coyote.

      If you want to critique the conspiracy angle that Scott Alexander says is plausible: it seems implausible that there is some vast conspiracy among an amorphous group of “elites” to hide results showing ivermectin’s benefits, but it is not so implausible that Merck recommends not taking ivermectin because it can get much more money for a drug that is still under patent. In fact that might be what Scott was actually saying but just in sloppy language (I mean… that’s a long post and when one writes hyperbolically for effect sometimes language gets sloppy.) So when critiquing the conspiracy angle, you want to recognize the more plausible parts vs. the more implausible parts.

      That’s not to say you should mischaracterize what others have said in trying to steel man the arguments of your interlocutor. You should represent their arguments faithfully, but if you find yourself responding solely to strawman-like arguments, you are either not engaging in good faith or your interlocutors are actually presenting strawman arguments and you should find yourself other people to engage with. On the internet, both of those are distinct possibilities.

  2. “Strawman” isn’t used correctly here. A “strawman” argument isn’t attacking a foe’s weakest argument. It’s attacking an argument they didn’t even make. It’s a bait and switch. According to Wikipedia:

    “the real subject of the argument was not addressed or refuted, but instead replaced with a false one”

    – usually one that’s easy to refute.

    I can’t see any sensible opposite extreme to this that we could call a “steelman”.

    A good example of a “strawman” is when Progressives claim that voter ID laws “attack voting rights”. Voter ID laws do not change anything about voting rights. Anyone who has the right to vote can still vote. The only difference is that with voter ID laws is that voters must present proof – readily available proof – of who they are.

    Another good example is the frequent argument that “the science” says “climate catastrophe” is upon is. Properly understood, “the science” does not universally hold that “climate catastrophe” is inevitable or even likely. In this argument “climate catastrophe” is substituted for “climate change”.

    A more muddy example is when, for example, John Kerry claims that every published study concludes it’s cheaper to control global temperatures through emission regulations than it is to adapt to climate change. It might (or not) be true that every published study says such a thing. But that doesn’t imply that these studies have enough precision to make their conclusions worth a damn (they probably don’t). It’s not *exactly* a strawman, but it’s analogous in the sense that Kerry is (likely intentionally) misrepresenting the weight of the argument in order to support his position.

    • “A good example of a “strawman” is when Progressives claim that voter ID laws “attack voting rights”. Voter ID laws do not change anything about voting rights. Anyone who has the right to vote can still vote. The only difference is that with voter ID laws is that voters must present proof – readily available proof – of who they are.”

      Not to get too distracted on this topic, but that is not a good example of a strawman. Took consider a related issue, the 1964 Civil Rights Act and case law in the US have long acknowledged that a law can be discriminatory even if it’s facially neutral. That is, you must consider not simply the text of the law, but how its provisions can be used negatively affect protected classes. Indeed, this is the basis for striking down Jim Crow laws, which on their face treated everyone “equally”, however we designed to decrease the number of Black people who could vote. For example, the payment of a poll tax as a prerequisite for voter registration was a law that was applied “equally” but impacted Black people on average far more than white people.

      The voter ID laws are criticized on the same basis as poll taxes, which were struck down. They pose an administrative and financial burden that, while on its face is neutral, has the impact of preventing poor people from exercising their right to vote. Note: if any of these voter ID laws were paired with programs that actively sought to register every eligible voter and help voters obtain valid voter IDs for free, then the arguments that voter ID laws are not unduly burdensome would be more believable. But they’re not. And the impact is greatest on the elderly, the disabled, those who live in rural communities, and the poor.

  3. Unfortunately, in trying to explain things, Andrew unconsciously manages to tilt the playing field:

    ” Express support for abortion to some people and they’ll put you in “social hell”; express opposition to other people and they’ll put you in “social hell.”

    Instead, I claim the sentence should read, ” Express support for abortion rights to some people and they’ll put you in “social hell”; express opposition to other people and they’ll put you in “social hell.”

    No one is an advocate for abortion. The extreme right [note my adjective] has successfully framed the issue and thus, tends to win the linguistics battle. Not that long ago, a typical commentary on a social issue would begin with, “How would you like it if your daughter ….?” Wordsmiths tend to make a good living by banging the emotional drums. As another example, note that far right [once again, the loaded adjective] always employs the “dog whistle” when it refers to the Democrat Party and never to the Democratic Party.

    • I can’t believe the hypocrisy of someone who claims the “extreme right” are playing linguistic games about abortion while doing the exact same thing themselves, asserting that pro-life stances are “opposition to people” and the territory of the “extreme right”. They’re an entirely reasonable conservative position, YOU’RE just trying to push the overton window.

  4. People think this Ivermectin kind of stuff is a conspiracy because they do not appreciate the profound confusion NHST has created.

    There really is no need to invoke fraud or conspiracy. In many cases, all researchers have been doing for decades is measuring the collective opinions of themselves and their funding institutions.

    It wouldn’t surprise me if we found out ivermectin doesn’t work as a horse dewormer either.

  5. Related to Matt’s comment, and that of Anonymous.

    I think the idea of “steelmanning” is fine such as it is; there’s nothing particularly wrong with addressing someone’s strongest point. I have to admit I have a bit of a reflexively negative reaction – as I’ve mostly encountered it in what I’ll call the heterodox-o-sphere – where I’ve also encountered another, what I consider to be a more significant problem (that I’ll try to describe below)

    And I think that “charity” is fine as well, when engaging with people with whom your exploring viewpoints.

    But, IMO, “charity” is kind of orthogonal to what’s really needed, and I’d say somewhat secondary to what’s perhaps more important: Make sure that you actually understand the arguments being presented, and that you don’t mischaracterize them in such a way to elevate the validity of your own view. There is an inherent human tendency to reframe someone else’s viewpoint in a way to strengthen your sense of security about your own views. Charity is great – but it doesn’t really address that problem head on.

    IMO, when engaged in sharing differing perspectives, a key guideline for having a productive exchange is that you can express someone’s view in a way that they would agree is an accurate representation of their view. Some of this comes also relates to the difference between a “debate” and a discussion.

    But even there, you can’t really debate an issue with someone, IMO, if you haven’t accurately understood their argument – and the way to check for that is to say, simply, something like: “let me see if I’ve got this right. What you’re saying is XYZ. Did I get that right? If not, what did I get wrong. If I did get it right, is there anything else you’d like to add?” (And if they add or correct anything, you recursively go through the same process).

    That formula can feel awkward and contrived, but I think it’s extremely important, and necessarily a model for good communication.

    That said, it can get tricky. I had a discussion the other day with a friend who, I felt didn’t really understand what I was trying to say. But he said that he understood, but just disagreed with me. My feeling was that when he conveyed my argument back to me, it missed what I was trying to say, and that either he was misinterpreting what I was trying to say or I wasn’t articulating my view very clearly. But I had to allow for the possibility that either (1) my argument didn’t really make sense (or was incoherent or internally inconsistent) or (2) he really did get exactly what I was trying to say but he just disagreed (or that he was “right” and I was just resisting that).

    • Checking with others and refining your understanding of their argument is great when they are willing to engage. Steelmanning is supposed to be a best practice where they won’t engage, or are otherwise unavailable.

  6. Anoneuoid –

    > People think this Ivermectin kind of stuff is a conspiracy because they do not appreciate the profound confusion NHST has created.

    You need to distinguish between the variety of arguments.

    One argument found commonly on many high profile media platforms is that billions of people have been hypnotized into getting vaccinated – because they’re sheeple and/or incapable of independent thought – by people with an explicit and malicious intent to control the populous for personal gain (indifferent to the suffering of millions and millions). Regardless of views about the “profound confusion NHST has created,” that’s an independent view, and it necessarily involves a vast conspiracy.

    • The people claiming there would be herd immunity and/or there should vaccine passports made sense had no idea what they were talking about. There was never really any chance that would work.

      It is correct to distrust the “expertise” of anyone advocating that, since they obviously put no effort into understanding how respiratory viruses and immunity towards them works.

      And we are all still waiting for that all cause mortality data in vaccinated vs not. That is another red flag since it was higher in vaccinated during the RCTs and excess mortality has not gone down.

  7. Andrew (or whoever is managing comments)

    For some reason the captcha is no longer there when I try to comment. Then, after I press “post” I get an error message:

    >>Error: You have entered an incorrect reCAPTCHA value.
    Click the BACK button on your browser and try again.

    and when I hit the back key the comment is still there, there’s still no captcha, and when I press post again the comment goes through but the nesting fails and it goes to the bottom of the comment thread.

  8. The principle of charity is a tricky thing. When a linguist studies a unknown language, she must assume that the speakers of that language share the same world. When she sees a cat, and they say, “Wap”, she can translate “Wap” to cat only by assuming they also see the cat. When we apply the principle of charity (or steelman) to arguments, we make the other person’s arguments plausible to us, that is we assume that the other side shares certain values that we find to be basic. However, what makes the argument plausible to them may not be what makes it plausible to us. Sometimes, people express beliefs that seem crazy, racist, or fascist from our point of view because they are crazy, racist, or fascist. When applying the principle of charity, we have to ask to whom are we being charitable. Restating people’s arguments to make them more plausible to us, is distorting other people’s beliefs. I doubt that most of the people promoting the ivermectin thing did so because of the data. There is a group of people who genuinely believe that all of science, academia, the media, and the government is rigged against them. So, having experts say ivermectin doesn’t work is a reason to use it. That may seem crazy to us, but pretending that everyone has the same basic worldview is equally delusional. We need to understand that some people have fundamentally different world views from ours.

  9. I think it’s worth pointing out that Alexander comes down on the side of Ivermectin not working. For example, “Ivermectin supporters were really wrong” or “Ivermectin doesn’t reduce mortality in COVID a significant amount”

    To conclude that he fell into strawmanning the anti-Ivermectin side of things seems really strange given his conclusions are, in fact, anti-Ivermectin.

    On the larger point, I would argue that steelmanning X so hard that you overshoot and strawman Y feels like a bankshot situation. I worry far more about the first-order effect of strawmanning X (as we have all seen countless times) than the second-order effect you’re describing.

  10. “No, my point here is just what I said earlier, that in bending over backward to be fair to the proponents of ivermectin, Alexander has painted himself into the corner in which he is engaging in a ridiculous strawman, not just of opponents of ivermectin, but of the entire public health establishment. Without realizing that his clincher (“Your only goal is to save your patient”) contradicts his main argument.”

    What Alexander did wasn’t steelmanning. Creating obvious caricatures of someone’s opponents isn’t “the strongest argument for their position.” It’s probably one of the weakest.

  11. It seems to me that these arguments are, ironically, strawmanning the process of steelmanning; specifically,

    > in bending over backward to be fair to the proponents of ivermectin, Alexander has painted himself into the corner in which he is engaging in a ridiculous strawman, not just of opponents of ivermectin, but of the entire public health establishment. Without realizing that his clincher (“Your only goal is to save your patient”) contradicts his main argument.

    If a steelman of argument A leads to a strawman of argument B, then there is room to improve the steelman to deal with both arguments.

    To steelman against this, perhaps you are arguing that by invoking that one is steelmanning, some readers might misunderstand that, it is, often, only steelmanning argument A. If so, this seems to be unfair to steelmanning because it is quite difficult (perhaps provably impossible) to steelman all possible arguments within a single argument. Steelmanning is a tool to help in a dialogue among people trying to empathetically find the truth, so it should be an iterative process of steelmanning.

    • Anon:

      I don’t see how I’m strawmanning anything! I’m quoting directly from Alexander’s post, where in the act of steelmanning he’s implicitly dismissing other perspectives. I guess I agree that steelmanning doesn’t have to dismiss other perspectives—there is diversity in steelmanning as there is in everything else in human behavior—but this dismissing-of-other-perspectives does seem to me characteristic of steelmanning.

      • > I’m quoting directly from Alexander’s post, where in the act of steelmanning he’s implicitly dismissing other perspectives

        I agree with your analysis of Alexander’s post. I just find it odd to generalize some people’s misuse of steelmanning into something inherently problematic within steelmanning. People misuse statistics, too. What specifically within the process of steelmanning inherently increases the probability of implicitly dismissing other perspectives more than other argumentation methodologies? It seems — by definition — that steelmanning should do this the least.

        > this dismissing-of-other-perspectives does seem to me characteristic of steelmanning

        This is the part that surprises me. The steelmanners I’ve met are some of the most empathetic people and trying the hardest they can not to implicitly or explicitly dismiss other perspectives. What is the alternative?

        • Anon:

          You ask, what is the alternative to steelmanning. To me, the preferred alternative is to just address arguments as they are. If you also want to address alternative arguments, that’s fine too, but I don’t think there’s any reason to do this in a “steelman” framework; I think this can be just done directly.

          Regarding your experience of steelmanners being empathetic and trying the hardest they can not to dismiss other perspectives: Sure, I’ll buy that. But, again, if you try the hardest not to dismiss perspective A, that can end up resulting in strawmanning or dismissing alternative perspective B.

        • Andrew:

          I think you are missing a big part of the context here. You say that you like to “to just address arguments as they are”, which is completely reasonable, but — as I see it — the context in which strategies such as strawmanning or steelmanning etc. are most relevant is when there’s enough ambiguity to make the arguments not directly addressable.

          This kind of ambiguity might arise due to for example differences in

          1) background assumptions
          2) background knowledge
          3) language/jargon used

          or just because sometimes expressing one’s argument can be tricky for whatever reason.

          (And one of Osmo Wiio’s humorous laws of communication was that communication usually fails and if it doesn’t, that’s just due to random stroke of luck.)

          One common case, where strawmanning/steelmanning is relevant, is when metaphors are used. Usually metaphors have their limitations, after all, they are not meant as a perfect map of the problem but just as a key to gaining superficial understanding of what’s being discussed.

          In my opinion “addressing arguments as they are” does not make sense in these kinds of contexts, since it’s really more about how we as recipients of the arguments construe them from the noisy information and background assumptions that are available to us. There are no “arguments as they are” that are being exchanged in this model of noisy communication: there’s always interpretation involved, and that is always based on partial information and non-matching background information/assumptions.

        • It seems like there’s a limit of 3 nested replies.

          > if you try the hardest not to dismiss perspective A, that can end up resulting in strawmanning or dismissing alternative perspective B.

          I think that’s the strawman. Steelmanning is trying the hardest not to dismiss any perspective. The fact that Alexander strawmanned perspective B was simply a failure in his steelmanning process.

          > To me, the preferred alternative is to just address arguments as they are.

          “Just addressing arguments” is very vague and doesn’t seem to have much methodology to it. Most people don’t know how to argue. Often, they strawman, dismiss, misunderstand, get emotional, use ad hominems, logical fallacies, non sequitors, poor analogies, rhetoric, etc. What I think makes steelmanning superior is that it gives a structure to argumentation: empathy. The goal is to try as hard as possible to understand the other person’s perspective.

          I think you’re right that some people misunderstand this by not empathizing with all people at once. That’s a fair point. Steelmanning should be emphasized to empathize with all people, not just the one you’re arguing with. But this is true of “just addressing arguments” as well, so I don’t see how that’s superior.

  12. If I’m not mistaken, multiple people have already described the difficulties in having all-cause-mortality as an endpoint in an RCT. It’s a little easier to evaluate this observationally.

    Refer to the the file “Deaths occurring between 1 January 2021 and 31 January 2022 edition of this dataset” and Table 3 in that file for ACM comparisons between vaccination status (UK data).

    • There are wrong end points/outcomes used all the time. For example, mean antibody titer is used as a proxy for protection. I guess it will take 57 boosters and possible medical issues to realize that vaccines were not invented to produce a negative PCR test or high antibody titer.

  13. Anoneuoid –

    Are you old enough to know what a broken record sounds like?

    And anyway, SQUIRREL! ?

    Particular views on COVID herd immunity have pretty much zero to do with the heavily promoted conspiracy theory that billions of sheeple have been “hypnotized” into getting dangerous and ineffective vaccines while it was abundantly obvious that a perfectly safe and cheap “miracle drug” was readily available.

      • “Breaking news: Man who claims to have been ill once in the past 10 years declares miracle drugs largely don’t exist. More at 10pm”

        I dunno…localized and general anesthesia before major surgical procedures are pretty nice. Antibiotics were pretty dope, too. Suppose the discovery of insulin an helping Type-1’s live more than a year or two wasn’t too bad. If there could be a cure for message board trolls posting incredibly stupid comments, that would also be considered a panacea. Get help, little Anonyd00d. I worry about you sometimes.

        • Anesthetics are fine but don’t cure anything. So I will give you “miracle drug”, but not in the sense I meant it. As in it cures a disease. LSD has profound effects as well, but would you really call it a “miracle drug”?

          Antibiotics can also be helpful but are no “miracle drug”, people are still dying of bacterial infections all the time. And they can have serious lifelong side effects when given at the doses required to treat the infection, particularly when given to children. Eg, my tooth enamel is likely weak due to antibiotics as a child.

          Insulin is basically the same idea as a vitamin/mineral. It is a deficiency being corrected.

        • Something else I noticed is all these examples were discovered pre-1950, with only minor modifications since.

          Thats the rough cut off I use for pre vs post NHST eras. Though of course it was more a gradual replacement as people died/retired, wrote new textbooks, and so on.

          But if I had to choose where to start over (unfortunately throwing out any babies with the bathwater), it would be circa 1950. So we would still have insulin, anesthetics, vitamins, and antibiotics.

  14. 《it’s not “extremely plausible” that there’s a vast conspiracy among medical researchers (many of whom stated open to the view that ivermectin has some benefit) and public health officials to perpetrate the most significant fraud in human history, with complete disregard for the welfare of tens of millions,》

    Why am I thinking how the drug war has destroyed my life?

  15. A related concept is “cognitive empathy,” (see Robert Wright or Responsible Statecraft) or “strategic empathy.”

    (“Empathy” can have a misleading connotation there. The point is just basically that you attempt to see an issue from another person’s perspective. It doesn’t mean that you have to agree with that perspective or even think it’s logical or viable.)

    The “Scout mindset vs. Soldier mindset” (see Julia Galef) is a similar framing.

  16. > Why am I thinking how the drug war has destroyed my life?

    Do you think there’s a vast collusion among medical researchers and public health officials to perpetuate the war on drugs with compete disregard for the welfare of (hundreds?) of millions?

  17. That voter ID laws are in any way meaningfully comparable to Jim Crow poll taxes may not be a strawman but it is a false analogy. Poll taxes were confined to the South, voter ID laws exist in almost all states. But let’s for the sake of argument assume that your argument is valid. Jim Crow laws were carefully designed to disenfranchise blacks. Whose interests are served by disenfranchising in Jim Crow style “the elderly, the disabled, those who live in rural communities, and the poor”?

    • Parties appealing to the rich, drivers, healthy people, etc?

      I generally adopt the solomon law principle on this. If the republicans proposing these voter ID laws think they truly are really a simple and non-partisan measure, then fine. *Let Democrats define which IDs count as valid voter ID.*

  18. I think Andrew has got steelmannning and strawmanning wrong in their essence.

    The essence of the straw man argument is to replace the opponents argument with one that superficially looks similar but is easy to defeat, and then to trounce this dummy argument and act as if the opponents real argument were trounced. It’s an act of bad faith.

    In steelmannning, you take your opponents argument, identify it’s weakest components, replace those weakest components with the best possible arguments which make the same conclusion, and then (hopefully) attack this bolstered version of the opponents argument and show that even with all it’s obvious flaws fixed in the best possible way, the argument still fails. If the steel man is defeated, It’s an act of decisive defeat.

      • Hmmm, I think my objection here is that Alexander isn’t really steelmanning in the example you gave, he’s just accepting a really dubious argument as true. I think proper steelmanning would just avoid the whole conspiracy theory argument completely – it doesn’t make me even a little bit more likely to agree with them.

        Steelmanning would here be about trying to soften claim that into “maybe reviewers are subconsciously more harsh about this kind of drug because of their commitments to other avenues of treatment, which reverses the usual issues of publication bias” etc.

      • Maybe, but let me give a concrete example of steelmanning: https://itschancy.wordpress.com/2019/02/05/the-sev-function-just-plain-doesnt-work/

        In this post Corey takes Mayo’s Severity function seriously, he uses it to analyze a variety of experimental designs of the same type that Mayo has used as examples, he analyzes it from every possible direction he can think of, searching for a case where it works well. He uses his significantly more sophisticted mathematical ability to absolutely try his hardest to find something where Mayo’s SEV function gives a reasonable and serious answer… and he finds serious, fatal flaws with each example.

        Now if I understand your point, by steelmanning SEV he has perhaps failed to bring in what might be decisive arguments from first principles against it (ie. he has “straw-manned” the Bayesian arguments related to Cox’s theorem etc). But that’s not the purpose of the argument. the purpose of the argument is to try to make SEV work using every tool he can think of… and in the end showing that it doesn’t.

  19. Mark –

    > That voter ID laws are in any way meaningfully comparable to Jim Crow poll taxes may not be a strawman but it is a false analogy

    Did you read this part of Daniel Hawkins’ comment above?

    >> Not to get too distracted on this topic, but that is not a good example of a strawman. Took consider a related issue, the 1964 Civil Rights Act and case law in the US have long acknowledged that a law can be discriminatory even if it’s facially neutral. That is, you must consider not simply the text of the law, but how its provisions can be used negatively affect protected classes. Indeed, this is the basis for striking down Jim Crow laws, which on their face treated everyone “equally”, however we designed to decrease the number of Black people who could vote.

    Indeed – if your comment is in response to Daniel’s referencing Jim Crow voting laws above, it may be an example of the utility steel manning – or “cognitive empathy,” a “scout mindset,” or simply just taking the time to make sure that you can convey an opposing argument accurately.

    In other words, (it seems to me that) you may have missed the main point of Daniel’s use of poll taxes as an example.

  20. “in trying to replace with “the best possible arguments” for A, you can be strawmanning position B.”

    I don’t believe so. In a very real, but meta- way, attacking a weak strawman is a very poor way to argue for your position. When reasonable people hear you attack a strawman, they are generally *less* inclined to believe your argument. After all, if the best argument you put forth is an obvious strawman attack, then you don’t have a good argument to make.

    As other have pointed out, there is always ambiguity in how you interpret communication, especially with regards to metaphors, background assumptions, etc. Part of steelmanning is to be as generous as possible in auxiliary assumptions needed to interpret an argument. Steelmanning goes further and also attempts to correct any explicit, but superficial, mistakes in the original argument.

    I’d maybe make the argument that proper steelmanning has the advantage of itself being a steelmanned argument. That is, by focusis

  21. I didn’t make this part explicit, but I thought it was a fairly obvious corollary. Black people are overrepresented among the poor. They are less likely to be financially and logistically able to get voter IDs, so any laws that increase that burden will have greater impact on Black people. If those laws were paired with significant efforts to get valid IDs in the hands of all eligible voters *before* the stricter laws were enforced, that would be different. But they’re not.

  22. “The only ‘miracle drugs’ ever discovered are vitamins and minerals in the case of a deficiency.”

    Hyperbolic and off topic, this one gets murderboarded!

    Ivermectin is a miracle drug if you suffer from river blindness, it is extremely toxic to parasitic worms and nontoxic to humans at low doses. Antibiotics are miracle drugs if you are septic. The list goes on and on.

    (I’m having the same problem as Joshua with recaptcha.)

    • Ivermectin is a miracle drug if you suffer from river blindness

      First paper I found was this meta-analysis:

      The lack of convincing evidence for the effectiveness of ivermectin in the prevention of onchocercal blindness was also noted by Abiose in a narrative overview of onchocercal ocular disease and the impact of ivermectin treatment (Abiose 1998). Abiose noted that ‘in none of the studies was there any evidence of a reduction in the prevalence of blindness and a few new cases due to [onchocercal ocular disease] were observed. It may take a long time for the effect of Mectizan on the incidence of blindness to become apparent’.

      https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4425412/

      That is hardly evidence for a “miracle cure”.

  23. I like this take, that when some flawed argument is out there it’s fair to explore it from any angle and take it at face value rather than trying to imagine some optimal version of it. There are risks to misconstruing things whenever you start trying to abstract the ‘core’ from the details. This seems like a kind of postmodern view actually… don’t try to distill your opponent’s argument to its essence or idealized version, because that presupposes you have some omniscient perspective from which things can be isolated. Instead explore multiple reactions you might have.

    On the other hand distilling arguments to their core logical tenets is sort of a knee jerk reaction among a lot of scientists so it can feel natural to proceed that way. But you could distinguish how the critic or opponent thinks through the argument they don’t like (where they could learn more about their reaction by trying to construct a steelman version) and how they communicate their critique which could take any form. I do think when you know other people are watching you critique another’s ideas it’s easy to feel pressure to do something like steelmanning, lest your critique should be perceived as ‘unfair’ and discounted by the outsiders watching. But this seems like an expectation born out of the same discomfort people have with negativity in critique and not a good reason to distort the discussion from what it would be if fairness wasn’t such a preoccupation.

  24. Anoneuoid –

    > Antibiotics can also be helpful but are no “miracle drug”, people are still dying of bacterial infections all the time.

    Classic. Let’s figure out what standard you’d use for “miraculous.”

    Millions of deaths prevented yearly and life expectancies extended all over the world. Imagine the associated positive global economic impact. The vast reduction in pain and suffering.

    But sure – it could all boil down to how you define “miracle.” I mean we all die sometime, right? We could just set the minimum standard at eternal life for everyone.

    This is no doubt how you’d get to “we can’t trust public health ‘experts'” despite their contributions to doubling life expectancy in the last 150 years. If “experts” haven’t prevented all deaths everywhere they can’t be trusted.

    I love the heterodox binary mindset.

    • Classic. Let’s figure out what standard you’d use for “miraculous.”

      The typical NNT for antibiotics appears to be 5-20: https://www.thennt.com/?s=antibiotics

      This is *far superior* to most medical interventions, but still only 5-20% chance of benefiting the patient. I would put miracle closer to 100% than that. At least better than 50/50 odds. Perhaps there are other illnesses not included on that site where it meets that criteria.

      But correcting/avoiding a deficiency has close to 100% success rate. Eg, see scurvy and rickets.

  25. So you’ll just dismiss millions and millions of years of life saved, untold suffering prevented, and vast economic and qualify of life benefits by virtue of availability of antibiotics because antibiotics aren’t perfect, don’t prevent all illness, and sometimes people will recover from Illness despite not taking them or irrespective of whether they’re taken.

    Truly, heterodox binary think at its finest.

    • So you’ll just dismiss

      Why do you waste time on creating these strawmen? Especially in a thread about strawman arguments…

      This is an honest question.

  26. @Joshua: Can you at least agree that my personal priors in dealing with doctors and police and courts and jail strongly inform my reluctance to agree with most of their prescriptions?

    When my doctor told me I should quit pot because the statistical association with mental illness was too risky, or that he was afraid he’d go to jail if he prescribed pain medication for me when I had some shooting pains and such, might he just as well have been conspiring against my well-being? What’s the difference to me?

    How did cannabis go from zero tolerance to essential business during the pandemic, in this state at least? Can it be that politicians cherry-pick statistics to support a political agenda? Do most academics simply go with the flow? If superpredators are all the rage, can’t politicians manufacture a scientific consensus to support that story? Are scientists as susceptible to arbitrary, fickle social pressures (Ehrlichman admitted the war on drugs was based on lies) as anyone else?

    Given my experience, if you walked a mile in my moccasins, would you have different priors on expert opinions, mayhap?

    @somebody: So the Surgeon General was not aware of the real scientific consensus on cannabis? Why should I trust the current one?

  27. We are witnessing a large-scale breakdown in trust in our society. Increasing numbers of people are skeptical of an ever-widening range of experts.

    Count me among them. For decades, going back to at least the Vietnam war, the government has repeatedly and consistently lied about all manner of aspects of foreign affairs. Politicians consistently misrepresent what they are doing and what their opponents do, e.g. the Florida parental rights in education law neither prohibits saying “gay,” nor is there any sort of “grooming” crisis in the schools that needs to be quashed. Large corporations manufacture deadly products and cover it up rather than taking responsibility (examples too numerous to count). Drug companies have, at times, falsified or suppressed adverse data. Readers of this blog don’t need me to recount the growing problem of phony science. I could go on and on.

    So why should we trust anybody? I trust only people with whom I have personally dealt and who have a track record of honest dealings with me. Or things I can verify. Were I not an epidemiologist, with the ability to read and understand the applications that were submitted to the FDA, I would have had little faith in a previously untested technology, developed and brought to market in a remarkably short time, as an effective vaccine against covid-19. The track records of Big Pharma and the FDA don’t inspire confidence.

    The problem, of course, is that the modern world is much too complex to function on a “trust only what you can verify” basis. Nobody can possibly acquire broad enough knowledge and understanding to apply that principle to the myriad complicated decisions we all must make. When a society has low levels of trust, we waste much effort in defensive activities that are otherwise unproductive. Our ability to move forward on anything is stymied by our well-justified suspicions of where our leaders really will take us. Such a society is probably on a death spiral.

    We are desperately in need of new institutions that visibly and accountably maintain high levels of integrity as sources of guidance. It is, unfortunately, unclear whether the existing corrupt institutions can be reformed in that way. We may need to “burn it all down and start over” to get there. But that strategy has usually ended badly in the past. Today’s world is not a pretty picture, and erosion of trust is at its very center.

  28. rsm –

    I’m having some difficulty following your point through all those questions.

    Bottom line, the war on drugs obviously combines some of the absolute worst elements of public policy development, “big pharma” corruption, the prison-industrial complex, bad medical practice and tons o’ other bad shit. Even assuming that some degree of bad shit inevitably happens in ways that make affixing blame complicated, when I look at places like Scandinavia or Portugal it’s obvious that there are smarter ways to address drug-related bad shit. Our society’s failures in dealing with the scourge of addiction is truly tragic and my own family has been tragically affected. I have great empathy for anyone who has been affected and work hard on not judging people who have run into trouble.

    But all that said, I think that the plausibility of any theory that does affix blame for complex societal problems should be held to a high level of scrutiny. It’s all to human to want to feel like we can find answers by distilling such problems into simplistic narratives.

  29. Clyde –

    > We are desperately in need of new institutions that visibly and accountably maintain high levels of integrity as sources of guidance […] We may need to “burn it all down and start over” to get there.

    IMO, this largely gets the direction of causality wrong.

    As evidence on motivated reasoning shows, the “trustworthiness” of expertise is not some subjective metric, but one that is largely predicted by one’s ideological orientation. Not to excuse mistakes that have been made, but no institutions could overcome a predisposition by large segments of the public to distrust, and to sieze on even the slightest errors to leverage that distrust in pursuit of an agenda.

    There’s a chicken/egg issue in play, IMO. Burning it all down would be treating the symptom, not the disease. And I think that the idea that the fault lies with the institutions (or “experts”) themselves is a manifestation of the underlying problem. Imo, the bigger problem is the constant need to generate and feed an impose towards antipathy.

  30. Yes, there’s a chicken-and-egg issue, and there is bidirectional causality. But I think Martin Gurri largely got it right: the rise of communication technologies enabled the dispositionally skeptical to amplify their signal. (There were always people who distrusted the New York Times at its most pontifical… but lacking a channel of equal bandwidth, they could be ignored.) And the crisis comes not from the amplification of skepticism alone, but from the *reaction* of the former sources of authority, who were discovered, in case after case after case, to be unable to defend (a) their methods; (b) their leveraging of authority from, say, newsgathering (which was undoubtedly high) to authority in opinion (which was entitled to virtually no respect); and (c) perhaps worst of all, their contempt for being attacked which they were unable to hide. News media, politicians, public health authorities, economists…. all of them. And then once the blood was in the water, you get Trump and Twitter and all the rest.

  31. Anoneuoid –

    > This is an honest question

    I didn’t mean to create a strawman. Maybe I misunderstood.

    It seemed that indeed you dismissed millions and millions of years of life saved, untold suffering prevented, and vast economic and quality of life benefits, as qualifying antibiotics to be a “miracle drug.” (Seemingly because antibiotics aren’t perfect, don’t prevent all illness, and sometimes people will recover from Illness despite not taking them or irrespective of whether they’re taken).

    So if that’s a strawman, and you didn’t dismiss all those benefits as not qualifying antibiotics as a “miracle drug,” then I’ll ask you again to please explain what is your standard for a miracle drug?

    How many more years of life saved would qualify? How much more pain and suffering avoided would qualify? How much more economic and quality of life benefits would qualify?

    Those are honest questions.

    I asked once before and your answer seemed unspecific to me.

    Maybe if you’re more specific I won’t misunderstand your answer.

    • Maybe if you’re more specific I won’t misunderstand your answer.

      I was very specific. To me a miracle drug is something that cures or prevents the disease nearly 100% of the time. The only real examples of this are correcting deficiencies. Eg, vitamins/minerals and insulin in the case of diabetes.

      To you it is apparently a “miracle” if something works ~10% of the time. That is fine, you are free to have low standards. I wouldn’t think a mechanic was a miracle worker if they only fixed the car 10% of the time, even if they fixed millions of cars (while failing to fix tens of millions).

      The rest is all stuff you made up.

  32. Anoneuoid –

    > I was very specific. To me a miracle drug is something that cures or prevents the disease nearly 100% of the time.

    Ok. So then you’re not agreeing, that the benefits I listed, are qualifications for a miracle drug. Saving millions and millions of years of life? Not sufficient. Preventing untold pain and suffering? Insufficient. Vast economic and quality of life benefits? Insufficient.

    Again, it certainly seems that you’re dismissing those enormous benefits as qualifications for a miracle drug. I fail to see what the strawman is.

    But let’s break down your argument a bit more. Let’s say that there’s a bacterial infection that kills 10% of those infected. 90% of those infected would recover without antibiotics. But let’s say that antibiotics save the lives of 90% of those who wouldn’t recover otherwise (for the sake of this example we’ll assume it doesn’t speed recovery or lessen severity for anyone). And 10% of those who would die, die despite treatment with antibiotics. Not very impressive, right – according to your standard the drug isn’t a “miracle” drug because it doesn’t prevent death anywhere near 100% of the time. It has a 90% mortality reduction but ONLY prevents death 10% of the time! Clearly not a miracle.

    And of course that’s ignoring baseline risk for specific populations. Or for specific, and horrific diseases. For example, an antibiotic that process a 100% mortality benefit for anyone who gets a particular infection wouldn’t be a miracle drug if it had a low NNT for other infections (and this had a relatively low NNT across the board).

    > To you it is apparently a “miracle” if something works ~10% of the time.

    Well, first, that depends on what you mean by “works.” If you’re saying it doesn’t “work” if it doesn’t prevent a death when the death wouldn’t occur irrespective of taking the antibiotic, then you and I are operating from a different definition of “works.” If a drug provides a 90% mortality reduction despite only saving 10% of those infected, yes, I’d say that fits a description of “works.”

    > That is fine, you are free to have low standards. I wouldn’t think a mechanic was a miracle worker if they only fixed the car 10% of the time, even if they fixed millions of cars (while failing to fix tens of millions).

    I know that you don’t have much experience with cars, but I’d say that it’s pretty unusual that brakes that are failing will just get better without a repair being done. Whereas an infection for which I might take an antibiotic might very well go away if I didn’t take the antibiotics. So I think your analogy is a pretty poor one.

    It’s interesting how often I encounter what seems to me to be a enormous sense of entitlement. Imagine living 150 years ago, when your life expectancy was significantly less, due in no small part to a lack of access to antibiotics. Where the chances that an infection would kill you would have been significantly higher (even if chances were you’d recover from your infection without antibiotics).

    I dare say that most people living back then, if told that a few pills could realize those benefits for them, would consider that a miracle drug. But maybe that’s just me.

    Ok, before Andrew yells as me and Raghu complains, I’ll leave it here.

    • I referred to antibiotics as “*far superior* to most medical interventions”… you create a strawman of someone who is “dismissing those enormous benefits”.

      In a thread about strawman arguments, no less. It is just strange behaviour.

  33. Joshua said:《the plausibility of any theory that does affix blame for complex societal problems should be held to a high level of scrutiny.》

    Where was that scrutiny when pot was blamed for so many things from the 1980s to the 2010s? Why should I believe vaccines, or ivermectin, have been scrutinized any more effectively than drug policy?

    Can you see that it is obvious to me both that weed was not the public health threat officials said it was, and that vaccines are not necessary to avoid Covid? Why isn’t the best policy simply to present your arguments and let individuals make their own choices about what they put in their own bodies?

  34. Jonathan (ao)

    I don’t exactly disagree, but I think there are questions.

    > and there is bidirectional causality

    I certainly agree there’s some bidirectionality.

    > the rise of communication technologies enabled the dispositionally skeptical to amplify their signal. (There were always people who distrusted the New York Times at its most pontifical… but lacking a channel of equal bandwidth, they could be ignored.)

    I’m inclined to agree – but I don’t know how we measure this phenomenon to confirm that amplification or the likely mechanism. There could be any number of causal chains.

    It does seem that this all has a heavier grip on our more of our society now than in previous manifestations – say the skepticism of the hippies versus the establishment. But I think it’s hard to trease out just how much that’s true or whether their might be spurious signals: for example, maybe it’s not so much that there’s increased skepticism but that it has become more uniformly associated with ideological orientation. In other words, “skepticism” has become more political and politicians are capitalizing on it more explicitly.

    > And the crisis comes not from the amplification of skepticism alone, but from the *reaction* of the former sources of authority, who were discovered, in case after case after case, to be unable to defend (a) their methods;

    Maybe to some extent you’re embedding a circular causal assumption there. For example, “skeptics” say that it’s the reaction of climate scientists to criticism that amplifies their skepticism, but people on the other side of the climate change aisle say that the reactions of climate scientists reinforces their confidence in the work of climate scientists. So the view of the reaction is a function of the predisposition (and its ideological root).

    > (b) their leveraging of authority from, say, newsgathering (which was undoubtedly high) to authority in opinion (which was entitled to virtually no respect);

    I think there can be a similar circular aspect there as well. You se it as an invalid appeal to authority because you start with a skepticism about the authority. Appealing to authority isn’t NECESSARILY fallacious.

    > (c) perhaps worst of all, their contempt for being attacked which they were unable to hide. News media, politicians, public health authorities, economists…. all of them.

    Again, I question how you isolate the origin of the phenomenon. Where is the stating point?
    The chicken or the egg? If you have a different ideology, the reaction to the attack is entirely justified – in fact more or less required.

    > And then once the blood was in the water, you get Trump and Twitter and all the rest.

    Well, I think It certainly true that Trump skillfully weaponozed existing antipathy for personal and political gain – but was that really unique? People in the right accused Obama of doing that before Trump became quite such a central figure.

  35. @Joshua “There’s a chicken/egg issue in play…”

    As I think about it, chicken and egg problem is not a good analogy for the contemporary problem of distrust. Yes, there is a cyclical dynamic going on where the propensity to distrust gets stoked by parties who wish to destroy trust, which then feeds on the prevalent mendacity, which then generates more distrust, the response to which is often more mendacity, and on and on. So there is that dynamic.

    But, if we were to set ourselves a goal of driving the chicken/egg cycle into extinction, were we able to either destroy all eggs before they hatch, or exterminate all chicks before they reach sexual maturity, either way, in a short time, chickens and eggs would be gone forever. The same is not true with mendacity and distrust. Even if we had a highly successful public relations campaign to rebuild trust in our leaders and institutions, without also reforming them to make them trustworthy, the trust would soon dissipate as people uncovered the ongoing corruption of the reigning sociopaths and charlatans. By contrast, if institutions were somehow made truly trustworthy, it might be possible to restore trust in them, and as long as we remained vigilant against outbreaks of corruption, that trust could be sustained long term.

    • The basic problem is this. No one who is trustworthy would want to work for these broken institutions. You have to reform the institutions in order to reform the institutions. It’s a catch 22 if not a chicken and egg.

      • RE: “No one who is trustworthy would want to work for these broken institutions.”

        It’s a good point and probably true as a general tendency but in fact, everyone trustworthy or not has to earn to a living. I’m sure even the most broken of institutions has employees who personally find the institutional B.S. distasteful but just do the best they can within a broken system in order to remain employed.

        My own career involved doing mostly the kind of cookbook NHST modeling that is derided every day on this blog. I recall being vaguely uncomfortable with that recipe as early as graduate school and I certainly had a continual nagging sense that “there must be a better way to use this data” throughout my career. That sense became very acute once I was unfortunate enough to start paying attention to Gelman, Greenland, et. al. quite a few years ago.

        But regardless of the validity of the paradigm, I was employed to make a best-faith effort in adhering to the prevailing method of analyzing the data presented to me. In the greater sense I was part of a broken system if by “system” we mean getting academic research funded and published but I do not feel there was anything untrustworthy about myself or my colleagues. We were constrained by the gigantic, monolithic “NHST Industrial Complex” which controls the publication and funding apparatus in our field of study but within that framework we tried our very best not to cut corners, fish for p-values, cherry pick chance associations and so forth.

        I would probably look back on my career and feel less unfulfilled in a counterfactual universe where NHST, power calculations, laundry lists of unnecessary covariates and Bonferroni adjustments did not rule the roost. But doing work in that universe was never going to pay me a living wage while working in the geographical area where I’ve chosen to live my life. Does that make me untrustworthy? I think that would be a harsh assessment.

        • You personally may not be untrustworthy, for example your personal dealings with your friends and acquaintances and debtors and so forth, but by your own admission, your work probably is. So in this sense “you as scientist” were untrustworthy. (That is to say, people probably shouldn’t trust that your work reached valid conclusions)

          Also just because someone doesn’t want to work in a field doesn’t mean they don’t work in that field. But if we are looking for reform we need people who wholeheartedly want to do the work and refuse to do it using a bullshit formula. That’s a different group of people than those who simply “would have preferred” to do something else but found themselves constrained by the institution. To reform an institution you need people who reject the institutional flaws and constraints not merely wish they weren’t there.

        • RE: “To reform an institution you need people who reject the institutional flaws and constraints not merely wish they weren’t there.”

          …and who have agency to cause anyone to care about their rejection of the prevailing paradigm.

  36. Clyde –

    Sorry if this is repititious, but near as I can tell you and I have a basic disageement here.

    I think that if there is a primary problem, you see the basic origin of the problem generated by the mendacity within the institutions. (I guess this would match Daniel’s belief that no one trustworthy would work for any of them).

    I think I see more of the origin of the mendacity originating in the motivation that people have to find mendacity in the institutions, rooted in unrealistic expectations, and often in direct association with their political orientation. (And related to Dan’s view, I have known a few people I consider trustworthy who have worked at such institutions).

    > Yes, there is a cyclical dynamic going on where the propensity to distrust gets stoked by parties who wish to destroy trust…

    Here, I think the problem is deeper. Yes, there are parties who seek to stoke the distrust – but that comes on top of the underlying mechanism, where people project identity-aggressive cognition (via motivated reasoning) into scientific issues that are easily polarized.

    Maybe chicken/egg is a bad metaphor, but that’s not particular relevant to me. My point is more about the locus of the problem.

    I will note that (at least the last time I looked) contrary to common perception much of the polling related to trust in our scientific institutions shows that there’s not that much of a decline, overall. There is some decline, but it’s mostly concentrated in a particular cohort – libertarian/tea party types, and it’s mostly in association with political controversies and not generalized. There hasn’t been much change among moderate types and there’s been something of a slight *increase* in liberal types – so overall it kind of balances out. It would be interesting to check up on the more recent data (which I haven’t done). Dan Kahan did a series of related blog posts a while back. Here’s one:

    http://www.culturalcognition.net/blog/2017/3/1/mistrust-or-motivated-misperception-of-scientific-consensus.html

    You might Google for Gordon Gouchat, as he’s done a lot of work on the topic (Kahan has some posts related to Gauchat’s work, and as I recall he wrote a guest post at Dan’s).

    I don’t see much by way of cure through the idea that it’s the institutions (or the people they comprise) that primarily create the problem through their mendacity, and thus they must be fixed to fix the problem.

    I think to the extent that there’s a cure it will come if people find a way to be less motivated in how they approach polarized scientific topics, and the institutions of science. Of course, I’m not going to hold my breath waiting for that to happen but I don’t think that waiting for that to happen is any less realistic than thinking that somehow scientific institutions aren’t going to be viewed as mendacious by some folks – particularly when those institutions are engaged in public communication on issues that map into political cleavages in our society.

    Consider that this kind of distrust of expertise and institutions of science may be fairly unique to the American landscape. If you believe the locus of the problem is primarily with the institutions themselves, then do you think that American institutions science are uniquely menndacious? If not, then is the American public uniquely perceptive in identifying thst mendacity?

    > Even if we had a highly successful public relations campaign to rebuild trust in our leaders and institutions, without also reforming them to make them trustworthy, the trust would soon dissipate as people uncovered the ongoing corruption of the reigning sociopaths and charlatans

    I think this embodies more of a top down conceptualization than the one that I hold. I don’t think there’s any way to rebuild trust in our leaders and institutions if we don’t reform our tendency to filter information in such a way that it holds leaders and institutions hostage to our ideological agendas. I think the reform has to be more bottom up. We need to find a way to be less inclined to confirm our deological preferences by leveraging trust/lack of trust in institutions and leaders. Everyone’s looking to justify outrage, everyone’s looking to prove their victimhood. Everyone projects mendacity as a way to confirm our beliefs.

    I think what we need is more of a societal contract. To a large degree, it used to be that we had unifying enemies (Germany, Russia). But with the loss of those unifying targets, I think we’ve also lost some of the basic mechanisms of unification. In our society used to be that demz and pubz could reach some compromise on issues because ideological cleavages didn’t align so perfectly with party identification. Now, pubz and demz are enemies and there’s very little ideological crossover. I don’t see how our institutions and leaders can enhance trust as long as there’s so much reason to use those entities as a means to savage our enemies, which are
    kw primarily fellow Americans with whom we disagree

  37. Anonyd00d, just curious: are you autistic? Or do you have an obsessive personality disorder? Not that there’s anything wrong with that–it could just help others with their interactions toward you. I’m trying to understand where your obsessive “tics” come from (others have described this as a “broken record”).

    I’m quite sure antibiotics are quite useful to folks with chlamydia, who have undergone recent invasive surgery, and for folks with strep throat, whooping cough, and urinary tract infections. They are useful to folks who are immunocompromised who are at risk of secondary infections. The consequences of letting any of those conditions go untreated/unprotected can be quite miserable and sometimes dangerous or even deadly. Many medical and dental procedures would not be possible without them.

    Sure, the examples I gave are of drugs created pre-1950. There are plenty of miracle drugs created post 1950, too, that help out a lot of people: the drug class of immune check-point inhibitors as cancer therapy, the development of non-drowsy antihistamines such as fexofenadine restoring normal life to severe allergy sufferers, the class of monoclonal antibody medicines such as Humira/adalimumab for auto-immune disorders, sapropterin for helping folks with phenylketonuria, and so on. If I wanted to, I could list a hundred more examples.

    • Anonyd00d, just curious: are you autistic?

      You are the one who hasn’t figured out how to hit the reply button correctly when a new obstacle arose.

      Your second paragraph is more strawmen.

      The third, I will check and get back to you.

    • This is what you consider to be a miracle drug (fexofenidine)?
      https://www.ajmc.com/view/apr99-948ps235-s247

      Like Joshua, you apparently have very low standards/expectations. Anything that slightly improves the situation for a few patients is “miracle”.

      These interventions are in a qualitatively different category than vitamin c for scurvy, vitamin d for rickets, insulin for diabetes, etc.

      Also, my allergies went away when I did a low carb (high fat) diet and have not returned. From a quick search, I see many people have reported similar. So I think that would be a better comparator than placebo.

  38. Yes, thanks for supplying a reference that even further proves my point. I’d like to add there are literally tens of millions of people across the country that have quality of life improvements from taking allergy medicine. Think also of secondary effects–reduction of transmissible disease due to reduced nose-wiping, sneezing, and coughing for individuals with both seasonal allergies and asymptomatic but infectious cold/flu.

    These drugs are much more important than those for vitamins and minerals. Firstly, vitamins and minerals are not even defined as drugs. Secondly, vitamins and minerals are naturally occurring in well-balanced diets. Your thought process makes you come off as a pirate from the 1500’s–sorry, my mentally ill friend, but most of the civilized world is not at risk of scurvy or rickets. Your argument is sorta like saying protein, carbohydrates, lipids, and oxygen are super-important drugs. It’s just laughably dumb, like most of your posts. Your “quick search” means nothing. Also, I think if you’ve made a convincing argument to anyone else on this board, they would have written something by now. You’re just a sad, delusional fool.

    • Your thought process makes you come off as a pirate from the 1500’s–sorry, my mentally ill friend, but most of the civilized world is not at risk of scurvy or rickets.

      I have no idea what strawman you even made up here. If you get sufficient vitamins from food why would there be a deficiency?

      It is bizarre to interact with you and Joshua who just create a constant stream of strawmen to argue with. And niether of you can figure out the reply button either, this seems related.

      • Vitamin C is very important for someone suffering from scurvy, but not very useful if given to people who eat sufficient quantities of fruit and vegetables. The NNT for vitamin C is probably very very high in the modern world. I bring this up because NNT is a poor way to determine whether a drug is a miracle drug. Every time someone goes to the doctor and requests antibiotics for their viral upper respiratory infection it raises the NNT but it doesn’t make the antibiotic any less miraculous for someone with a festering wound or a kidney infection for example.

      • Specifically, what happens once a drug has been found is that if it doesn’t have a terrible side effect profile it will get tried for a lot of uses outside the main use. So for example if you have severe seasonal allergies, fexofenadine is a great drug that lets you get through your days instead of lying in bed sick. But it also has basically zero serious side effects so people try it for hives, where it works well, but maybe it doesn’t work well for eczema yet people try it there to see if there’s some minor improvements, and maybe it doesn’t work well with atopic dermatitis but people try it there and it has some minor effects, and maybe if you have very mild seasonal allergies then the dry mouth you get from fexofenadine is almost canceling out the mild improvement from mild allergies… so the NNT goes up.

        You’re comparing “vitamin C for scurvy” to “fexofenadine in all the uses anyone has ever thought of” but if you want to compare them you really need “vitamin C for scurvy” vs “fexofenadine for people who can’t leave their HEPA filtered bedrooms without tears streaming down their face and sneezing at about 2 Hz.”

        • Yes, I am talking about correcting deficiences. That is when we see “miracle cures”. Nothing else comes close.

          Here was my original comment, with new emphasis added:

          The only “miracle drugs” ever discovered are vitamins and minerals in the case of a deficiency.

        • But what I’m saying is that Fexofenadine *in the case where you can’t leave your room due to allergies* is also a valid “miracle” in the same way as vit C in scurvy treatment. If you’re going to call out the treatment of a specific condition with a specific drug you have to also take the best specific condition for the comparison drug.

        • Look at the Fexofenadine study I linked to, it is barely more effective than placebo (which appears very effective for some reason, I bet theres reversion to the mean because they recruit subjects who recently were suffering from worse than usual allergies)…

          So while there may be a few people who it helps greatly, it is nothing close to resolving/preventing the issues of near 100% like vitamin c for scurvy, d for rickets, thiamine for beriberi, insulin for diabetes, and so on.

          Therefore, if we want to call that a “miracle” we need a new category even better than miracle for fixing deficiencies.

        • See the paragraph around this quote (I had to type it out):

          The key issue is not the magnitude of a given placebo response, but the significance of the the difference in symptom improvement between the active treatment and placebo. This study clearly shows statistically significant improvements in QO: measures with 60 mg bid fexofenadine, compared with placebo.

          https://www.ajmc.com/view/apr99-948ps235-s247

          Figure 3 shows overall 2 week improvement from a score of ~2.75 (out of 6) to ~1.8 in placebo and ~1.7 in treatment.

          I almost couldn’t think of a better example. In the NHST era, that apparently counts as a “miracle drug”.

          How many people are finally getting fed up with their allergies due to some extreme exposure or i similar then going taking this drug for the first time? Then they attribute the entire improvement to the drug, rather than accounting for the 90% of that improvement they would get from placebo or doing nothing?

        • Then, if you missed it, here was the other claim for a recent “miracle drug”: https://statmodeling.stat.columbia.edu/2022/04/28/the-challenge-of-bending-over-backward-to-see-things-from-the-other-persons-point-of-view/#comment-2050857

          Basically there is lack of evidence that ivermectin prevents/cures river blindness. Maybe it does, maybe it doesn’t. This is nothing like what we see for scurvy/rickets/beriberi/etc. Those essentially disappear in the absence of a deficiency.

          Why do people consider these miracle drugs?

  39. Here’s another kind of strawman: The self-strawman!

    Yes, that’s right, you can create a fake argument – a strawman – for your *own* position that is more defendable than your actual argument! Our local newspaper – one of the few remaining in the region – runs a daily column about how social media – FB, TWTR, etc – are destroying “local journalism”, because newspapers are closing everywhere, and arguing for Federal financial support for “local media” (e.g., print news). Sounds really scary to local people :o!! Are they taking our news? We’ve *got* to do something!

    The strawman is this: while the local newspapers are closing, local *websites* are popping up like maggots on a corpse. The “local journalism” argument is a strawman – a fake. Local journalism is doing just fine, its newspapers that are dying. This is hardly surprising given the relatively low quality of local talent compared to national talent (which isn’t exactly stunning either). But the “killing” of “local journalism” by Evil Big Tech sounds a lot worse than just admitting that print newspapers are outdated *and* unpopular, and no one wants to buy them or advertise in them. I

    Please note that this argument relies on the strength of the fake rather than its weakness. A strawman argument is about faking, not about the strength or weakness of a particular argument. The “strawman” is a fake of a “real man”, not a weak opposite of a “steel man”

  40. Anoneuoid –

    Well, because you referenced me further….

    > Like Joshua, you apparently have very low standards/expectations

    Apparently low expectations in your book = saving millions and millions of years of life, resulting in significant increases in life expectancy all over the world, along with amassing huge economic and quality of life benefits.

    All which you dismiss as criteria for a determining whether antibiotics are a “miracle drug” because antibiotics don’t prevent illness or save plives in people who wouldn’t get infected or would recover from infection without taking the antibiotics.

    Wish you would answer the question: What # of lives saved, or economic or quality of life benefits accrued, would you consider as meriting a label of “miracle drug?”

    It seems to me that you’re engaging a rather typical and banal heterodox binary mindset – in other words, despite unprecedented in human existence, and obviously massively beneficial impact, antibiotics aren’t a “miracle drug” because they aren’t perfect (and don’t extend all life eternally).

  41. Anoneuoid –

    Obviously, a determination of what is a “miracle drug” is subjective, because it would depend on subjective determinations of what is or isn’t a “miracle.”

    Someone who receives a heart transplant might consider immunosuppressant drugs a “miracle,” as they significantly increase their chances of survival. Whereas based on another measure they don’t impact a huge % of the world’s population. So depending on how people are affected, their perspective on “miraculous” might differ.

    And this is what you don’t seem to want to engage in – the degree to which your statements integrate to the full context.

    I don’t really know what a “miracle drug” is because I think the labor is pretty vague, but yes, a significant benefit to billions of people, over decades, would seem to me to be a meaningfu metric to consider as criteria.

    So again, I ask you what degree of significant benefits enjoyed by billions of people do YOU think merits consideration for applying a label of “miraculous?”

  42. Anoneuoid –

    > The NNT for vitamin C is probably very very high in the modern world

    I raised a similar point several times and asked you to address it – but you didn’t. I was hoping you’d address it when Daniel raised it. Why haven’t you?

    You’re basically cherry-picking to make your argument.

    That antibiotics don’t save lives for many people who would recover without them doesn’t mean that they don’t have an enormous benefit to contemporary society (that people from previous eras didn’t enjoy).

    Let’s consider that many people take vitamin C with little benefit.

    There are conditions for which the relative benefits of antibiotics are very high even if the absolute benefit isn’t.

    Here:

    After pooling, antibiotic prophylaxis was found to reduce the relative and absolute risk of wound infection by 73% and 17.5%, respectively. The number needed to treat to prevent one wound infection was 5.7 (95% confidence interval = 4.4-8.0).

    https://pubmed.ncbi.nlm.nih.gov/11095330/

    That puts an NNT of 6 into the meaningful context of a relative risk benefit of 73%. Considering the implications of wound infections and the ubiquity of wounds, that’s a huge benefit, imo, irrespective of the fact that many people would recover even without taking the antibiotics.

    And in another sense, the absolute benefits of antibiotics are just enormous. They make a significant contribution to life expectancy all over the world. Why would you dismiss saving the lives of so many children, for example, as you determine what is or isn’t a “miracle drug?”

    Is there a reason that you aren’t addressing these aspects of the discussion? It’s getting hard to avoid a conclusion (since you’ve commented many times without addressing this aspect) that the reason you don’t respond on point is that you just can’t reconcile that aspect with your arguments.

  43. Yes, that’s true. If people won’t engage in good faith trying to clarify won’t work, and often people won’t engage in good faith.

  44. “For example, the payment of a poll tax as a prerequisite for voter registration was a law that was applied “equally” but impacted Black people on average far more than white people. ”

    This is another strawman argument – and interestingly as you point out one that has been effectively encoded into law by the judiciary.

    The strawman here is equating “Poll Tax” with “Identification Card”. The identification required in voting laws – a driver’s license or analogous state ID card – is widely available at virtually no cost ($9/year in my state) and is necessary anyway for a normally functioning individual in the United States and probably any Western country. For that matter state ID cars could be made available for free.

    It’s great that you brought this up though because it shows how widely “strawman” arguments are distributed and that even supposedly sophisticated thinkers like judges and law school graduates can fall for them or use them to promote their own views and interests.

  45. Anonymous –

    > The strawman here is equating “Poll Tax” with “Identification Card”.

    I didn’t read Daniel’s comment as equating the two, but as arguing that both are similar in having a greater impact on black communities. Of course, it would be legit to ask him for evidence of such.

    And just because there might be disproportionately greater impact on black communities doesn’t prove that was the intent. But even if that wasn’t the explicit intent that effect might be considered important -and as I recall, wasn’t there some evidence of at least some Republicans stating that as an explicit intent?

  46. “I didn’t read Daniel’s comment as equating the two, ”

    I read his argument as claiming the two had been equated by others, and that such an equation was justified by the doctrine of proportionality; and because it is justified by the doctrine of proportionality, my claim (that the statement “voter ID laws are undermining voting rights” is a strawman argument) is incorrect.

    It’s a lot of wind to get through, which emphasizes the tenuous nature of the connection.

    BTW, he said voter ID laws disproportionately impact the poor, the elderly and the disabled, not blacks.

    For my money, the proportionality doctrine is also underlain by a strawman: that the claim that disproportional impact is equal to injustice. It’s used as a strawman like this: A young man climbs on to a bus. He doesn’t pay. He is, in fact, stealing from the transit agency and the public. His theft is discovered and he is cited. His guilt is not in question. However, he defends himself and wins his case because he was “disproportionately targeted”. Mind you, however he was targeted, the targeting was accurate: he was, in fact, guilty. But his guilt for committing an *actual crime* is now subservient to proportionality.

    • Lucius:

      I read what you wrote, and I think your argument all comes down to your claim that Alexander in his post is doing the good steelmanning and not the bad steelmanning, which you call tinmanning.

      Alexander wrote:

      Sometimes these people even have a specific theory for why elites are covering up ivermectin, like that pharma companies want you to use more expensive patented drugs instead. This theory is extremely plausible.

      The term “elites” is vague enough to cover all sorts of people, but given the vast publicity that ivermectin received, it doesn’t seem like anyone was covering it up. My impression was that much of the ivermectin hype was all about hope, which is fine. Ivermectin doesn’t seem to have done much, but it was out there and there was lots of hope. It wasn’t being covered up, so I disagree with Alexander that this theory “is extremely plausible.” As I wrote, in steelmanning the cover-up theory, Alexander was in effect swallowing whole a “strawman” about pharmaceutical companies and public health officials.

      You write:

      By analogy, tobacco companies, medical researchers, and public health authorities caused a “significant fraud in human history, with complete disregard for the welfare of tens of millions” in regards to smoking and lung cancer plausibly without a vast conspiracy; instead, it was plausibly due to a mix of greed, regulatory capture, and a science funding system that took a while to find the truth.

      That’s a strange thing to do, to lump together “tobacco companies, medical researchers, and public health authorities” as one category. Let’s separate them:

      1. Tobacco companies really did have a “vast conspiracy” to suppress the evidence that cigarettes cause cancer. There’s lots of documentation of this, going back to the 1950s and continuing for decades. Given the huge pile of evidence—actual documents from these companies—it is not at all “plausibly without a vast conspiracy.” The conspiracy was there.

      2. Medical researchers are not a unified group. Some medical researchers were working for cigarette companies, others were working in other places. Different medical researchers focused on different things.

      3. Public health authorities, at least in the U.S., took some major steps to spread the word about cigarettes and cancer. They were fought on this for decades by the cigarette companies.

      Getting back to ivermectin, you bounce between three theories as to why “elites” (again, not clearly defined) were not advocating mass use of the drug for covid: “Elites lied about positive ivermectin studies” or the more mild “a simpler explanation of incompetence and myopia, along with greed combined with regulatory capture” or some sort of “Machiavellianism.” This seems to me to be incredibly cynical of you, and you don’t consider the third option that these studies did not, in fact, provide good evidence for the use of ivermectin for covid. It’s striking that you don’t consider the third option, given that followup studies did not find ivermectin to be effective; i.e. the skeptics seem to have been right on this one.

      I don’t know if “strawmanning” is the right term here for what you’re doing, but whatever it is, I think it reveals a strong bias on your part, that your only explanations are “lying,” “incompetence,” “myopia,” “greed,” “regulatory capture,” and “Machiavellianism.” How about “scientific judgment”? Maybe an appropriate term would be “selective steelmanning,” where you decide to think the worst of public health officials. I agree that public health officials have made many mistakes; I disagree with your stance of ruling out the possibility that they were showing sound judgment in this case to not get swayed by hype.

      You also write, “the history of scientific incompetence and regulatory capture (e.g. esketamine, tobacco, etc.).” You use the passive term “regulatory capture.” Again, with cigarettes there was the Surgeon General’s report, the cigarette companies were sitting on lots of evidence on smoking and cancer, and they did a decades-long effort to confuse people on the matter.

      Getting back to ivermectin, you just seem really committed to this “elites are lying” story, without being clear on who are the elites or without evidence that they are lying. I’d ask why don’t you strawman the elites (whoever they are) and consider that they might have been using solid scientific judgment, but this doesn’t even seem like a strawman.

      Given that this is what “steelmanning” my argument looks like, I’d prefer you just address the arguments as they arise. Your attempt to steelman my argument just led you into making one claim (that there was no conspiracy to hide the cigarette-cancer link) that was flat-out false and another claim (that skepticism of ivermectin was due to some combination of “lying,” “incompetence,” “myopia,” “greed,” “regulatory capture,” and “Machiavellianism”) that is at best extremely uncharitable and is, indeed, to my mind, ridiculous.

      To step back a moment, I think this reveals a more general problem, which is that you have to be careful about defending any method of reasoning. “Steelmanning” can be useful in some settings and counterproductive in others; it depends on context. For that matter, the post from Alexander that I linked to has lots of really good things (indeed, I wrote that it had “lots of interesting things in it, and I recommend you read the whole thing—it’s in the style of some of some of our posts here, but more interesting”); it just wasn’t perfect. I think understand why Alexander was doing the steelmanning he was doing—for one thing, he wanted to be persuasive to the proponents of ivermectin, and one way to do this is to express sympathy for various arguments they had raised. I thought it was interesting how this steelmanning didn’t come for free, one could say. To be open to all the arguments of ivermectin proponents, Alexander had to be open to some pretty bad arguments, and he didn’t go steelmanning the other sides of those arguments. Again, I can see why he did this, as this could lead to an infinite regress of steelmanning.

      So, my recommendation to you is, first to consider that you might be wrong in your belief that all those ivermectin skeptics were lying, and second to recognize that steelmanning can involve some misrepresentation of arguments on the other side. Which might be fine on balance; you should just then consider making that “on balance” argument.

      • Hi Dr. Gelman,

        Thank you for taking the time to respond. As you could probably tell, I’m trying to start something new with Steelman Anything, as I think steelmanning has a lot of potential, so this first meta argument about steelmanning itself is very useful. And if you can convince me that steelmanning is deeply flawed, then I’ll give up on it and you might save me a lot of time :)

        I will try my best to respond point by point (your quotes are prefixed with >, mine with >>, and my responses inline) and I understand if you won’t have the time to respond any further or not in much depth, but I would greatly appreciate it if you do. In either case, thanks again for taking the time.

        > it doesn’t seem like anyone was covering it up

        Dr. Alexander’s claims related to this are:

        1. “if you say anything in favor of ivermectin you will be cast out of civilization and thrown into the circle of social hell reserved for Klan members and 1/6 insurrectionists. All the health officials in the world will shout “horse dewormer!” at you and compare you to Josef Mengele.”
        2. “and the elites saying ‘no don’t take it!'”

        Dr. Alexander doesn’t provide any evidence for these claims.

        I’m not immediately aware of any systematic evidence for these claims. We could try to look up evidence (e.g. an analysis of media and government statements on ivermectin), but for the purposes of this argument, I don’t think it’s necessary because we just need to establish whether Dr. Alexander’s premise of a potential cover-up is plausible. I can’t speak for Dr. Alexander, but I’m guessing one large piece of evidence Dr. Alexander was alluding to (particularly with the “horse dewormer” quote) — and perhaps why it was left un-cited by Dr. Alexander given it was a large media story — was the large controversy involving Joe Rogan who is one of the largest podcasters in the world with millions of views per show. The impression many seemed to get was that many news networks and the FDA were referring to ivermectin as a veterinary medicine, implying that it was not suitable for human use, which is something like a lie or cover-up (i.e. why not just say that it’s ineffective?). Here is one analysis of this situation including an interview by Joe Rogan with Dr. Sanjay Gupta from CNN and Dr. Gupta admitting that these statements by CNN and the FDA were inaccurate: https://www.youtube.com/watch?v=8YIhSOL_23A ; Subsequently, CNN seemed to do a poor job of defending itself, and that seemed to be a cover up and another lie: https://www.youtube.com/watch?v=FOVA9UDLjkk

        In regards to the lumping of “tobacco companies, medical researchers, and public health authorities” as one category, my goal was the narrow one of responding to a commenter on your post (which you generally agreed with) that wrote:

        > But it’s not “extremely plausible” that there’s a vast conspiracy

        What my post noted was that:

        >> Dr. Alexander never claimed a “vast conspiracy” and such a conspiracy isn’t necessary

        My point was that the commenter’s claim that Dr. Alexander was claiming a vast conspiracy was a strawman. The tobacco incident may have occurred “plausibly without a vast conspiracy; instead, it was plausibly due to a mix of greed, regulatory capture, and a science funding system that took a while to find the truth”. I did not claim that public health authorities or scientists were part of that conspiracy; my goal was the opposite: to note that they didn’t need to be. I will look to see how I can clarify this.

        > Getting back to ivermectin, you bounce between three theories as to why “elites” (again, not clearly defined) were not advocating mass use of the drug for covid: “Elites lied about positive ivermectin studies”

        This goes back to the point above that many people will think that “covering up” (horse dewormer) is similar to “lying”.

        > This seems to me to be incredibly cynical of you, and you don’t consider the third option that these studies did not, in fact, provide good evidence for the use of ivermectin for covid.

        That is premise #1 in that argument:

        >> Most good ivermectin studies showed positive results.

        Dr. Alexander spent most of his post arguing for that claim and I quoted part of that:

        >> Dr. Alexander’s post argues that the scientific studies did show positive results of ivermectin overall, although Dr. Alexander concludes they were confounded by being done in parasite-prone areas

        I was not judging Dr. Alexander’s claim but just taking it as a plausible premise. Are you saying that at the time of Dr. Alexander’s article, it was completely implausible that ivermectin showed good evidence for the use of ivermectin for covid? If so, what is your evidence? Dr. Alexander does quote the Cochrane Collaboration metastudy, summarizing it as “They end up with a point estimate where ivermectin cuts mortality by 40% – but say the confidence intervals are too wide to draw any conclusion.”

        > given that followup studies did not find ivermectin to be effective

        My goal was not to debate the science of ivermectin efficacy but the meta-argument of whether Dr. Alexander was strawmanning. If such studies occurred after Dr. Alexander wrote his post, and if it’s plausible that Dr. Alexander’s analysis was correct that ivermectin might have been plausibly seen as effective (despite Dr. Alexander’s opinion of confounding by studies in parasite-stricken areas), then Dr. Alexander could not be strawmanning a study that did not yet exist. This seems to be a major misunderstanding in our debate here, so I’ll tag this paragraph as TAG1 and refer to it in subsequent responses.

        > that your only explanations are “lying,” “incompetence,” “myopia,” “greed,” “regulatory capture,” and “Machiavellianism.” How about “scientific judgment”? Maybe an appropriate term would be “selective steelmanning,” where you decide to think the worst of public health officials.

        See TAG1 above; these were the only explanations because the question was whether Dr. Alexander’s argument was a strawman. If you can show that Dr. Alexander’s premise that most good ivermectin studies showed positive results at the time he made his post was a strawman; then, yes, of course, Dr. Alexander’s whole argument would be a strawman, and we can just assume the best of public health authorities.

        > I disagree with your stance of ruling out the possibility that they were showing sound judgment in this case to not get swayed by hype.

        I did consider that:

        >> They may have thought the evidence was inconclusive, required larger experiments, or simply thought the vaccines were a more likely path to containment in an emergency situation

        But if it’s plausible there was a cover up or lying (back to the previous points above), then that would need to be squared with using the other premises in that section (“This sort of medical Machiavellianism is plausible and potentially justifiable”).

        > You also write, “the history of scientific incompetence and regulatory capture (e.g. esketamine, tobacco, etc.).” You use the passive term “regulatory capture.” Again, with cigarettes there was the Surgeon General’s report, the cigarette companies were sitting on lots of evidence on smoking and cancer, and they did a decades-long effort to confuse people on the matter.

        Are you saying that it’s implausible that scientific incompetence and regulatory capture could have occurred (including without a conspiracy)? Dr. Alexander notes in his post and in the eskatimine link some evidence of this sort of plausibility.

        > Getting back to ivermectin, you just seem really committed to this “elites are lying” story, without being clear on who are the elites or without evidence that they are lying. I’d ask why don’t you strawman the elites (whoever they are) and consider that they might have been using solid scientific judgment, but this doesn’t even seem like a strawman.

        I agree that this was a failure of Dr. Alexander to not show the evidence of this claim but my guess about the Joe Rogan anecdote should probably help better frame the debate. I’ll make an update to my post.

        > Given that this is what “steelmanning” my argument looks like, I’d prefer you just address the arguments as they arise.

        I tried to cover the more theoretical point about addressing arguments as they are because I still don’t understand how they help to reduce strawmen:

        1. Steelmanning strives to reduce or eliminate all strawmen (premise; by definition).
        2. A steelman that worsens a strawman is a failure of steelmanning; instead, it’s a tinman (premise; by definition).
        3. “Addressing an argument as it is” does not explicitly involve striving to reduce or eliminate strawmen because “addressing” may involve any form of argumentation (premise; by definition, or lack thereof).
        4. “Addressing an argument as it is” – with a strawman in it – is less likely than steelmanning to reduce or eliminate a strawman (follows from 3).
        5. Therefore, proper steelmanning should reduce rather than worsen strawmen, and more than “addressing an argument as it is” (follows from 1, 2 and 4).

        > Your attempt to steelman my argument just led you into making one claim (that there was no conspiracy to hide the cigarette-cancer link) that was flat-out false

        See above discussion

        > and another claim (that skepticism of ivermectin was due to some combination of “lying,” “incompetence,” “myopia,” “greed,” “regulatory capture,” and “Machiavellianism”) that is at best extremely uncharitable and is, indeed, to my mind, ridiculous.

        See above discussion

        > To be open to all the arguments of ivermectin proponents, Alexander had to be open to some pretty bad arguments, and he didn’t go steelmanning the other sides of those arguments. Again, I can see why he did this, as this could lead to an infinite regress of steelmanning.

        I agree that it does seem to be difficult to steelman all positions at once; however, I look at it less as an infinite regress, and more as a constantly growing argument that tries to take all perspectives into account.

        > So, my recommendation to you is, first to consider that you might be wrong in your belief that all those ivermectin skeptics were lying

        I didn’t bring this up earlier so as not to distract, but at many points in your response, you made reference to my statements and beliefs (and in this above quote, “your belief that all those ivermectin skeptics were lying”), but at no point does my post state my beliefs. I do not share the opinions in Dr. Alexander’s post and I’m not an ivermectin supporter. My goal was to try to accurately steelman yours and Dr. Alexander’s arguments and see whether Dr. Alexander was strawmanning. I have yet to be convinced that he was, but I’m open minded. If he was, it truly would be a big blow to steelmanning given his prominence. This would be along the lines of Ozy Brennan’s argument that I also cover in that post:

        >> You can say “but neither of those are actually steelmanning! Real steelmanning is being able to put other people’s viewpoints in words they themselves find more compelling than their own arguments!” However, that is an extraordinarily rare and difficult skill; even most people who do it once can’t do it consistently. Saying “to steelman position X…” should be interpreted the same way as saying “to express perfect loving kindness for all beings…” It’s certainly a nice ideal which people might want to approach, and some people even manage to pull it off sometimes, but it’s a bit arrogant to declare that you’re definitely doing it. Even when you think you are, you usually aren’t.

        > and second to recognize that steelmanning can involve some misrepresentation of arguments on the other side. Which might be fine on balance; you should just then consider making that “on balance” argument.

        I agree but how are other forms of argumentation any better at misrepresenting arguments when they don’t have an explicit goal of not misrepresenting arguments?

        • Lucius:

          I agree with you that “steelmanning” can be a useful technique in discourse, moving from argumentation to more cooperative conversation. Like all useful techniques, it can sometimes be counterproductive.

          How did things go in the particular post from Alexander that I linked to? There, he steelmanned the arguments of ivermectin proponents while strawmanning the arguments of ivermectin skeptics. I can see why it made sense for him to do so: He ultimately came down on the ivermectin-skeptical position, and perhaps he felt that the steelmanning-and-strawmanning of the argument would make his post more palatable to the ivermectin proponents in his readership. Or perhaps he felt that, once he came to position A, his argument would be stronger if he steelmanned the views of the other side. That may have been a reasonable strategy. Nonetheless, his steelmanning, at least in the form he did it, came at a cost, and this was that he strawmanned ivermectin skeptics, indeed he strawmanned the entire public health establishment. Perhaps he didn’t need to do this—maybe he could’ve steelmanned both sides. That would’ve made his post longer but it could’ve been worth it. So if you want to salvage Alexander’s post for your pro-steelman position, I’d suggest that you argue that he could’ve made his post stronger by steelmanning all the positions in the debate, rather than steelmanning one side and strawmanning the other.

          Regarding the details, I just think you’re twisting yourself into knots. Just for example, you talk about people “referring to ivermectin as a veterinary medicine, implying that it was not suitable for human use, which is something like a lie or cover-up.” Huh? Just for example, here’s the first link that came up for ivermectin. It’s from the U.S. Food and Drug Administration, dated Dec 2021. I’m going to copy in all of it, just to show you what the actual public health authorities were saying. You can disagree with some of the recommendations if you’d like, but in any case it’s sober, reasonable advice with no “lying.” Again, I think what’s happening is that you’re coming into this discussion with a bunch of talking points (“regulatory capture,” etc.) that are not so relevant to what’s going on here.

          I see no lies or coverups in the FDA statement below; its discussion of veterinary medicine is completely reasonable; and, despite your claim, they do not imply that it was not suitable for human use. They accurately say, “Currently available data do not show ivermectin is effective against COVID-19. Clinical trials assessing ivermectin tablets for the prevention or treatment of COVID-19 in people are ongoing.” If you want to disagree with this and argue that the data at that time did actually show ivermectin is effective against COVID-19, then, well, I’d disagree with you—the studies at that time had lots of problems—but that’s a judgment call, the FDA is not “lying” or doing “something like a lie or cover-up,” they’re just coming to a conclusion that you might not like.

          Here’s the FDA statement, in its entirety:

          Why You Should Not Use Ivermectin to Treat or Prevent COVID-19

          COVID-19. We’ve been living with it for what sometimes seems like forever. Given the number of deaths that have occurred from the disease, it’s perhaps not surprising that some consumers are turning to drugs not approved or authorized by the Food and Drug Administration (FDA).

          One of the FDA’s jobs is to carefully evaluate the scientific data on a drug to be sure that it is both safe and effective for a particular use. In some instances, it can be highly dangerous to use a medicine for the prevention or treatment of COVID-19 that has not been approved by or has not received emergency use authorization from the FDA.

          There seems to be a growing interest in a drug called ivermectin for the prevention or treatment of COVID-19 in humans. Certain animal formulations of ivermectin such as pour-on, injectable, paste, and “drench,” are approved in the U.S. to treat or prevent parasites in animals. For humans, ivermectin tablets are approved at very specific doses to treat some parasitic worms, and there are topical (on the skin) formulations for head lice and skin conditions like rosacea.

          However, the FDA has received multiple reports of patients who have required medical attention, including hospitalization, after self-medicating with ivermectin intended for livestock.

          Here’s What You Need to Know about Ivermectin

          – The FDA has not authorized or approved ivermectin for use in preventing or treating COVID-19 in humans or animals. Ivermectin is approved for human use to treat infections caused by some parasitic worms and head lice and skin conditions like rosacea.
          – Currently available data do not show ivermectin is effective against COVID-19. Clinical trials assessing ivermectin tablets for the prevention or treatment of COVID-19 in people are ongoing.
          – Taking large doses of ivermectin is dangerous.
          – If your health care provider writes you an ivermectin prescription, fill it through a legitimate source such as a pharmacy, and take it exactly as prescribed.
          – Never use medications intended for animals on yourself or other people. Animal ivermectin products are very different from those approved for humans. Use of animal ivermectin for the prevention or treatment of COVID-19 in humans is dangerous.

          What is Ivermectin and How is it Used?

          Ivermectin tablets are approved by the FDA to treat people with intestinal strongyloidiasis and onchocerciasis, two conditions caused by parasitic worms. In addition, some topical forms of ivermectin are approved to treat external parasites like head lice and for skin conditions such as rosacea.

          Some forms of animal ivermectin are approved to prevent heartworm disease and treat certain internal and external parasites. It’s important to note that these products are different from the ones for people, and safe only when used in animals as prescribed.

          When Can Taking Ivermectin Be Unsafe?

          The FDA has not authorized or approved ivermectin for the treatment or prevention of COVID-19 in people or animals. Ivermectin has not been shown to be safe or effective for these indications.

          There’s a lot of misinformation around, and you may have heard that it’s okay to take large doses of ivermectin. It is not okay.

          Even the levels of ivermectin for approved human uses can interact with other medications, like blood-thinners. You can also overdose on ivermectin, which can cause nausea, vomiting, diarrhea, hypotension (low blood pressure), allergic reactions (itching and hives), dizziness, ataxia (problems with balance), seizures, coma and even death.

          Ivermectin Products for Animals Are Different from Ivermectin Products for People

          For one thing, animal drugs are often highly concentrated because they are used for large animals like horses and cows, which weigh a lot more than we do— up to a ton or more. Such high doses can be highly toxic in humans. Moreover, the FDA reviews drugs not just for safety and effectiveness of the active ingredients, but also for the inactive ingredients. Many inactive ingredients found in products for animals aren’t evaluated for use in people. Or they are included in much greater quantity than those used in people. In some cases, we don’t know how those inactive ingredients will affect how ivermectin is absorbed in the human body.

          Options for Preventing and Treating COVID-19

          The most effective ways to limit the spread of COVID-19 include getting a COVID-19 vaccine when it is available to you and following current CDC guidance.

          Talk to your health care provider about available COVID-19 vaccines and treatment options. Your provider can help determine the best option for you, based on your health history.

          Again, consider Alexander’s post, which in many ways was eminently reasonable but also had this incendiary bit:

          All the health officials in the world will shout “horse dewormer!” at you and compare you to Josef Mengele. But good doctors aren’t supposed to care about such things. Your only goal is to save your patient. Nothing else matters.

          Pretty clear, huh? “All the health officials” on one side, “good doctors” on the other. This is the kind of ridiculous he was led to by steelmanning one position and strawmanning the other. Again, I think you’d be making a stronger case for steelmanning by just arguing that Alexander should’ve steelmanned both sides.

        • Andrew
          Thank you for quoting the entire FDA statement. I am not going to say anything about steelmanning, strawmanning, or conspiracies/lies. But I think the FDA statement is fairly bad in a number of ways. It is overly repetitious (is that redundant?). Distinguishing between animal and human Ivermectin once should be enough. The entire tone of the statement is classic overly cautious – it reads like the endless warning labels on drugs, household items (e.g. lawnmowers), or virtually anything regulated by the government. Of course, there are good reasons for advising caution. But one thing that has been abundantly clear from COVID is that the FDA and CDC have repeatedly thwarted their messages by over-emphasizing the dangers of anything not officially approved. This has led to suspicions that they are just protecting themselves against potential blame. And their message always comes back to “get vaccinated,” “wear masks,” “get tested,” etc.

          All of that bothers me, but I don’t see it as a conspiracy or coverups. Instead I see it as a continuing pattern of ineffective communication that has fed a number of conspiracy theories. I can’t speak for Alexander or Asclepius – indeed, I haven’t even read what they have written. But I think the FDA statement easily leads to such interpretations, erroneous in my mind. But I don’t agree with your presentation of the statement as “sober, reasonable advice.” I think it is factual but overbearing, repetitive, and ineffective. It would have been fine in a different world than we live in. But given the polarization that has occurred over everything (especially regarding COVID), I don’t think that statement works the way they may have intended.

        • Dale:

          Fair enough. Let me just say that, contrary to what Lucius was saying, the FDA statement mentioned veterinary uses of ivermectin for a reason, and also it doesn’t see anything like a lie or a cover-up. If Lucius or Alexander wanted to make the case that the FDA statement could’ve been better, I’d be fine with that.

      • Hi Dr. Gelman, (I can’t seem to reply to your response below – I guess there’s a thread depth limit – so I’m responding to your previous comment)

        It seems to me that the steelman you would have liked Dr. Alexander to have made would have been:

        1. There was little or no lying, covering up, or manipulating by public health authorities and scientists about ivermectin.
        2. Seemingly the best meta-study available at that point by Cochrane found that “ivermectin cuts mortality by 40% – but say the confidence intervals are too wide to draw any conclusion” leading to a reasonable conclusion that, “Currently available data do not show ivermectin is effective against COVID-19. Clinical trials assessing ivermectin tablets for the prevention or treatment of COVID-19 in people are ongoing”.
        3. Public health authorities were concerned about:
        a) potential risks of taking things unapproved for COVID,
        b) people potentially not taking approved and efficacious treatments because they think ivermectin might work instead, and
        c) people take a veterinary form of ivermectin that’s different from the human form and has additonal risks.
        4. The media’s repetition of the phrase “horse dewormer” was mostly a combination of:
        a) alerting people of the risks of 2(c) in a memorable way, and
        b) simplifying 2(b) for a general audience
        c) rather than any sort of coordinating talking point, media editors watch each other and they found “horse dewormer” was effective so they copied it

        Everything really hinges on premise 1. In the first YouTube clip I linked in my last post, Dr. Gupta referenced the FDA’s tweet that linked to the FDA statement that you quoted which wrote: https://twitter.com/US_FDA/status/1429050070243192839

        > You are not a horse. You are not a cow. Seriously, y’all. Stop it.

        However, the steelman for that would be that this wasn’t some grandly orchestrated manipulation campaign by public health authorities but just some editorializing by a low-level social media person at the FDA along the lines of premises 3 and 4.

        Therefore, significantly more evidence would be needed to substantiate premise 1 about lying, covering up, and manipulation.

        It seems to me that Dr. Alexander simply rejected premise 1 (either he himself rejects it or he believes in premise 1 but he was rejecting it as part of his steelman) and therefore the rest of the above steelman would have been implausible. In addition to the two quotes I pointed out in my last repsonse, he also wrote, “even the people who don’t usually lie were saying it was ineffective, and they were saying it more directly and decisively than liars usually do”.

        The major failing here is that Dr. Alexander never substantiated his claims against premise 1. And since he is the one claiming lying, covering up, and manipulation, the burdern of proof is on him to substantiate this, and you do not bear a burden to disprove such a claim.

        Therefore, your argument is that in his zeal to steelman one position, he strawmanned another position.

        For the sake of argument, let’s just assume premise 1 is true and Dr. Alexander had a colossal failure of substaniating that it wasn’t true. In this case, I agree that Dr. Alexander’s post did strawman public health authorities and scientists.

        The question for me though is: Why do you think something particular about the steelmanning mindset caused this failure? It seems to me that, given Dr. Alexander’s general propensity to maximally substantiate, Dr. Alexander’s lack of substantiation against premise 1 suggests that he assumed that premise 1 being false is common knowledge.

        If so, then if Dr. Alexander was simply “addressing the argument as it was”, he would have still had that same assumption against premise 1, no? I’m assuming “the argument” in this case (as evidence by the second paragraph in his post) was to try to understand the supposed scientific disagreement about the efficacy of ivermectin. It seems like if he was just “addressing the argument as it was,” that he would reach the same conclusion because he disagreed with premise 1.

        If so, then I think the failing isn’t steelmanning, but a lack of it.

        If so, then this comes back to my second steelman of your argument (since I haven’t seen any counter-argument to the first steelman of your argument; i.e. that there’s nothing in “addressing an argument as it is” that would reduce strawmanning any more):

        >> 1. Dr. Alexander is a major figurehead of the steelmanning movement (premise).
        >> 2. A field’s general soundness can be gauged informally by its experts’ use – or misuse – of that field’s tools (premise).
        >> 3. Dr. Alexander committed two major instances of tinmanning (Dr. Gelman’s arguments).
        >> 4. Therefore, something is rotten in steelmanning (follows from 1, 2, and 3).

        I’m now willing to grant premise 3. However, and perhaps this is just me trying to justify a sunk cost (I’m not sure how to evaluate that), I will instead reject premise 1. Given the lack of substantiation by Dr. Alexander for such a controversial premise that public health authorities and scientists are massively lying, covering up, and manipulating, it seems to me that Dr. Alexander is simply not very good at steelmanning (despite being a figurehead).

        I content that if he had written out that premise very clearly, then I think he would have seen the need to substantiate it.

        The deeper issue for me, though, is still my first (admittedlly quite theoretical) steelman against your argument: Steelmanning is, by nature, against strawmanning, whereas other types of argumentation are not necessarily. I’m guessing that if Dr. Alexander was not a steelmaner but was an “addressing the argument as it is”-er, he would have made the same mistake, and, he probably would have strawmanned other premises of the pro-ivermectin crowd.

        Thank you very much for this discussion. It has been very thought-provoking and I’ll be revising my post (and I’ll reference this discussion).

        > I think you’d be making a stronger case for steelmanning by just arguing that Alexander should’ve steelmanned both sides.

        I agree.

      • Hi Dr. Gelman,

        Perhaps this is a better steelman of your argument against steelmanning per se (and charitably assuming Dr. Alexander didn’t really believe public health authorities were lying):

        1. Imagine that person A would like to steelman person B’s argument.
        2. Imagine that person B’s argument has a premise P1 that person A does not accept beforehand.
        3. Imagine that in the process of steelmanning, person B accepts premise P1 for the sake of argument to empathize with them.
        4. Imagine that premise P1 is a strawman of person C.
        5. Therefore, due to steelmanning, person A implicitly strawmanned person C whereas if person A was simply addressing person A’s argument as it was, they would have denied premise P1 and not caused the strawmanning of person C,

      • FYI, completely updated the page based on our discussion (and made a note of the update): https://steelmananything.com/topics/introduction/#steelmanning-one-argument-may-strawman-another

        In summary:

        Steelmanning you (hopefully): If person A is steelmanning an argument of person B, person A might accept one of person B’s premises as part of steelmanning (i.e. empathizing “for the sake of argument”). If person A didn’t otherwise believe this premise and another person C considers this premise a strawman, then steelmanning introduced a strawman that Person A wouldn’t have otherwise done if they were just “addressing the argument as it was”.

        Response: Accepting any strawman while steelmanning – even of someone who’s not necessarily involved in the argument – is a failure of steelmanning; instead, it’s tinmanning. Person A should have also steelmanned Premise P1. In contrast, “addressing an argument as it is” is not necessarily designed to reduce strawmen. Even if such an approach did not include Premise P1, it may have strawmanned other premises.

        Steelmanning againt that: Steelmanning an argument so that all people do not consider that it contains any strawmen may be impossible due to peoples’ contradictory premises. Even if we grant in the above example that, theoretically, Person A should have steelmanned Premise P1, there may be no way to steelman it in such a way that both Person B and Person C are happy.

        Response: While it may be impossible to steelman some arguments in a way that everyone accepts, if the goal is to empathize with other people for the purpose of improving relations between all people and helping achieve the best possible world for everyone, then there is no better method than steelmanning (see the all truth-seeking is hard response), and we should try our best to steelman as much as possible.

        • Lucius:

          I don’t think any of what you’re saying makes much sense, but I guess different people have different tools for understanding the world.

Leave a Reply

Your email address will not be published. Required fields are marked *