A balanced view of ad hominem judgments

by Carl V Phillips

Tap tap tap. Is this thing on?

Welcome back to this blog. As many of you know, The Daily Vaper, where I published most of my good material for a year, has ceased publication (the articles, fortunately, are still archived at my author page at dailycaller.com, and they redirect from the original links if you have used those somewhere). I also recently did a “best of” Twitter thread highlighting some of what I wrote there (and here and elsewhere). There is something simultaneously atavistic and postmodern about watching (now nonexistent) the DV website slip slowly down the list of top guesses for where I might want to go when I open a new browser tab.

I am writing most of my subject-matter analysis under contract these days, with a bit of freelancing for commercial websites. Deep-think tangents will start to reappear here. Like this one. (I thought about doing it as a Twitter thread, but I realized that would never work.)

The purest form of ad hominem fallacy is to criticize an analysis based entirely on who it was that wrote or endorsed it. As in, “I read this paper, the methods and analysis were clearly presented, and the stated conclusions followed from the results, but I still don’t believe them because I don’t like the author or who he associates with.” This is, obviously, moronic. A minor variation — the most common form of “reasoning” used by tobacco controllers — is, “I did not even try read this paper [often meaning: it was is beyond my ability to understand], so I have no idea whether it was tight, but I am going to declare the results are wrong because I don’t like the author or who he associates with.” This is moronic for the same reason, and is also pretty pathetic to implicitly admit.

But these bits if idiocy do not mean that there is nothing to be learned from what might be called an ad hominem analysis. This requires far more effort, knowledge, and wisdom than most knee-jerk ad hommers are capable of, but its value should not be dismissed just because it partially resembles the knee-jerk crowd’s anti-scientific approach (such a dismissal, ironically, might be thought of as a second-order ad hominem attack).

Should you consider the known history of the author when assessing a paper? (Note: I am using “paper” as a shorthand for any published analysis where the reader can follow the reasoning; it could be a research paper, essay, book, position statement, or even a detailed Twitter thread. This contrasts with an unsupported free-standing assertion of a conclusion, which is necessarily judged entirely based on how much you trust the author.)

Yes, of course. This obviously does not mean you should let the author’s history override what you learn from actually reading the paper. But the methods and analysis are almost never fully presented in papers. Unless you are a huge expert on the subject matter, you may not recognize there is cherrypicking of inputs or references. Unless you are an expert on the methodology, you might not recognize tricks that were used in it, even if they are openly reported. Thus, you almost always have to decide how much you trust the author. Science (which I define very broadly) is all about trust, contrary to the simplistic lessons from grade school that it is all about skepticism. It is a rare paper that is such a complete presentation that you should be willing to think, “I don’t trust this author further than I can throw him, but the analysis is clearly correct,” (and a paper of that quality is almost certainly written by someone who is impeccably trustworthy).

On the subject of authors worthy of throwing, consider an extreme case: Stanton Glantz and his colleagues and minions at UCSF. (The groups at Bath and Karolinska are pretty similar, along with, of course, most of the anti-tobacco groups and government agencies. The UCSF group are the extreme case in this space, but are not outliars [sic].)  Basically everything they publish about tobacco products is a lie. That is not hyperbole. They use junk science methods, reported poorly, and draw conclusions that do not follow from the evidence in basically everything they do. You should never believe any conclusion from one of their papers. This is not because of their funding (usually, but not always a red herring — see below) or because they are they are “on the other team.” It is because there is overwhelming documented concrete evidence that most of what they write is lies.

You might choose to read one of their papers in detail to try to mine some genuine information, which often appears despite the overall dishonest narrative. You might want to read it just to know what they are on about now, in the same way (sensible) people read Trumps tweets. But you should never believe any conclusions, nor assume that unreported methods and opaque steps in the analysis were done correctly. You know this by virtue of the authorship alone. This is proper ad hominem analysis.

For authors who are known to often do junk work, but not always lie, a bit of deference should be offered, but not much. You might not want to assume the whole thing is junk. But when you read it and wonder if they might have skewed their analysis by doing X (and omit any mention of whether or not they did X), you should assume they did based on their history. It is important to keep in mind that X usually includes running their data through various models and reporting only the one that produced their preferred result, pretending that was the only result they got (I have written a lot about this elsewhere — example.)

It is a bit tricky to sort this out from knee-jerk ad homming, in the tobacco research space, because there is such a high correlation between “tobacco controller” and “most of what they write is lies”. It is useful to think of some other realm where you have strong political views but there are some with opposing views whose analyses you trust to be solid, though you still might read the details and find fault. (If you have no views fitting this description, such that there are some authors in the space who you usually disagree with but whose analyses you assume will be honest and reliable, consider that perhaps you are a knee-jerk ad hommer.) I wanted to provide an example of someone “on the other team” who you might disagree with but should generally trust, but I literally cannot think of anyone. You might have to carve out individual pieces that a few of us on “our side” write that express skepticism about some of the cheerleading and sloganeering on our side. Those are examples of someone you (presumably) trust writing things that you might not agree with, but you are inclined to believe they must have something valid to say.

The proper ad hominem analysis contrasts sharply with what tobacco controllers do. They do not identify any serious flaws in the corpus by a particular author (and in many cases could not even if they were not too lazy to try, because they do not exist), but merely dismiss his work because they do not like his views. I have watched with amusement as they attacked fatally flawed papers from “our side” — pieces that can easily be shown to fail on their merits — based only on ad hominem attacks on generally reputable authors.

Tobacco controllers (and many others) typically focus on funding in moronic ad hominem attacks. This is not surprising because actually addressing issues of whether the author is honest and does good science is beyond their ken; they understand about as much about honesty and good science as a cat. But the funding they talk about is usually, not always, a lousy basis for an ad hominem analysis.

First, the fact that someone funded a particular paper is completely uninteresting. That money was already paid or promised before the research was published, and so offered no further incentives. However, trying to please funders who you expect to renew ongoing “center” type operations, but who have not promised the money, or fishing for new funding can affect present conduct. I had a senior colleague once whose policy was to never take research funding from the same entity twice (a luxury that no longer exists, except for maybe Nobel Prize winners). This was the perfect proof against funding-based conflict of interest, even at the subconscious level. But it is a bit overkill. It is also another case where anyone who would do this is obviously so honest and scrupulous that he is last person who actually needs to do it.

This means that the usual “this paper was funded by X, and therefore…” always falls onto the moronic side of ad homs. An ad hominem analysis of funding needs to look at whether the prospect of future funding is being dangled in front of the researchers. That takes more analysis — only a tiny bit more, but even that is beyond the abilities of the usual ad hommers. Of course this only only the case when (a) the funder is the type who would pull funding if the researchers published legitimate work that produced conclusions the funder did not like and (b) the researcher venal enough (and/or vulnerable to peer-pressure) to let that override whatever integrity he might have. The latter is certainly the case for many tobacco control researchers. The former is the case for basically all tobacco control funders but — and this is important — not for industry funders. There are documented cases of tobacco controllers pulling someone’s funding for getting the “wrong” results, whereas no company would dare do that even if they were so inclined (which few are).

Notice this in particular this means that, in the tobacco research space, the people that are most likely to be affected by funding are employees of anti-tobacco operations: CDC, NIH, FDA, health departments, FCTC, anti-tobacco quangos, and Legacy/Truth and their ilk. Authors with that employment know that their future paychecks depend on staying on the good side of some seriously monomaniacal and unforgiving bosses. I am always floored (though never surprised) when one of them publishes a paper and declares “the authors have no conflicts of interest.” There is barely any greater conflict of interest.

Coming in second in the space are those receiving potentially ongoing grants from those organizations or junior researchers who hope to get such grants or staff positions. An obvious example of this is where someone has a big-money “center” grant (as with FDA CTP funding of shops at a dozen universities), making the careers of the lead investigator and many of the employees dependent on renewals. Junior researchers who are sucked into this space are told, in no uncertain terms, that toeing the party line is the only way they will get their turn to suckle at this teat. (Quite literally — I’ve watched it happen, with students bailing on working on THR with me to go work with third-rate hacks who intimidated them with threats (probably valid) about their career prospects.) There is good reason to believe that people in these positions will alter their “research” to please the funders.

Of course, not everyone is willing to sell their intellectual integrity. A lot of the correlation between “funder’s goals” and “researcher’s views” is caused by funders and perfectly honest researchers, who happen to already have similar views, seeking out one another. An ideal ad hominem analysis should thus look at evidence about just how willing to sell out the individual is. The clearest example in the tobacco space might be Jack Henningfield, who was a ferocious opponent of THR when he was being paid by Glaxo and then became a big fan of low-risk substitution (never admitting he had ever been otherwise) when he started being paid by Njoy. But this can be observed even among those who seem to not switch sides. Glantz et co., despite all their junk science, were never really part of the anti-THR cabal until FDA started doling out big money for anti-THR. Then they became all about anti-THR and the war on nicotine.

There is plenty more to say about all of this, but just planting this basic ideas about proper ad hominem analysis is a good start.