Previous Month | RSS/XML | Current | Next Month
WEBLOG
July 31st, 2023 (Permalink)
Nullius in Verba1 & Thinking for Yourself
- Nate Silver, "Journalists should be skeptical of all sources―including scientists", Silver Bulletin, 7/21/2023. Warning: Contains some vulgar language.
In March 2020, a group of scientists―…Andersen [et al.]―published a paper in Nature Medicine that seemingly contradicted their true beliefs about COVID's origins and which they knew to be misleading. The paper…was enormously influential in shaping the debate about the origins of COVID-19.
We know this because of a series of leaked and FOIAed emails and Slack messages…. The messages show that the authors were highly uncertain about COVID's origins―and if anything, they leaned more toward a lab leak than a spillover from an animal source. But none of that was expressed in the…paper, which instead said that “we do not believe that any type of laboratory-based scenario is plausible”. Granted, there is a little bit of ass-covering—“More scientific data could swing the balance of evidence to favor one hypothesis over another,” they also wrote in the paper. But the message—natural origin good, lab leak bad—was received clearly enough by mainstream news outlets. “No, the new coronavirus wasn't created in a lab, scientists say”, reported the CBC in covering the paper. “COVID-19 coronavirus epidemic has a natural origin” was the headline at Science Daily.
In the Slack and email messages, the authors worked to manipulate the media narrative about COVID-19’s origins and to ensure that their private uncertainty wasn’t conveyed in conversations with reporters. They also thought they were going to get away with it. … This went beyond mere motivated reasoning. There was an enormous gap between what the authors believed privately and what they stated publicly, including in the…paper….
What were the authors' motivations to mislead the public? I think that's also pretty straightforward. In fact, you can find prominent virologists quoted on record as to why the lab leak theory was so problematic―even if it wasn’t necessarily wrong. The problems fall into three buckets:
- Evidence of a lab leak could cause a political backlash—understandably, given that COVID has killed almost 7 million people—resulting in a reduction in funding for gain-of-function research and other virological research. That’s potentially important to the authors or the authors’ bosses—and the authors were very aware of the career implications for how the story would play out;
- Evidence of a lab leak could upset China and undermine research collaborations;
- Evidence of a lab leak could provide validation to Trump and Republicans who touted the theory—remember, all of this was taking place during an election year, and medical, epidemiological and public health experts had few reservations about weighing in on political matters.
To be clear, I'm not sure how COVID originated either. … But I think this is a big scandal either way. …I'm deeply disappointed by the scientists' conduct here and how unmoored they were from any attempt at truth-seeking.
The COVID origins story has also been a journalistic fiasco, with the lab leak having been dismissed as a "conspiracy theory" and as misinformation even though many prominent scientists believed it to be plausible all along. Perhaps it's tempting to give the media a pass―they were manipulated by the…authors, after all. But I’m not inclined to, for two reasons.
First, the coverage of the recently leaked emails and Slack messages at major center-left outlets like The New York Times has been pathetic. The Times portrayed Andersen as the victim of a Republican witch-hunt—rather than someone at the center of a major scientific scandal of his own making.
And second, journalists ought to have decent [BS] detectors—including toward scientists, academics and other experts.
Maybe you think Andersen et. al. are bad apples, but the messages make clear that they were speaking for a pretty broad swath of the scientific community. Still—and maybe this is wishful thinking—but I’m going to assert that people like him are in the minority among scientists. I fairly often speak with scientists and academics myself…and those experiences are overwhelmingly positive.
And yet, even if the incidence of bad apples is relatively rare among scientists and academics—rarer than it might be among politicians or other groups that journalists intrinsically treat with more skepticism—it’s clearly not exceedingly rare. It was just this week that the president of Stanford was forced to resign in a research scandal. (Perhaps not coincidently, the scandal was broken by the student newspaper, The Stanford Daily, and not by a major center-left outlet like The Times.)
I also think journalists are more prone toward being manipulated by bad apples in academia and science than they were ten or twenty years ago. As a result of increasing educational polarization, both journalists and the expert class of scientists and academics are far more aligned politically than they once were (the very large majority are left-of-center and vote Democratic in American elections).
I'm a couple of decades older than Silver and I think he's wrong about this. The "very large majority" of academics, scientists, and journalists have been "left-of-center" Democrats as long as I recall, though the situation may be marginally worse than it was, say, fifty years ago. What seems to have changed is the attitude towards ideological dissent within their ranks: it used to be accepted among liberals that free speech and intellectual diversity are good things, but the current ideology favors an enforced orthodoxy. Why this has happened, I don't know.
Even if “trust the science” or “trust the experts” is usually right—and I think it usually is right!—it leaves an opening for bad apples…to exploit the trust that honest scientists have worked so hard to earn.
"Trust the experts” and “trust the science” mean entirely different things. First of all, you can't just trust the experts: which experts? Different experts were saying different things. We should certainly trust the science, but the science was not on the side of the establishment "experts" who became prominent during the pandemic. We now know that the science was on the side of the dissident experts, not just in the case of the lab leak hypothesis, but also on such matters as economic and social lockdowns, school closings, and masks. Too often what "trust the experts" means is "trust government bureaucrats", and "follow the science" means "follow the government's orders".
There’s also a generational divide in journalism, with younger journalists tending to be more openly left/progressive than their older peers—and tending to be more Manichean in dividing the world between good and evil rather than proceeding from the notion that people and news stories are complicated and it’s not particularly their job to pass moral judgment. … But this really isn’t complicated. All I’m suggesting is that journalists ought to treat scientists like they do any other source—that is to say, with an appropriate dose of skepticism.
Scientists and other experts are human―all too human―and they will fail in exactly the same situations and ways that other human beings do. There's no reason to think that scientists are any better than anyone else at resisting the temptations to lie, cheat, and steal when their careers or financial interests are at stake. This is why we need to follow the science, not the scientists.
- Matt Taibbi, "Are Authorities Using the Internet to Sap Our Instinct for Freedom?", Racket News, 7/14/2023
I don't agree with Taibbi's suggestion, in the title of this article, that "authorities" are doing this intentionally. Unfortunately, the situation is worse: as I wrote earlier this month2, we're doing this to ourselves.
…Imagine that instead of Google, we invent in the near future an internalized search engine, one in which our thoughts can move a cursor. … You’re walking down the streets of Washington and you visibly encounter a demonstration in front of the White House. … Instead of flashing labels for “harm” or “age inappropriate” content, this new app tinges certain images it finds dangerous with the same warning colors animals use in nature to warn others to stay away, like yellow or red. These protesters therefore look a bit like poisonous insects to you, bathed ever so slightly in visual menace. Then, when you try mentally to remember what you know about the group you’re looking at, you again get the search engine, which calls up a string of condemnatory headlines and words like “fascist” or “extremist” or “antivax,” which are just the verbal equivalents of black and yellow stripes, transparent code for “threat.”
With this app you’ll also have internalized access to communications services…. Do you silently capture an image and forward it on social media to tell people what you’ve seen? Likely as not, you won’t, unless it’s to attach a negative comment, which you may do instinctively, in search of the dopamine hit of return affirmation the Internet has already proven so effective at delivering. …
It's not the internet per se that does this, but anti-social media.
This seemingly far-out situation I’m describing that would go on entirely inside your head is not different from what people already experience. …[W]e lean more and more on machines to do our thinking for us. However, the worst part is, we often do not distinguish between thinking that is ours, and thinking that is someone else’s.
We Americans once cherished independence, and lived off folk tales about going off on one’s own, on the open road. … That was then. Now instead of giving the world something invigorating and freeing…, we’re exporting mass neurosis. At home we’ve become afraid to walk even a few steps without our electronic helpers. Our sense of self is now inextricably tied to a huge global entourage of prying commentators who live in those phones of ours that are always in our pockets and whose good opinion we never stop seeking, whether we admit it or not. …
But thinking for yourself is hard work, and political interests in the Internet age have preyed on another very American instinct: laziness. Their sophisticated programs begin with the premise that the Internet always punishes difference and rewards conformity. This is the core principle at work in shadow-banning and de-amplification algorithms. These automated surveillance tools look for phrases like “Open-minded” or “I like to do my own research” or “I’m generally apolitical” and don’t score the people saying such things as tolerant, creative freethinkers. What the algorithm instead detects is someone harboring a dangerous willingness to embrace unorthodox ideas, or look at a forbidden thing and not flee.
It was once a virtue for Americans to say, when asked about their politics, “None of your damn business.” Nobody thinks that way anymore, either. Young people especially are worried to the point of mental illness about their likes and ratios. We not only want people to know what we think, we’re terrified of people not knowing what we think, lest we be suspected of harboring something unsavory underneath.
This is how it is for Americans trying to be themselves now. First they became addicted to the Internet as a tool of convenience. Then it became a cheap substitute for real-life interaction. Finally they learned to submit to the wisdom of crowds, which on the Internet, as we also found out, is really an artificial representation of a crowd, generated by political and social engineers…. The results have not been good.
Even if it's a real crowd, its alleged wisdom is a myth, since a crowd is only as intelligent as its least intelligent member.
If they can preemptively extinguish that fire in us, formal censorship will become unnecessary. The population will become too fearful of difference to ever risk punishment in the first place. That moment is close at hand.
Notes:
- Translation: "Nothing in words", Latin. This motto of England's Royal Society was adopted in 1660, and means "take no one's word for it". The motto emphasized the importance of experimentation as opposed to the authority of Aristotle and the Bible. No scientific claim should be taken on the word of scientists; all must be independently replicated. See: "History of the Royal Society", The Royal Society, accessed: 7/31/2023.
- Two Books on George Orwell's Nineteen Eighty-Four, 7/15/2023.
Disclaimer: I don't necessarily agree with everything in these articles, but I think they're worth reading as a whole. In abridging them, I have sometimes changed the paragraphing of the excerpts.
July 26th, 2023 (Permalink)
Headline Hype
Experts Say Decades of Recycling Hype Has Backfired Dramatically*
Yeah, we need some new hype.
Also, has decades of teaching English grammar backfired?
*Michaela Barnett, et al., "Experts Say Decades of Recycling Hype Has Backfired Dramatically", Science Alert, 7/25/2023
July 23rd, 2023 (Permalink)
From Fact Checking to Censorship
Quote: "A casual political observer may take the 'fact-checker' label at face value. But the entire industry (with little exception) serves as a Trojan horse to justify censorship for the political left."1
Comment: That's a rather sweeping generalization. Unfortunately, some of the fact-checking groups have indeed cooperated with the censorship programs of anti-social media, but not all2. For instance, Snopes was established in 19943, long before any of the major "social" media came into existence. If the fact checkers have since been compromised, that's unfortunate and should be exposed, but it's no reason to condemn the entire "industry".
Title: Fact-Checking the Fact-Checkers
Subtitle: How the Left Hijacked and Weaponized the Fact-Checking Industry
Comment: Clearly, the subtitle announces that this book is a conservative critique of the "fact-checking industry", which must be a selling point for the book's target audience. I'm disappointed that, as with everything nowadays, fact checking has become so politically polarized, and I would prefer a less partisan approach. However, I'm still interested in the book's criticisms.
Author: Matt Palumbo
Comment: Palumbo is the author of a half-dozen previous books, all of which appear to have a conservative slant, and none of which I've read.
Date: 2023
Comment: This book was just published earlier this month.
Summary: Judging from the table of contents, the book is divided into two parts: the first part is entitled "The Fact-Checking Industry" with sections on PolitiFact, Snopes, and other major fact checkers, and then what appear to be four sections on "social" media censorship. The second part consists of over a hundred discussions of individual fact checks divided into sections on economics and COVID-19, among other subjects.
The Blurbs: The book has three blurbs: first, from James Agresti, the president of the "Just Facts" conservative fact-checking site; second, from Charles Lipson, an emeritus professor from the University of Chicago; and, third, from Dan Bongino, a conservative radio host. There's a conflict of interest in the endorsement from Bongino, whose imprint, Liberatio Protocol, published the book4, so of course he would recommend it.
General Comments: I'm currently reading a book on fact checking by a leftist academic5―but I repeat myself―and I'd like to see criticism of the practice from the right. I've often read general comments from conservatives complaining that various fact-checking groups are biased, but these generalities are seldom backed up by specific charges. As noted in the Summary, above, this book appears to contain a large number of critiques of individual fact checks.
I've said before6, and I'll say again, that I think the major fact checkers―Annenberg, Glenn Kessler, PolitiFact, and Snopes―usually do a decent job, at least when it comes to fact-checking specific claims. This doesn't mean that they're perfect―no one is, and it's unfair to expect perfection―nor does it mean that they are completely unbiased. Kessler, for instance, writes for The Washington Post, an establishment liberal newspaper, so it's inevitable that some of that political slant rubs off on his fact checks. However, if a fact check lives up to its name, then it is usually easy to make allowances for bias. So, I don't care what the personal politics of the fact checker are if the checker gives me the facts; I'd prefer to get just the facts, ma'am, without a dose of the checker's opinions, but they're easy enough to discount.
I'm already on record against the use of ratings systems such as the "Truth-O-Meter" and Pinocchios6, as well as the "Lie of the Year" that some fact checkers announce, which are a source of both subjective bias―as some of the fact checkers themselves have admitted7―and inconsistency. While it's easy for a reader to ignore or downplay such gimmicks8, the fact-checkers should stop using them.
A more difficult type of bias for a reader to deal with is selection bias, that is, bias in selecting what to check. If a fact checker checks many more claims made by members of the Republicrat party than those from the Demoblican party that is evidence of bias even if the individual checks are factually correct. Palumbo provides pretty strong evidence that PolitiFact is biased in its selection of what to check in a way that favors Democrats over Republicans. The only solution to this problem, outside of PolitiFact changing its ways, is to not rely upon a single fact checker. In general, it is a good idea to check more than one source, and that's as true of fact checkers as anything else. There are right-leaning checkers, such as the previously mentioned Just Facts, to counter-balance the left-leaning ones. In addition, there's now this book.
Copyediting: The Kindle version of this book that I've been reading has some weird text, specifically, occurrences of the "words" "text-d", "text-dd", and "text-ds". For instance, the title of one section in the Table of Contents is given as: "CNN Adds Words to text-d So They Can Rate It False"―what does that mean? A later sentence in the book reads: "Riedl said that bias at PolitiFact was out in the open, and reporters would explicitly ask for text-ds negative towards Republicans." There are a total of six such occurrences in the excerpt that I read. My guess is that these strange "words" are the result of a text conversion program used to convert a scan of the printed text into e-book format. Whatever happened, it would appear that no copyeditor checked the final Kindle version of the book.
Disclaimer: I haven't read this book yet, so can't review or recommend it, but its topic interests me and may also interest my readers. Also, the above comments and criticisms are based only on reading a sample of the book.
Notes:
- "Intro and Acknowledgments"
- For an example of a "fact checker" used by Facebook to suppress unwelcome truths, see: John Tierney, "This Article Is 'Partly False'", City Journal, 5/17/2021.
- "About Us", Snopes, accessed: 7/22/2023.
- Jeffrey Wernick, "Why AlignPay", The Dan Bongino Show, 6/30/2021.
- Lucas Graves, Deciding What's True: The Rise of Political Fact-Checking in American Journalism (2016). I haven't finished reading this book, but it appears to be more of a history than a critique of the practice of fact checking.
- Fact-Checkers ≠ Lie-Detectors, 8/27/2021.
- Kessler has said that such decisions are "fairly subjective"; "Fact Checker: The biggest Pinocchios in politics", The Washington Post, 8/24/2011.
- According to Graves, this is what the editor of PolitiFact's parent newspaper called the Truth-O-Meter; p. 41.
July 15th, 2023 (Permalink)
What's New?
I've added a review of two books about George Orwell's novel Nineteen Eighty-Four to the book review page. Check it out:
Two Books on George Orwell's Nineteen Eighty-Four
July 13th, 2023 (Permalink)
Affect or Effect?
Confusing the words "affect" and "effect" is so common that all but one of the reference books I usually consult on such confusions discusses it. There are several reasons why this confusion persists, despite such frequent warnings. One obvious reason is that the spelling of the two words differs by only a single letter, but they are also pronounced nearly identically unless one emphasizes their initial vowel sounds. More importantly, while they don't mean the same thing, their meanings are related.
"Affect" is a verb meaning "to influence", that is, to change something. It does occur as a noun, but with an unrelated meaning having to do with emotion, and then it is usually pronounced with the accent on the first syllable1.
"Effect" occurs as both a noun and a verb. As a verb, it means "to accomplish" or bring about, which is a stronger meaning than to affect; as a noun, it refers to the results of such causation, as in the phrase "cause and effect": a cause is an event that brings about another event, the effect2.
I initially had no intention of writing about this confusion because it's too common to be interesting, but then I came across the following sentence in America's most prestigious newspaper: "Estelle Freedman, a history professor at Stanford University who advised Dr. Petrzela on her Ph.D. thesis, described her as 'a very serious scholar and a public intellectual who is quite unique in imagining, "How do we get scholarship out into the world and affect social change?"'"3
What is it to "affect social change"? Is it to change social change? While it could mean to influence social change, my guess is that Freedman meant to effect social change, that is, to bring it about. Notice that Freedman is being quoted by the author of the article, Katherine Rosman, so probably Rosman misheard "effect" as "affect", and then Rosman's editors failed to catch the mistake, which is still uncorrected.
Since both "affect" and "effect" are correctly spelled English words, a pure spell-checking program will not be able to tell them apart. "Affect" usually occurs as a verb and "effect" as a noun, so it's possible that a program that also checks grammar will notice the substitution if the verb is put into a grammatical position requiring a noun, or vice versa.
I checked the example sentence in my old copy of Word, Microsoft's word-processing program, and the only thing it had a problem with was the name "Petrzela"―who can blame it? I also checked it on several on-line spelling and grammar checkers and none of them flagged "affect". One did object to the phrase "quite unique", which is quite a good point since uniqueness does not come in degrees. Since the grammar of the example sentence requires a verb in the position of "affect", it's unsurprising that no programs were able to catch the error.
My experience suggests that this confusion goes mostly one way, namely, that "effect" is used when "affect" is intended, though the example sentence shows that the reverse mistake does occur. Given that a spelling or grammar checking program is unlikely to stop you from mistaking one for another, and even the human copyeditors of The New York Times may fail to do so, this is a distinction to remember.
Notes:
- "Affect", Britannica Dictionary, accessed: 7/13/2023.
- "Effect", Britannica Dictionary, accessed: 7/13/2023.
- Katherine Rosman, "A History Professor Takes On Hollywood", The New York Times, 3/12/2023.
July 5th, 2023 (Permalink)
How to Solve a Problem: Climbing Up that Hill
As usual, let's start out with a problem, but this time it's not a puzzle; rather, it's a thought experiment. Imagine that you are at the foot of a large hill and you want to get to the top of it. Suppose, further, that you do not have a map, and there is no trail that you can follow to the top. Moreover, the hill above you is high enough that it is hidden in clouds, so you can't see the summit or a route to it. How can you get to the peak?
This is an easy problem and the answer should be immediately obvious: just climb. That is, start walking uphill, in such a way that each step takes you up the slope. If you keep doing this long enough, eventually you'll find yourself in a position in which every step that you can take is either level or downhill. At that point, stop climbing for you have reached the summit.
The problem solving technique that you just used in your imagination is called "hill-climbing". Despite its name, hill-climbing is a general problem-solving strategy, and climbing hills is just one application of it1. Here's how the strategy is generalized: Suppose that you have a problem and you lack an immediate solution to it, but you do have a way of gauging your progress towards the solution. Then, to "climb the hill" is to act in such a way that you are always advancing towards the solution. As long as the problem is finite, you'll eventually find yourself in a position where you can advance no farther, and that will mean you have found the solution.
Applied to actual hill-climbing, your problem is to get to the summit of the hill, and you have no immediate way of doing so, but you know that the summit is up. Since you can tell whether you're going up, down, or on the level, you have a way of gauging your progress towards the summit. The hill-climbing algorithm tells you to ABC: "Always Be Climbing".
One familiar type of recreational puzzle that hill-climbing can be used to solve is jigsaws. There are other strategies that you can use for jigsaw puzzles, and we'll see at least one of those in a future entry, but the basic problem-solving approach is hill-climbing.
I assume that you've worked on a jigsaw puzzle at some time in your life, at least as a child. The goal of such a puzzle is, of course, simply to put all the pieces together so that they fit and, usually, make a picture. Starting with a completely unassembled puzzle, you can't immediately see how to put them all together. However, you do have a way of telling whether you are advancing towards your goal: as long as you're putting pieces together, you're making progress. Since all jigsaw puzzles are finite, if you keep putting pieces together, eventually you'll finish it. When you run out of pieces you have solved the puzzle―assuming that you haven't lost any pieces or put them in the wrong places.
As is the case with the other problem-solving techniques we've looked at, hill-climbing is not guaranteed to work. For instance, if the hill is too high, you may just run out of time or energy before you get to the summit. This is primarily what accounts for the difficulty of jigsaw puzzles, that is, the bigger the puzzle, the more pieces it has, the longer it will take to solve it. If a jigsaw puzzle is too big, you may give up before solving it.
In addition to the height of the hill, there are three other problems that may cause hill-climbing to fail:
- Plateaus2
Suppose that the hill you are climbing has a plateau lower than the summit, that is, there is a flat area on the side of the hill lower than the peak. If your climb leads you onto the plateau then you're stuck. The hill-climbing algorithm tells you to keep climbing until you can climb no further, then you've reached the solution. So, the algorithm will falsely tell you that the plateau is the peak.
- Ridges2
A ridge is similar to a plateau, except that it's narrow. Ridges fall away on the sides, but in front and back are flat. If you end up on a ridge during your climb, the algorithm will again falsely inform you that you've reached the summit.
- Multiple Peaks3
Some hills and mountains have more than one peak. If your goal is simply to get to any peak, then hill-climbing will work; but if your goal is to get to the highest peak, hill-climbing may lead to a lower one.
As long as the "hill" that you're trying to climb does not have plateaus, ridges, or multiple summits, hill-climbing is guaranteed to get you to the top eventually. However, it's not guaranteed to get you there quickest or by the shortest route: that's a different problem.
But what do you do if the "hill" that you want to climb does have plateaus, ridges, or multiple peaks? Or, what if you don't know whether it has them? I'll address these questions in a future entry.
So far, other than climbing a hill itself, I've only mentioned jigsaw puzzles as a type of recreational puzzle in which hill-climbing is useful. Another familiar type of puzzle which uses hill-climbing is the crossword puzzle. You can't solve a crossword in one stroke of the pencil; instead, you must solve the individual numbered clues until all the blank squares are filled in.
Hill-climbing is another tool in your problem-solving toolbox which now contains, in addition, contraction, thinking backwards, and elimination. Of course, the more tools you have, the more of a problem you'll have selecting the right tool for the job, and none of these tools is right for every problem.
Here's a final problem for this entry: among the puzzles appearing in previous entries in this series4, which ones required hill-climbing to solve?
The "logic" problems from the previous entry on reasoning by elimination. Once the different possible solutions have been identified and listed in a table or grid, eliminating possibilities is a process of hill-climbing. As long as you're crossing out lines in the table, or putting Xs in boxes of the grid, you're making progress towards the solution. Once you've eliminated all but one possibility, you've reached the summit and the solution.
Notes:
- For hill-climbing as a general problem-solving method, see: Wayne A. Wickelgren, How to Solve Problems: Elements of a Theory of Problems and Problem Solving (1974), chapter 5.
- Patrick Henry Winston, Artificial Intelligence (2nd edition 1984), p. 95.
- Wickelgren, pp. 83-84.
- For previous entries in this series, see:
- Contraction, 4/6/2023
- Think Backwards, 5/5/2023
- Solving a Problem by Elimination, 6/20/2023
