This seems to come down to a misunderstanding of what is meant by “science.” Horgan seems to be focused on the big, bombastic, super collider type of science. I can only hope that kind of science winds down sooner rather than later. But the scientific method as a problem solving tool isn’t going anywhere, and i doubt Horgan thinks it is ( but I could be mistaken).
I’m a pool guy, and scientists at Cal Poly San Luis Obispo ran a years long study to determine the actual source of calcium deposits forming on pool plaster. Was it the initial mix when the plaster was applied, or was it due to poor water chemistry maintenance? Pool guys everywhere rejoiced, and Plasterers rent their garments in lamentation when it turned out that it was the mix and not the water chemistry that caused the calcium deposits. Scientists could go on doing this sort of low cost, practically useful science for millennia, no problem.
Yes, I haven't read the book so I don't know what the criteria is for what is and isn't scientific, but currently Wall Street is booming in part due to an advance in AI/large language models. Are LLM's science?
Good post, though, and I laughed out loud at the Ehrlich tweet.
LLMs are no more science than a psychic talk show host is at speaking to your dead father. An electronic calculator is science because you can accurately predict its results and they're infinitely repetable. LLMs produce spooky stuff that is impossible to understand how it got those results or repeat them. LLMs are just the 21st century version of those Mechanical Turks that played chess and wowed the perfumed wig-wearers of European courts.
LLMs are the result of peer reviewed and corporate research (Google's "Attention is all you Need" paper for instance) into natural language, neural networks, and transformer architecture. So they're the applied result of basic research, which is what the original paper is about.
LLMs *are also* the result of a system of normal people going to normal jobs making the overall system run - this does not make it true that LLM's "are plumbing" or "are babysitting".
> So they're the applied result of basic research, which is what the original paper is about.
> Scientists could go on doing this sort of low cost, practically useful science for millennia, no problem.
This very much depends on the definition of "could" - I "could" go to the gym (I am not constrained physically), but I've had a membership for 4 years and have never shown up, because it seems I am not able to make it happen.
I've enjoyed reading your newsletters for a while now but haven't commented before . I'm a retired literature professor who enjoys reading (some) science writing (my dad, the geologist, who got me hooked on science fiction when I was in grade school, is probably the one to credit). I have always remembered a mini-lecture to me about plate tectonics that he gave one day when I was helping him label rocks for a quiz for his first-year class, mostly focusing on how long and difficult a process it was for the theory to be accepted.
The only thing that makes me feel qualified to comment on this post which is great is Horgan's bizarrre doomsday predictions about "science" following the (apparently downward spiralling and failing) paths of "literature, art, music, and philosophy."
Journalists (and others) have been yelling about the demise of literature for decades (in the 1990s, it was accusations of being "politically correct," "cultural Marxists," and oh, yeah, the destruction of "Western Civilization" by us getting rid of Shakespeare in order to teach Alice Walker!
Nowadays, it's accusations of being "woke" and (still) "cultural Marxists" and a complete failure to notice that Shakespeare is still taught and analyzed by scholars all over the world) because everybody is so busy freaking out about how "literature" is no longer limited only to straight white men (not to mention Shakespeare's hots for the young man in the sonnet sequence is openly discussed!).
So Horgan's argument collapses completely for that reason (and I won't be reading his book anytime soon). Although for an interesting approach to geology and literature, let me recommend someone I just discovered through a friend's recommendation: _Romantic Rocks, Aesthetic Geology_ by Noah Heringman because, you know, all these disciplinary boundaries people try to hard to police are also a major problem!
Oh, and my current scholarly projects involve the questions of racisms and J. R. R. Tolkien's legendarium, both the body of work and its reception (including the growing awareness of white supremacist fans of the Legendarium who are throwing temper tantrums all over the internet over the mere idea of a Black elf). So, prejudice, yep, still a relevant topic!
Ah, peer review. Here’s a fun complication: if you work in a niche field, the only likely reviewers of your papers will be people in labs competing directly with yours [1]. The potential is therefore present for the reviewer to torpedo a couple of years of your work because they’re doing the same thing [2] but their paper isn’t quite ready to send out. This was suspected to have occurred a couple of times in my last lab, but is clearly difficult to prove. The mere suspicion of it has consequences in reduced trust and scope for future collaboration.
Oh, and those “secret” lab meetings held in the absence of a postdoc who’s about to join a rival group because the PI doesn’t want a new research interest to get out. Farcical.
[1] The problem here is that pre-emptive publication wins. “We found this, too” or “We found subtle differences” [3] are deemed redundant. Like you say, our worth / livelihoods are dependent on publication, which is a pretty crappy motivation.
[2] If we know A and someone discovers B, everyone will suddenly wonder if A+B=C.
[3] Let’s have a journal of negative results. My publication list would be HUGE if we did.
I *really* expected this title to be referring to metaphorical licking, but no, it's actually physical licking, and the reason why will stick in my mind for quite a while.
"The realists don't have a plan because they can't have a plan." It sounds like Horgan is deep into the status quo mindset. There is no history, only the present and the present is as it is because there was no other way it could have been. This is an incredibly limiting mindset as you rightly point out, it's ahistorical with a reliance on decontextualized data to make points about how the world is, without being able to imagine any possibilities beyond the status quo. This also directly parallels broad sectors of U.S. society (political economy has been obsessed for decades with marginal improvements/efficiency rather than systems change).
a lot of talk about woo-woo physics. string theory was an unfortunate wrong turn that still hasn't been acknowledged as such. peter woit's "not even wrong" blog is never-ending in its snark about this. it was always the next energy level which would reveal the super-symmetric particles predicted at the last energy level. so a lot of math and a lot of handwaving makes that particular part of physics a worthy subject of criticism. maybe something would be revealed if only we could build a collider as wide as the orbit of neptune, but it won't get funded.
more seriously, quantum gravity is an unsolved riddle, and perhaps unsolvable. explanations may lie at the level of planck lengths or perhpas quantum phenomena themselves obscure our abiity to tease it out - i'm really out of depth saying anything at all on this subject.
otoh, other areas of physics such as, e.g. materials science and astrophysics continue to make advances. hogan only wants to talk about the flashy stuff.
as an m.d. i couldn't help but notice the lack of any mention of biology,, and as a psychiatrist i couldn't help but notice the lack of mention of neuroscience in particular. when i was in medical school nerve cell receptors were a theoretical construct to explain the binding of chemicals to vesicles formed by disrupted membranes. now we not only know receptor amino acid sequences, we know that they penetrate the membrane 7 times, and we know how they are folded and how a transmitter is bound and distorts the membrane to affect its permeability. for another example, i have a patient with a serious tremor who has arranged to go to boston to have a very small portion of his brain destroyed as a solution to this problem. over the course of my long career our knowledge and understanding has grown enormously, although what remains unknown is greater still. nonetheless, you'd be hard pressed to argue that biology is at a dead end.
For what it's worth, Horgan has chapters on both neuroscience and biology that I didn't mention. He doesn't think things are better there, but the gist of his argument for biology is "you can only discover evolution once"
again, he wants to point to one big edifice, e.g. evolution. this morning there was an article in the nytimes about the transplantation of a kidney from a genetically modified pig into a human recipient. the genetic modifications involved knocking out 3 genes which produced substances which could trigger rejection, and the insertion of 7 human genes to enhance the probability the kidney would survive, and thus so would the patient. just piffle for hogan, i guess. nothing to see here, move along.
What looks to Horgan like "the end of science" is actually its beginning. Most fields of science -- those that deal with complex systems, open and adaptive ones most of all -- are pre-paradigmatic. As Dennett says, "philosophy is what we do until we figure out the right question to ask"; what he witnessed at SFI is people grappling with what the right question to ask. And because until you figure that out you keep going in circles, it's actually really important to remember what Kant (or Merleau-Ponty) says, because otherwise you do smash yourself into the same boulder as the previous generation.
FWIW, I think we're finally making progress. Stuart Kauffman's recent work on "the adjacent possible" is the philosophical mirror of recent advances like the ones by Michael Levin in biology, Chris Fields in information theory, and Karl Friston in neuroscience -- all pointing to something like a unified account of agency, observation, measurement and meaning. Those are the building blocks for the next thing.
One thing that really drives me nuts is that most people who discuss prejudice don't have an essential understanding of what prejudice actually is. As a computer professional who specializes in Machine Learning, prejudice, to use computer terminology, is a feature, not a bug. Prejudice is a good thing. Inductive inference, or learning, is the ability to generalize from previous experience to future experience. This is done by drawing generalizations. Therefore, when faced with some new experience, we compare a set of parameters that we have determined to be most relevant in this experience and we draw certain expected conclusions. This is the essence of prejudice. Prejudice is not limited to humans, bigoted or not. We all do it. All living things do it. Birds do it, bees do it,even trees do it. And computers do it too. Cognition and the ability to generalize is based on biases. So you can't fix prejudice - actually fixing it is a dumb idea and doomed to failure. What you really need to focus on is not the fact of prejudice, but the ability to reassess and recalibrate as new information comes in. This is sometimes referred to as cognitive bias. if done incorrectly. The essential problem is that we find it difficult to admit when we are wrong in our prejudices in light of further information that makes each new experience unique and somewhat different from the past. The ability to appropriately incorporate new information allows us to see each individual experience for what it is: a case that, if our prejudices are well-founded, is most likely to be supported by further evidence, but with characteristics all its own. These differences may be minor, or they may just be a one-off counter-example such as the rare black swan. But if there are enough counter-examples, this indicates that we need to adjust our prejudices. Not dicard prejudice, mind, but change them so as to change our expectations.
I think the word you should be using is "discrimination" rather than "prejudice". In colloquial language, they are often used interchangeably. But discrimination is essential, while prejudice is irrational.
Generalization, classification, etc. from observations rely on discrimination, not prejudice, which is defined as "preconceived opinion that is not based on reason or actual experience."
I am trying to argue that this distinction is incorrect.
Thus, a wine connoisseur is known for their discrimination. But if they reject Ripple out of hand, is this prejudice?
I want to analyze the nature of prejudice. What is going on that makes it bad and discrimination good. I don't think it is helpful to make the problem go away by defining terms: using the word "discrimination" instead of "prejudice".
One big problem is that we tend to associate certain behaviors and the words that describe them with connotations that are actually are not an essential feature of that named behavior. What I want to do here is to use a bare-bones definition of "prejudice". Prejudice is the act of classifying something and inferring characteristics about that something that have not yet been observed. Based on my background in Machine Learning, I claim that prejudicing is done in the following manner:
1. At a given time, we have developed a library of general classifications.
2. Given a novel input, we determine some characteristics of this input.
3. We then classifiy this input as a member of one of our given general classifications
4. We prejudge this input to also have a number of other characteristics, because our experience has shown this to be generally true in the past.
I make the following claims:
A. There is nothing essentially irrational about prejudice. The act of discrimination is just as likely as the act of prejudice to be irrational.
B. Preconceived opinions that are not based on reason or actual experience are probably based on a person's emotions, which is determined genetically. Even if the emotional outlook came through the accumulation of lived experience, this is an act of generalization based on actual experience.
I urge you to put aside the attempt to make a distinction between "discrimination" and "prejudice" just because the emotional valence in colloquial usage is different (discrimination = Good, prejudice = Bad).
Lets explore deeper:
You claim discrimination is essential but prejudice is not. I claim that the two terms refer to the same action outside of the emotional valence. My claim that someone makes a discrimination when they draw a conclusion you agree with, and they are prejudiced when they draw a conclusion you don't like. I claim that they are used interchangeably because they are in fact interchangeable in terms of their process. The difference is in your agreement with the outcome.
You claim prejudice is irrational but discrimination is rational. I don't see how that conclusion can be drawn. Please justify that claim. I can well imagine that prejudice is based on personal experience and is based on a set of rational conclusions. I can also well imagine that said wine connoisseur has an emotional snobbery about their discrimination against certain kinds of wines that is irrational.
You claim that prejudice is not based on experience. Nor is it based on reasoning. Then what is it based on?
Personally, I will stick with my main thesis: the words discrimination and prejudice are different words for the same thing. Discrimination on the basis of race is just another term for racial prejudice. Discrmination for or against a certain type of wine can be considered an act of prejudice if you are drawing conclusions based on the label before you have even drawn out the cork (or twisted off the top). Instead, the problem is further downstream: the unwillingness or inability to update your classification (or discirmination or prejudice) based on further information that either shows that this particular individual does not fit all of the inferred properties of the classification, or that your generalized class is in error and your models need to be updated.
You are trying to redefine perfectly good and useful concepts of the English language in a bizarre and idiosyncratic manner for which there is no need.
The definition of prejudice that I quoted is not mine; I just grabbed it from an online dictionary. And I made no claim that discrimination is always rational; it can be rational or irrational. Indeed, one can say that prejudice, a.k.a. bigotry, is a form of irrational discrimination. Each of these words has its distinct uses. There is no need to conflate them, or to use one in a context for which the other is better suited. "Discrimination" fits the usage in your original comment better than "prejudice" does.
You are saying this as if redefinition is a bad thing.
This is getting rather discouaging. My use of the terms, while idiosyncratic, is certainly not bizarre. The need for my redefinition is to point out that the current consensus could well be misplaced or in error.
For example, you conflate prejudice as irrational discrimination. I am putting forth the hypothesis that prejudice is not alwayss irrational of necessity and that the real problem with prejudice is later on, with the inability to change one's mind.
One of the problems with language is that, in its ability to communicate thought from one person to another, it also constrains thought. You end up being forced into thinking of a problem a certain way because of the extra connotations that go along with words in common usage. No wonder people like Whitehead coined new terms when he invented the concept of Process Metaphysics. If he had used the regular terms, no doubt people would have been arguing vocabulary instead of coming to terms with the ideas he was presenting.
No: discrimination does not fit my original comment better then prejudice does because my original comment was directed to what is commonly considered to be prejudice, not discrimination in general.
What would you have me do? Coin a new word "grebish" and say that "grebish" refers to any and only any act of prejudice, but I am using a different word because I am claiming that the root cause analysis of the nature of prejudice (such that it is claimed to be irrational) could well be in error and that there is another explanation that is connoted by the word "grebish" even though the word "grebish" and "predjudice" apply to the exact same events?
You objection is cause to suggest anotherr hypothesis: Adam claims that science is a strong-link problem.
This points to the fact that most science does not move a field forward all that much. But the exceptional papers make a big difference. It may well be true that the run-of-the-mill paper uses concepts and terminology in the standard ways with the distinct usage that is in common practice, and that the important and creative papers are more likely to use common terminology in bizarre and idiosyntratic ways. Adam draws a comparison to music that is helpful to bring out this point: some of the greatest music changes the current music idiom and leads us to hear things in new ways.
But talking about language instead of the underlying concepts misses the point: what is the nature of prejudice, whatever you call it?
Addendum: This problem with language is something that seems to come up a lot. Just recently, I was having a debate with someone on a particular issue in genomics: a paper I published in the Journal of Theoretical Biology laid out a hypothesis that transposons have the purose of helping to define three-dimensional structures of body parts. This other person had difficulty accepting this because the common connotation of transposons is that they are parasitical and do not play a helpful part in the biology of animals and plants. This was a case that the "perfectly good and useful" concepts associated with transposons made it difficult to see that there are other ways of looking at them. And, about 25 years ago, I was discussing with a highly regarded computer scientist about a paper I had written on the problem of overfitting in Machine Learning had had been rejected for publication. I had a nice mathematical result on the inevitability of overfitting based on Ramsey Theory. His response was that I would have better success getting published if I used the commonly accepted definition of over-fitting instead of my idiosyncratic usage. But after looking at the generally accepted definition, you could tell that my result was not derivable from that definition. Both cases just go to show that when you use the commonly accepted distinct usages, you end up with the commonly accepted conclusions: you don't end up with anything really new, you just end up repeating the same old tropes.
It seems to me that there are people who are able to see the problems language cause clearly and intuitively, and then there is the other 98% of people.....and, most "genuinely" smart people (on a relative scale, *because that's how we're trained to perceive "reality") are within the 98%. Should this bizarre resulting state of affairs be that surprising after running such a simulation for decades?
This question of the use of the word "prejudice" versus "discrimination" in the context of redefinition is worth exploring further.
In general, when a concept relating to some human state of being is defined, the essential features of the concept and the resultant connotations often fall into the following categories:
A set of characteristics, such as the symptoms that characterize a disease.
The causes that brought this state of being into existence
The causes that would result in the resolution of change of this state of being
The outcomes that result from this state of being
Any evaluations, judgemental or moral, or emotional states in response as a valence or sentiment, such as whether this state of being is good or bad.
I used the word "prejudice" deliberately and intend to consider prejudice to be a bare-bones case of pre-judging some aspects of a situation beyond the aspects that led to the classification. But I also want to attach the judgment that prejudice is considered to be a "bad thing" - something to avoid or ameliorate.
Discrimination does not fit either the characteristics or the judgment as well as prejudice does.
Discrimination is the process of making a distinction. But prejudice is more than that. Once you have made a distinction, prejudice kicks in when you draw inferences about the situation that are not directly observed. Instead, they are believed to be associated with the situation.
Also, discrimination does not always have a negative judgment.
A wine connoisseur shows discrimination in (hopefully) a good way: by upholding the best quality wines and denigrating the inferior plonk.
Racial discrimination is usually considered a bad thing, quite often because it involves prejudice.
Someone attuned to sight and sound, such as the ability to discriminate between different shades of color, is neither considered good nor bad - it just is.
Prejudice, though, is considered to be a type of sociopathy.
And here is where the slavish adherence to "common usage" can get you into trouble.
Let's draw a comparison to medical pathologies - diseases.
A stomach ulcer is simply a wound (Greek elkos = wound). But there was a connotation associated with it in relation to its cause. It was assumed to be due to stress. Marshall and Warren back in 1982 proposed that the real cause was H. Pylori infection, but the connotation of stress as a causal agent was so strong that it took a decade before people accepted this.
Sometimes, the name of a disease is the name of the cause. For example, as we can see by the name of the disease, "malaria" = "bad air" is caused by miasma. Of course, we know that to be false now.
This is the problem around the analysis of prejudice. The connotation that you mention is that prejudice is irrational. But it may well be very rational. My thesis is that the "pre-judging" aspect of prejudice is fundamental to human cognition, to the point where some prejudices are genetic. These prejudices are (or were) good for you, which is why they are not termed prejudice in the derogatory use of the term.
For example, we humans have a natural prejudice against predators such as the wolf or the lion. We observe an animal and make the discrimination of it as a wolf. Our pre-judging is that getting too close can be dangerous. Similarly, we have a prejudice against spiders and snakes.
So, I claim that prejudice is neither irrational nor inherently bad. It is a ubiquitous part of human reasoning. What makes the cases that we denigrate as "prejudice" (bad, sociopathic) may come not from the initial discrimination nor from the consequential conclusions of secondary characteristics and connotations, but in our inability to "update our priors" - to either fail to say that our assumptions do not apply in a particular case, or that our generalizations are misplaced.
I love comprehensive claims of nonexistence - have you a proof for this fact, or is it more of a faith based belief, or a colloquialism (that shall not be restated in meaningful language, because "that's pedantic")?
> You claim that prejudice is not based on experience. Nor is it based on reasoning. Then what is it based on?
If you pushed him into a corner you'd eventually find he's dealing with a tautological truth - all exceptions to it do not count, *because they would then no longer be a member of the class*. You can see the same behavior in all ideological classifications, like science fans who claim science is perfect, because any wrongdoing by a scientists *is therefore not science*.
21st Century Western Consciousness is a hell of a drug.
What you're talking about is generalizations based on past experiences. Prejudice,however , is the set of generalization rules that fail to update when new data arrives. It is the set of degenerate weights that were overfitted on irrelevant data and just wouldn't budge; the pixel in the center of the image that makes it impossible to distinguish between plane and monkey.
Thanks. This is an interesting response. To respond in turn, I am going to get meta on you.
First, let's try to define what is meant by prejudice and give some examples.
By prejudice, I mean that, given a situation where a person observes an entity with defining characteristics A, B and C, they classify this entity as an example of Class X and infer secondary characteristics D, E and F.
For example:
It is prejudicial to assume that, if a person is a conservative, they are opposed to abortion, and if they are a liberal, they are pro-choice, even though this happens quite often.
It is prejudicial to assume that, if a person is black, they are good at basketball, even though the best players in the National Basketball Association are quite often black.
It is prejudicial to assume that, if a person is Asian, they have high SAT scores, even though Asians tend to have higher scores than other groups.
Therefore, to prejudge is to draw an association, based on experience or belief, of a trait not yet observed, based on the defining characteristics of a class of entities. This is also termed "connotation":
"The scientific method is an iterative, cyclical process through which information is continually revised.[l][47][48] It is generally recognized to develop advances in knowledge through the following elements, in varying combinations or contributions:[49][50]
Characterizations (observations, definitions, and measurements of the subject of inquiry)
Hypotheses (theoretical, hypothetical explanations of observations and measurements of the subject)
Predictions (inductive and deductive reasoning from the hypothesis or theory)
Experiments (tests of all of the above)"
Therefore, my claim is that prejudice usually connotes an error or an overgeneralization, but in actuality, a prejudice may be well-founded and appropriate.
Seen this way, Halley's prediction of the return of his Comet using Newton's theory is a form of prejudice as much as the prediction that a black person is more likely to listen to soul or rap rather than Rossinni's operas.
So you say:
"What you're talking about is generalizations based on past experiences. Prejudice, however, is the set of generalization rules that fail to update when new data arrives. It is the set of degenerate weights that were overfitted on irrelevant data and just wouldn't budge; the pixel in the center of the image that makes it impossible to distinguish between plane and monkey."
So we are in agreement that I am claiming that prejudice is generalizations based on past experience. Where we differ is whether or not prejudice involves failing to update when new data arrives.
Where this gets meta is that I would claim that this second statement is itself prejudicial. Not all prejudices fail to be updated. This happens sometimes, especially if there is an emotional valence attached to the prejudices that make it hard to give up those beliefs. I point this out because there typically is a strong emotional reaction when the word "prejudice" is mentioned. This obscures the fact that prejudice can often be overcome when someone is presented with disconfirming evidence.
There are actually two different ways to address our differences. Although I am claiming that prejudices occur before the act of failure or success in updating the generalization, another approach is to claim that my definition of prejudice is overly broad and that "failure to update" is a defining characteristic.
But that is an unpalatable approach. Going about redefining terms so as to justify our prejudices is unfortunately all too common. That would be like arguing that if someone from China got low SAT scores, then they are "not really Asian".
But a major point that your objection obscures is the common case where a prejudice is actually well-founded, but does not apply in a particular situation. What I was trying to point out is that not everyone fits the same mold. This is not related to updating the model (your reference to "weights" applies to the current fad of Neural Nets - having worked in Computational Learning Theory, I am more interested in learning in general). Instead, I was trying to get at a more subtle point. In applying a general classification rule, it is important to keep in mind that, even though an entity fits a class, this does not mean that all secondary characteristics are true in every case. What this means is that it could be true that your prejudices are well-founded and essentially correct, but in each particular case there are unique differences.
We obviously agree on the fundamental level. And I also agree that "failed to update" is not exhausting the possible scenarios linking prediction to prejudice. I just grabbed the opportunity and used the analogy to show why I think that it is not equivalent to prediction (of course bad predictions are a subset of predictions, so we may also start discussing what is "equivalent" 😇).
Even if it isn't so on the pure semantic level (I am not capable of judging this), in its global use the word "prejudice" carries the meaning of "judging before the (relevant) facts are known". More abstractly, this translates to " assigning value based on outdated, incomplete or irrelevant data". Also , in every day use at least, prejudices are generally thought to be rigid and not easy to change and shift (which is evident e.g. by the often use of the "no true Scotsman"-fallacy to harden one's prejudice once they were shown wrong, as you also pointed it out). Given those two properties of prejudice - "bad" data and rigidness, I think that, while not exhausting, my analogy was pretty precise. At least considering the general use of the word. 😎
This gives us the grounds for an interesting discussion.
One reason that I am trying to reformulate the concept of prejudice so as to be closer to the Machine Learning concept of generalization (besides my being a computer geek) is that this type of reframing can be helpful sometimes by allowing us to see things in a different way. Of course, we should always stick to reality. I am not trying to present "alternate facts". What I hope I can do is to look at prejudice in a way that help ameliorate the rigidity and other negative associations.
I used the term "connotation" advisedly. What I want to get at is to question the "everyday" use: the notion of regidity is considered an essential part of prejudice. If we use this everyday way of looking at things, we may be making a judgment call about the nature of prejudice. We are implying that we have a lack of agency in our prejudice. Thinking of prejudice in our current everyday way, as involving "bad" data and rigidness, means that we may be drawing implications that are not generally true. Which was why I said I was going to get meta on you: using the general use of the term makes you prejudiced angainst prejudice. I am trying to get us away from that.
My claim is that using "bad" or outdated data and sticking rigidly to a viewpoint is not an essential feature of prejudice. Being prejudicial about prejudice in this way is Calvinistic: the prejudiced amongst us are doomed to their incapacity.
But this is not necessarily shown to be true in reality. For instance, people who are prejudiced against a group of people tend to be less prejudiced the more thery actually interact with that group. This implies that they may not be so rigid. People who rely on outdated data may actually be conservative. This reliance is actually sketicism about new-fangled ways: they will give up their reliance if you can prove the new way is better.
So I am arguing that the general use of the term needs to be discarded. To get meta again, thinking of prejudice this way is outdated, incomplete and irrelevant. Outdated, because studies have shown the ability of prejudical people to change if given the chance. Incomplete in the sense that not all prejudice is rigid.
And I claim that these extra connotations are irrelevant. Worse, they make us focus on the wrong problem. A point I want to make is that prejudice is a good thing: it is an essential heuristic for living. Considering prejudice as just another way of drawing generalizations shows that it is the natural and default approach to experinecing new things. That allows us to focus on the rigidiy and the overly conservative clinging to bad or outmoded data as the essential problem, but not an essential part of prejudice itself. My reframing has split off the negative connotation from what is good in prejudice allowing us to keep the good and avoid the bad. It also lessens the tendency to separate people into "good" and "bad" people in terms of their degree of prejudice. Like the song goes in Avenue Q: "everybody is a little bit racist". I explain that by noting that we go through life effieicently, drawing quick generalizations as a matter of course, and only going back and rethinking our prejudices if we have to. Reframing things this way makes it less likely that we overthink things *(like what happened recently with the ACLU) and it also acknowledges that perfection is unattainable and is, in fact, not desireable. Prejudice is a heuristic and heuristics by their nature are subject to error. But they are computationally efficient and "good enough for government work". You put in the extra effort when the need arises.
Admittedly my reframing can come up against our prejudices about prejudice, which we have invested a whole lot of emotional baggage. So, yes, your analogy was a proper characterization of prejudice the way the currently think of it. I am advocating that we have an attitude adjustment when it comes to prejudice: consider prejudice as a given and not inherently bad. Instead focus on the real problem: clinging rigidly to outdated, incomplete or irrelevant data. This points to an essential feature of that problem: we all have a tendency to want to have to aviod having to admit we were wrong.
What do I mean by reality? Wow, that's a tough one.
"Reality, what a concept" - Mork from Ork
A glib answer is that reality is the pain you feel when you decide that the Theory of Gravity is just a theory the way that the Theory of Evolution is, and decide to defy it by jumping out the second floor window.
To get beyond a glib answer takes you down the rabbit-hole. Philosophically I am a Taoist. Taoism teaches that human understanding is a phenomenological substitute for the underlying noumena - the Tao.
Some thoughts I have about the nature of reality are written up in an essay "Hylomorphic Functions"
This work attempts to relate metaphysics, like that of Aristotle, to physics, especially Quantum Mechanics.
For the purposes of this discussion, reality is a kind of "ground truth" where you can test your concepts and observe how well they work, both descriptively and predictively. What I mean by that is that your ideas must be testable. One of the most influential essays on my way of thinking about science is Richard Feynman's Cargo Cult Science
Unfortunately, I have a problem with semiotics. Pierce is considered the founder of that field and I have the highest regard for Pierce. There are other people who I respect are enamoured of semiotics and write about it. But whenever I try to understand semiotics, I could never figure out what it is about. And, I confess to admit, those times I have read articles about semiotics, the results I can understand turn out to be commonplace. Maybe I'm stupid - I don't know. But semiotics just seems to elude me. I guess you could say I can't see what it signifies.
"The strictest limits are inside our own minds—the questions we never ask because they seem too obvious, the explanations we accept without sufficient evidence, the studies we never run because they’re too far outside the consensus."
From my own interests, I think that gender may be important to the evolution of language. A few linguists and anthropologists have argued the same, but a big hurdle has been that sex differences in the brain are taboo. For my money, that taboo set us back decades in understanding who we are and where we came from.
The problem of the book is who is interviewed: only people who the author has heard about as journalist (a superficial profession), which means people who say crazy, attention grabbing things. As to Wittgenstein, it’s because his work is word salade that everyone (some people) like him so much.
I fail to understand your comment. But, if anything, I was raised in a culture reasonably close to that of Wittgenstein. But I believe only in science as a method of knowledge, but not in culture.
You are perfectly entitled to like Wittgenstein. And you are perfectly entitled to think that it is not word salad. But I think it is word salad, which is why I thought that I could infer that some people like it because word salad can be interpreted in every which way, and if you interpret it in a way you find interesting you will like Wittgenstein. Note, however, that I am also entitled to have an opinion about Wittgenstein and no appeal to the culture of Wittgenstein or the culture of Wittgenstein lovers can change that.
You call Wittgenstein "word salad", but then you state your opinions as facts - is this not rather paradoxical?
Do you believe that it may be possible that at runtime, you may sometimes forget that your opinions are facts?
Do you ever wonder what is true?
> Note, however, that I am also entitled to have an opinion about Wittgenstein and no appeal to the culture of Wittgenstein or the culture of Wittgenstein lovers can change that.
Are you claiming that *it is not possible* for your opinion of Wittgenstein to be changed by external forces?
What I think needs to be recognized is that often, due to frustrations with the limits of science (as well as its previous achievements), believers in scientific orthodoxy go rogue in order to gain influence (or funding), notoriety, or because they just can't help themselves.
If you think of science from the perspective of yet another human ideology, it shouldn't be too surprising that the faithful are going to misbehave, hallucinate, etc...it goes with the territory of humans!
This seems to come down to a misunderstanding of what is meant by “science.” Horgan seems to be focused on the big, bombastic, super collider type of science. I can only hope that kind of science winds down sooner rather than later. But the scientific method as a problem solving tool isn’t going anywhere, and i doubt Horgan thinks it is ( but I could be mistaken).
I’m a pool guy, and scientists at Cal Poly San Luis Obispo ran a years long study to determine the actual source of calcium deposits forming on pool plaster. Was it the initial mix when the plaster was applied, or was it due to poor water chemistry maintenance? Pool guys everywhere rejoiced, and Plasterers rent their garments in lamentation when it turned out that it was the mix and not the water chemistry that caused the calcium deposits. Scientists could go on doing this sort of low cost, practically useful science for millennia, no problem.
Yes, I haven't read the book so I don't know what the criteria is for what is and isn't scientific, but currently Wall Street is booming in part due to an advance in AI/large language models. Are LLM's science?
Good post, though, and I laughed out loud at the Ehrlich tweet.
LLMs are no more science than a psychic talk show host is at speaking to your dead father. An electronic calculator is science because you can accurately predict its results and they're infinitely repetable. LLMs produce spooky stuff that is impossible to understand how it got those results or repeat them. LLMs are just the 21st century version of those Mechanical Turks that played chess and wowed the perfumed wig-wearers of European courts.
LLMs are the result of peer reviewed and corporate research (Google's "Attention is all you Need" paper for instance) into natural language, neural networks, and transformer architecture. So they're the applied result of basic research, which is what the original paper is about.
LLMs *are also* the result of a system of normal people going to normal jobs making the overall system run - this does not make it true that LLM's "are plumbing" or "are babysitting".
> So they're the applied result of basic research, which is what the original paper is about.
"About" is a similarly misleading word.
> Scientists could go on doing this sort of low cost, practically useful science for millennia, no problem.
This very much depends on the definition of "could" - I "could" go to the gym (I am not constrained physically), but I've had a membership for 4 years and have never shown up, because it seems I am not able to make it happen.
I've enjoyed reading your newsletters for a while now but haven't commented before . I'm a retired literature professor who enjoys reading (some) science writing (my dad, the geologist, who got me hooked on science fiction when I was in grade school, is probably the one to credit). I have always remembered a mini-lecture to me about plate tectonics that he gave one day when I was helping him label rocks for a quiz for his first-year class, mostly focusing on how long and difficult a process it was for the theory to be accepted.
The only thing that makes me feel qualified to comment on this post which is great is Horgan's bizarrre doomsday predictions about "science" following the (apparently downward spiralling and failing) paths of "literature, art, music, and philosophy."
Journalists (and others) have been yelling about the demise of literature for decades (in the 1990s, it was accusations of being "politically correct," "cultural Marxists," and oh, yeah, the destruction of "Western Civilization" by us getting rid of Shakespeare in order to teach Alice Walker!
Nowadays, it's accusations of being "woke" and (still) "cultural Marxists" and a complete failure to notice that Shakespeare is still taught and analyzed by scholars all over the world) because everybody is so busy freaking out about how "literature" is no longer limited only to straight white men (not to mention Shakespeare's hots for the young man in the sonnet sequence is openly discussed!).
So Horgan's argument collapses completely for that reason (and I won't be reading his book anytime soon). Although for an interesting approach to geology and literature, let me recommend someone I just discovered through a friend's recommendation: _Romantic Rocks, Aesthetic Geology_ by Noah Heringman because, you know, all these disciplinary boundaries people try to hard to police are also a major problem!
Oh, and my current scholarly projects involve the questions of racisms and J. R. R. Tolkien's legendarium, both the body of work and its reception (including the growing awareness of white supremacist fans of the Legendarium who are throwing temper tantrums all over the internet over the mere idea of a Black elf). So, prejudice, yep, still a relevant topic!
Link to the book: https://www.cornellpress.cornell.edu/book/9780801441271/romantic-rocks-aesthetic-geology/
Ah, peer review. Here’s a fun complication: if you work in a niche field, the only likely reviewers of your papers will be people in labs competing directly with yours [1]. The potential is therefore present for the reviewer to torpedo a couple of years of your work because they’re doing the same thing [2] but their paper isn’t quite ready to send out. This was suspected to have occurred a couple of times in my last lab, but is clearly difficult to prove. The mere suspicion of it has consequences in reduced trust and scope for future collaboration.
Oh, and those “secret” lab meetings held in the absence of a postdoc who’s about to join a rival group because the PI doesn’t want a new research interest to get out. Farcical.
[1] The problem here is that pre-emptive publication wins. “We found this, too” or “We found subtle differences” [3] are deemed redundant. Like you say, our worth / livelihoods are dependent on publication, which is a pretty crappy motivation.
[2] If we know A and someone discovers B, everyone will suddenly wonder if A+B=C.
[3] Let’s have a journal of negative results. My publication list would be HUGE if we did.
I *really* expected this title to be referring to metaphorical licking, but no, it's actually physical licking, and the reason why will stick in my mind for quite a while.
Dude, stop ruining my PhD! Whenever I get an email from your substack, I immediately have to stop writing my thesis to read it cuz it's SO GOOD!
By the way, I think you should eventually assemble these posts into a neat book with a nice name.
You know, like "The End of Science", but better.
Title suggestion: "Reinventing Science"
"The realists don't have a plan because they can't have a plan." It sounds like Horgan is deep into the status quo mindset. There is no history, only the present and the present is as it is because there was no other way it could have been. This is an incredibly limiting mindset as you rightly point out, it's ahistorical with a reliance on decontextualized data to make points about how the world is, without being able to imagine any possibilities beyond the status quo. This also directly parallels broad sectors of U.S. society (political economy has been obsessed for decades with marginal improvements/efficiency rather than systems change).
It’s Poughkeepsie, not Poughkipsee. Haha! Your argument is thereby invalidated!
devastating, I thought spellcheck didn't like it because the town is too inconsequential to be in the dictionary
Very amusing! Always love reading your thoughts!
a lot of talk about woo-woo physics. string theory was an unfortunate wrong turn that still hasn't been acknowledged as such. peter woit's "not even wrong" blog is never-ending in its snark about this. it was always the next energy level which would reveal the super-symmetric particles predicted at the last energy level. so a lot of math and a lot of handwaving makes that particular part of physics a worthy subject of criticism. maybe something would be revealed if only we could build a collider as wide as the orbit of neptune, but it won't get funded.
more seriously, quantum gravity is an unsolved riddle, and perhaps unsolvable. explanations may lie at the level of planck lengths or perhpas quantum phenomena themselves obscure our abiity to tease it out - i'm really out of depth saying anything at all on this subject.
otoh, other areas of physics such as, e.g. materials science and astrophysics continue to make advances. hogan only wants to talk about the flashy stuff.
as an m.d. i couldn't help but notice the lack of any mention of biology,, and as a psychiatrist i couldn't help but notice the lack of mention of neuroscience in particular. when i was in medical school nerve cell receptors were a theoretical construct to explain the binding of chemicals to vesicles formed by disrupted membranes. now we not only know receptor amino acid sequences, we know that they penetrate the membrane 7 times, and we know how they are folded and how a transmitter is bound and distorts the membrane to affect its permeability. for another example, i have a patient with a serious tremor who has arranged to go to boston to have a very small portion of his brain destroyed as a solution to this problem. over the course of my long career our knowledge and understanding has grown enormously, although what remains unknown is greater still. nonetheless, you'd be hard pressed to argue that biology is at a dead end.
For what it's worth, Horgan has chapters on both neuroscience and biology that I didn't mention. He doesn't think things are better there, but the gist of his argument for biology is "you can only discover evolution once"
again, he wants to point to one big edifice, e.g. evolution. this morning there was an article in the nytimes about the transplantation of a kidney from a genetically modified pig into a human recipient. the genetic modifications involved knocking out 3 genes which produced substances which could trigger rejection, and the insertion of 7 human genes to enhance the probability the kidney would survive, and thus so would the patient. just piffle for hogan, i guess. nothing to see here, move along.
Your performance was exemplary until that last sentence.
Don't forget that consciousness is always lurking undetected in the background, messing things up.
What looks to Horgan like "the end of science" is actually its beginning. Most fields of science -- those that deal with complex systems, open and adaptive ones most of all -- are pre-paradigmatic. As Dennett says, "philosophy is what we do until we figure out the right question to ask"; what he witnessed at SFI is people grappling with what the right question to ask. And because until you figure that out you keep going in circles, it's actually really important to remember what Kant (or Merleau-Ponty) says, because otherwise you do smash yourself into the same boulder as the previous generation.
FWIW, I think we're finally making progress. Stuart Kauffman's recent work on "the adjacent possible" is the philosophical mirror of recent advances like the ones by Michael Levin in biology, Chris Fields in information theory, and Karl Friston in neuroscience -- all pointing to something like a unified account of agency, observation, measurement and meaning. Those are the building blocks for the next thing.
One thing that really drives me nuts is that most people who discuss prejudice don't have an essential understanding of what prejudice actually is. As a computer professional who specializes in Machine Learning, prejudice, to use computer terminology, is a feature, not a bug. Prejudice is a good thing. Inductive inference, or learning, is the ability to generalize from previous experience to future experience. This is done by drawing generalizations. Therefore, when faced with some new experience, we compare a set of parameters that we have determined to be most relevant in this experience and we draw certain expected conclusions. This is the essence of prejudice. Prejudice is not limited to humans, bigoted or not. We all do it. All living things do it. Birds do it, bees do it,even trees do it. And computers do it too. Cognition and the ability to generalize is based on biases. So you can't fix prejudice - actually fixing it is a dumb idea and doomed to failure. What you really need to focus on is not the fact of prejudice, but the ability to reassess and recalibrate as new information comes in. This is sometimes referred to as cognitive bias. if done incorrectly. The essential problem is that we find it difficult to admit when we are wrong in our prejudices in light of further information that makes each new experience unique and somewhat different from the past. The ability to appropriately incorporate new information allows us to see each individual experience for what it is: a case that, if our prejudices are well-founded, is most likely to be supported by further evidence, but with characteristics all its own. These differences may be minor, or they may just be a one-off counter-example such as the rare black swan. But if there are enough counter-examples, this indicates that we need to adjust our prejudices. Not dicard prejudice, mind, but change them so as to change our expectations.
I think the word you should be using is "discrimination" rather than "prejudice". In colloquial language, they are often used interchangeably. But discrimination is essential, while prejudice is irrational.
Generalization, classification, etc. from observations rely on discrimination, not prejudice, which is defined as "preconceived opinion that is not based on reason or actual experience."
I am trying to argue that this distinction is incorrect.
Thus, a wine connoisseur is known for their discrimination. But if they reject Ripple out of hand, is this prejudice?
I want to analyze the nature of prejudice. What is going on that makes it bad and discrimination good. I don't think it is helpful to make the problem go away by defining terms: using the word "discrimination" instead of "prejudice".
One big problem is that we tend to associate certain behaviors and the words that describe them with connotations that are actually are not an essential feature of that named behavior. What I want to do here is to use a bare-bones definition of "prejudice". Prejudice is the act of classifying something and inferring characteristics about that something that have not yet been observed. Based on my background in Machine Learning, I claim that prejudicing is done in the following manner:
1. At a given time, we have developed a library of general classifications.
2. Given a novel input, we determine some characteristics of this input.
3. We then classifiy this input as a member of one of our given general classifications
4. We prejudge this input to also have a number of other characteristics, because our experience has shown this to be generally true in the past.
I make the following claims:
A. There is nothing essentially irrational about prejudice. The act of discrimination is just as likely as the act of prejudice to be irrational.
B. Preconceived opinions that are not based on reason or actual experience are probably based on a person's emotions, which is determined genetically. Even if the emotional outlook came through the accumulation of lived experience, this is an act of generalization based on actual experience.
I urge you to put aside the attempt to make a distinction between "discrimination" and "prejudice" just because the emotional valence in colloquial usage is different (discrimination = Good, prejudice = Bad).
Lets explore deeper:
You claim discrimination is essential but prejudice is not. I claim that the two terms refer to the same action outside of the emotional valence. My claim that someone makes a discrimination when they draw a conclusion you agree with, and they are prejudiced when they draw a conclusion you don't like. I claim that they are used interchangeably because they are in fact interchangeable in terms of their process. The difference is in your agreement with the outcome.
You claim prejudice is irrational but discrimination is rational. I don't see how that conclusion can be drawn. Please justify that claim. I can well imagine that prejudice is based on personal experience and is based on a set of rational conclusions. I can also well imagine that said wine connoisseur has an emotional snobbery about their discrimination against certain kinds of wines that is irrational.
You claim that prejudice is not based on experience. Nor is it based on reasoning. Then what is it based on?
Personally, I will stick with my main thesis: the words discrimination and prejudice are different words for the same thing. Discrimination on the basis of race is just another term for racial prejudice. Discrmination for or against a certain type of wine can be considered an act of prejudice if you are drawing conclusions based on the label before you have even drawn out the cork (or twisted off the top). Instead, the problem is further downstream: the unwillingness or inability to update your classification (or discirmination or prejudice) based on further information that either shows that this particular individual does not fit all of the inferred properties of the classification, or that your generalized class is in error and your models need to be updated.
You are trying to redefine perfectly good and useful concepts of the English language in a bizarre and idiosyncratic manner for which there is no need.
The definition of prejudice that I quoted is not mine; I just grabbed it from an online dictionary. And I made no claim that discrimination is always rational; it can be rational or irrational. Indeed, one can say that prejudice, a.k.a. bigotry, is a form of irrational discrimination. Each of these words has its distinct uses. There is no need to conflate them, or to use one in a context for which the other is better suited. "Discrimination" fits the usage in your original comment better than "prejudice" does.
You are saying this as if redefinition is a bad thing.
This is getting rather discouaging. My use of the terms, while idiosyncratic, is certainly not bizarre. The need for my redefinition is to point out that the current consensus could well be misplaced or in error.
For example, you conflate prejudice as irrational discrimination. I am putting forth the hypothesis that prejudice is not alwayss irrational of necessity and that the real problem with prejudice is later on, with the inability to change one's mind.
One of the problems with language is that, in its ability to communicate thought from one person to another, it also constrains thought. You end up being forced into thinking of a problem a certain way because of the extra connotations that go along with words in common usage. No wonder people like Whitehead coined new terms when he invented the concept of Process Metaphysics. If he had used the regular terms, no doubt people would have been arguing vocabulary instead of coming to terms with the ideas he was presenting.
No: discrimination does not fit my original comment better then prejudice does because my original comment was directed to what is commonly considered to be prejudice, not discrimination in general.
What would you have me do? Coin a new word "grebish" and say that "grebish" refers to any and only any act of prejudice, but I am using a different word because I am claiming that the root cause analysis of the nature of prejudice (such that it is claimed to be irrational) could well be in error and that there is another explanation that is connoted by the word "grebish" even though the word "grebish" and "predjudice" apply to the exact same events?
You objection is cause to suggest anotherr hypothesis: Adam claims that science is a strong-link problem.
https://www.experimental-history.com/p/science-is-a-strong-link-problem
This points to the fact that most science does not move a field forward all that much. But the exceptional papers make a big difference. It may well be true that the run-of-the-mill paper uses concepts and terminology in the standard ways with the distinct usage that is in common practice, and that the important and creative papers are more likely to use common terminology in bizarre and idiosyntratic ways. Adam draws a comparison to music that is helpful to bring out this point: some of the greatest music changes the current music idiom and leads us to hear things in new ways.
But talking about language instead of the underlying concepts misses the point: what is the nature of prejudice, whatever you call it?
Addendum: This problem with language is something that seems to come up a lot. Just recently, I was having a debate with someone on a particular issue in genomics: a paper I published in the Journal of Theoretical Biology laid out a hypothesis that transposons have the purose of helping to define three-dimensional structures of body parts. This other person had difficulty accepting this because the common connotation of transposons is that they are parasitical and do not play a helpful part in the biology of animals and plants. This was a case that the "perfectly good and useful" concepts associated with transposons made it difficult to see that there are other ways of looking at them. And, about 25 years ago, I was discussing with a highly regarded computer scientist about a paper I had written on the problem of overfitting in Machine Learning had had been rejected for publication. I had a nice mathematical result on the inevitability of overfitting based on Ramsey Theory. His response was that I would have better success getting published if I used the commonly accepted definition of over-fitting instead of my idiosyncratic usage. But after looking at the generally accepted definition, you could tell that my result was not derivable from that definition. Both cases just go to show that when you use the commonly accepted distinct usages, you end up with the commonly accepted conclusions: you don't end up with anything really new, you just end up repeating the same old tropes.
It seems to me that there are people who are able to see the problems language cause clearly and intuitively, and then there is the other 98% of people.....and, most "genuinely" smart people (on a relative scale, *because that's how we're trained to perceive "reality") are within the 98%. Should this bizarre resulting state of affairs be that surprising after running such a simulation for decades?
This question of the use of the word "prejudice" versus "discrimination" in the context of redefinition is worth exploring further.
In general, when a concept relating to some human state of being is defined, the essential features of the concept and the resultant connotations often fall into the following categories:
A set of characteristics, such as the symptoms that characterize a disease.
The causes that brought this state of being into existence
The causes that would result in the resolution of change of this state of being
The outcomes that result from this state of being
Any evaluations, judgemental or moral, or emotional states in response as a valence or sentiment, such as whether this state of being is good or bad.
I used the word "prejudice" deliberately and intend to consider prejudice to be a bare-bones case of pre-judging some aspects of a situation beyond the aspects that led to the classification. But I also want to attach the judgment that prejudice is considered to be a "bad thing" - something to avoid or ameliorate.
Discrimination does not fit either the characteristics or the judgment as well as prejudice does.
Discrimination is the process of making a distinction. But prejudice is more than that. Once you have made a distinction, prejudice kicks in when you draw inferences about the situation that are not directly observed. Instead, they are believed to be associated with the situation.
Also, discrimination does not always have a negative judgment.
A wine connoisseur shows discrimination in (hopefully) a good way: by upholding the best quality wines and denigrating the inferior plonk.
Racial discrimination is usually considered a bad thing, quite often because it involves prejudice.
Someone attuned to sight and sound, such as the ability to discriminate between different shades of color, is neither considered good nor bad - it just is.
Prejudice, though, is considered to be a type of sociopathy.
And here is where the slavish adherence to "common usage" can get you into trouble.
Let's draw a comparison to medical pathologies - diseases.
A stomach ulcer is simply a wound (Greek elkos = wound). But there was a connotation associated with it in relation to its cause. It was assumed to be due to stress. Marshall and Warren back in 1982 proposed that the real cause was H. Pylori infection, but the connotation of stress as a causal agent was so strong that it took a decade before people accepted this.
Sometimes, the name of a disease is the name of the cause. For example, as we can see by the name of the disease, "malaria" = "bad air" is caused by miasma. Of course, we know that to be false now.
This is the problem around the analysis of prejudice. The connotation that you mention is that prejudice is irrational. But it may well be very rational. My thesis is that the "pre-judging" aspect of prejudice is fundamental to human cognition, to the point where some prejudices are genetic. These prejudices are (or were) good for you, which is why they are not termed prejudice in the derogatory use of the term.
For example, we humans have a natural prejudice against predators such as the wolf or the lion. We observe an animal and make the discrimination of it as a wolf. Our pre-judging is that getting too close can be dangerous. Similarly, we have a prejudice against spiders and snakes.
So, I claim that prejudice is neither irrational nor inherently bad. It is a ubiquitous part of human reasoning. What makes the cases that we denigrate as "prejudice" (bad, sociopathic) may come not from the initial discrimination nor from the consequential conclusions of secondary characteristics and connotations, but in our inability to "update our priors" - to either fail to say that our assumptions do not apply in a particular case, or that our generalizations are misplaced.
> for which there is no need
I love comprehensive claims of nonexistence - have you a proof for this fact, or is it more of a faith based belief, or a colloquialism (that shall not be restated in meaningful language, because "that's pedantic")?
> You claim that prejudice is not based on experience. Nor is it based on reasoning. Then what is it based on?
If you pushed him into a corner you'd eventually find he's dealing with a tautological truth - all exceptions to it do not count, *because they would then no longer be a member of the class*. You can see the same behavior in all ideological classifications, like science fans who claim science is perfect, because any wrongdoing by a scientists *is therefore not science*.
21st Century Western Consciousness is a hell of a drug.
> But discrimination is essential, while prejudice is irrational.
This itself is an example of prejudice.
> But discrimination is essential, while prejudice is irrational.
Prejudice precedes (prejudice based.....*and is there any other kind*?) discrimination though.
> Generalization, classification, etc. from observations rely on discrimination, not prejudice....
Have you a proof for this judgment of fact?
> which is defined as "preconceived opinion that is not based on reason or actual experience."
Consider the *minimal* tail of the spectrum of meaning of the phrase "based on".
What you're talking about is generalizations based on past experiences. Prejudice,however , is the set of generalization rules that fail to update when new data arrives. It is the set of degenerate weights that were overfitted on irrelevant data and just wouldn't budge; the pixel in the center of the image that makes it impossible to distinguish between plane and monkey.
Thanks. This is an interesting response. To respond in turn, I am going to get meta on you.
First, let's try to define what is meant by prejudice and give some examples.
By prejudice, I mean that, given a situation where a person observes an entity with defining characteristics A, B and C, they classify this entity as an example of Class X and infer secondary characteristics D, E and F.
For example:
It is prejudicial to assume that, if a person is a conservative, they are opposed to abortion, and if they are a liberal, they are pro-choice, even though this happens quite often.
It is prejudicial to assume that, if a person is black, they are good at basketball, even though the best players in the National Basketball Association are quite often black.
It is prejudicial to assume that, if a person is Asian, they have high SAT scores, even though Asians tend to have higher scores than other groups.
Therefore, to prejudge is to draw an association, based on experience or belief, of a trait not yet observed, based on the defining characteristics of a class of entities. This is also termed "connotation":
https://www.merriam-webster.com/dictionary/connotation
1 a: something suggested by a word or thing: IMPLICATION
1 b: the suggesting of a meaning by a word apart from the thing it explicitly names or describes
My claim is this: prejudice is an essential part of the scientific method. It is a form of prediction:
https://en.wikipedia.org/wiki/Scientific_method
"The scientific method is an iterative, cyclical process through which information is continually revised.[l][47][48] It is generally recognized to develop advances in knowledge through the following elements, in varying combinations or contributions:[49][50]
Characterizations (observations, definitions, and measurements of the subject of inquiry)
Hypotheses (theoretical, hypothetical explanations of observations and measurements of the subject)
Predictions (inductive and deductive reasoning from the hypothesis or theory)
Experiments (tests of all of the above)"
Therefore, my claim is that prejudice usually connotes an error or an overgeneralization, but in actuality, a prejudice may be well-founded and appropriate.
Seen this way, Halley's prediction of the return of his Comet using Newton's theory is a form of prejudice as much as the prediction that a black person is more likely to listen to soul or rap rather than Rossinni's operas.
So you say:
"What you're talking about is generalizations based on past experiences. Prejudice, however, is the set of generalization rules that fail to update when new data arrives. It is the set of degenerate weights that were overfitted on irrelevant data and just wouldn't budge; the pixel in the center of the image that makes it impossible to distinguish between plane and monkey."
So we are in agreement that I am claiming that prejudice is generalizations based on past experience. Where we differ is whether or not prejudice involves failing to update when new data arrives.
Where this gets meta is that I would claim that this second statement is itself prejudicial. Not all prejudices fail to be updated. This happens sometimes, especially if there is an emotional valence attached to the prejudices that make it hard to give up those beliefs. I point this out because there typically is a strong emotional reaction when the word "prejudice" is mentioned. This obscures the fact that prejudice can often be overcome when someone is presented with disconfirming evidence.
There are actually two different ways to address our differences. Although I am claiming that prejudices occur before the act of failure or success in updating the generalization, another approach is to claim that my definition of prejudice is overly broad and that "failure to update" is a defining characteristic.
But that is an unpalatable approach. Going about redefining terms so as to justify our prejudices is unfortunately all too common. That would be like arguing that if someone from China got low SAT scores, then they are "not really Asian".
But a major point that your objection obscures is the common case where a prejudice is actually well-founded, but does not apply in a particular situation. What I was trying to point out is that not everyone fits the same mold. This is not related to updating the model (your reference to "weights" applies to the current fad of Neural Nets - having worked in Computational Learning Theory, I am more interested in learning in general). Instead, I was trying to get at a more subtle point. In applying a general classification rule, it is important to keep in mind that, even though an entity fits a class, this does not mean that all secondary characteristics are true in every case. What this means is that it could be true that your prejudices are well-founded and essentially correct, but in each particular case there are unique differences.
We obviously agree on the fundamental level. And I also agree that "failed to update" is not exhausting the possible scenarios linking prediction to prejudice. I just grabbed the opportunity and used the analogy to show why I think that it is not equivalent to prediction (of course bad predictions are a subset of predictions, so we may also start discussing what is "equivalent" 😇).
Even if it isn't so on the pure semantic level (I am not capable of judging this), in its global use the word "prejudice" carries the meaning of "judging before the (relevant) facts are known". More abstractly, this translates to " assigning value based on outdated, incomplete or irrelevant data". Also , in every day use at least, prejudices are generally thought to be rigid and not easy to change and shift (which is evident e.g. by the often use of the "no true Scotsman"-fallacy to harden one's prejudice once they were shown wrong, as you also pointed it out). Given those two properties of prejudice - "bad" data and rigidness, I think that, while not exhausting, my analogy was pretty precise. At least considering the general use of the word. 😎
This gives us the grounds for an interesting discussion.
One reason that I am trying to reformulate the concept of prejudice so as to be closer to the Machine Learning concept of generalization (besides my being a computer geek) is that this type of reframing can be helpful sometimes by allowing us to see things in a different way. Of course, we should always stick to reality. I am not trying to present "alternate facts". What I hope I can do is to look at prejudice in a way that help ameliorate the rigidity and other negative associations.
I used the term "connotation" advisedly. What I want to get at is to question the "everyday" use: the notion of regidity is considered an essential part of prejudice. If we use this everyday way of looking at things, we may be making a judgment call about the nature of prejudice. We are implying that we have a lack of agency in our prejudice. Thinking of prejudice in our current everyday way, as involving "bad" data and rigidness, means that we may be drawing implications that are not generally true. Which was why I said I was going to get meta on you: using the general use of the term makes you prejudiced angainst prejudice. I am trying to get us away from that.
My claim is that using "bad" or outdated data and sticking rigidly to a viewpoint is not an essential feature of prejudice. Being prejudicial about prejudice in this way is Calvinistic: the prejudiced amongst us are doomed to their incapacity.
But this is not necessarily shown to be true in reality. For instance, people who are prejudiced against a group of people tend to be less prejudiced the more thery actually interact with that group. This implies that they may not be so rigid. People who rely on outdated data may actually be conservative. This reliance is actually sketicism about new-fangled ways: they will give up their reliance if you can prove the new way is better.
So I am arguing that the general use of the term needs to be discarded. To get meta again, thinking of prejudice this way is outdated, incomplete and irrelevant. Outdated, because studies have shown the ability of prejudical people to change if given the chance. Incomplete in the sense that not all prejudice is rigid.
And I claim that these extra connotations are irrelevant. Worse, they make us focus on the wrong problem. A point I want to make is that prejudice is a good thing: it is an essential heuristic for living. Considering prejudice as just another way of drawing generalizations shows that it is the natural and default approach to experinecing new things. That allows us to focus on the rigidiy and the overly conservative clinging to bad or outmoded data as the essential problem, but not an essential part of prejudice itself. My reframing has split off the negative connotation from what is good in prejudice allowing us to keep the good and avoid the bad. It also lessens the tendency to separate people into "good" and "bad" people in terms of their degree of prejudice. Like the song goes in Avenue Q: "everybody is a little bit racist". I explain that by noting that we go through life effieicently, drawing quick generalizations as a matter of course, and only going back and rethinking our prejudices if we have to. Reframing things this way makes it less likely that we overthink things *(like what happened recently with the ACLU) and it also acknowledges that perfection is unattainable and is, in fact, not desireable. Prejudice is a heuristic and heuristics by their nature are subject to error. But they are computationally efficient and "good enough for government work". You put in the extra effort when the need arises.
Admittedly my reframing can come up against our prejudices about prejudice, which we have invested a whole lot of emotional baggage. So, yes, your analogy was a proper characterization of prejudice the way the currently think of it. I am advocating that we have an attitude adjustment when it comes to prejudice: consider prejudice as a given and not inherently bad. Instead focus on the real problem: clinging rigidly to outdated, incomplete or irrelevant data. This points to an essential feature of that problem: we all have a tendency to want to have to aviod having to admit we were wrong.
> Of course, we should always stick to reality.
What do you mean by "reality"?
Have you heard of the term semiotics? Because it seems like that is the perspective you're arguing for here.
What do I mean by reality? Wow, that's a tough one.
"Reality, what a concept" - Mork from Ork
A glib answer is that reality is the pain you feel when you decide that the Theory of Gravity is just a theory the way that the Theory of Evolution is, and decide to defy it by jumping out the second floor window.
To get beyond a glib answer takes you down the rabbit-hole. Philosophically I am a Taoist. Taoism teaches that human understanding is a phenomenological substitute for the underlying noumena - the Tao.
Some thoughts I have about the nature of reality are written up in an essay "Hylomorphic Functions"
Hylomorphic Functions. Researchers.One
https://researchers.one/articles/18.11.00009v1
The first part of this was published under the title
"Causally Active Metaphysical Realism"
Quantum Speculations (supplement to the International Journal of Quantum Foundations), Volume 1, Number 1, October 2019, Pages 1-31.
https://ijqf.org/archives/5704
I wrote a Medium essay for non-technical readers:
Hylomorphic Functions: Is consciousness quantized?
https://medium.com/science-and-philosophy/is-consciousness-quantized-15e249aac558
This work attempts to relate metaphysics, like that of Aristotle, to physics, especially Quantum Mechanics.
For the purposes of this discussion, reality is a kind of "ground truth" where you can test your concepts and observe how well they work, both descriptively and predictively. What I mean by that is that your ideas must be testable. One of the most influential essays on my way of thinking about science is Richard Feynman's Cargo Cult Science
https://calteches.library.caltech.edu/51/2/CargoCult.htm
Feynman wasn't the type to philosophize but I think he had a better grip on what reality really is in reality than most people.
Unfortunately, I have a problem with semiotics. Pierce is considered the founder of that field and I have the highest regard for Pierce. There are other people who I respect are enamoured of semiotics and write about it. But whenever I try to understand semiotics, I could never figure out what it is about. And, I confess to admit, those times I have read articles about semiotics, the results I can understand turn out to be commonplace. Maybe I'm stupid - I don't know. But semiotics just seems to elude me. I guess you could say I can't see what it signifies.
> Given those two properties of prejudice - "bad" data and rigidness, I think that, while not exhausting, my analogy was pretty precise.
Careful though: precision and (exhaustively comprehensive) accuracy are two very different things.
"In college, one of my professors warned me not to go into studying prejudice because, in her opinion, there wasn’t much left to learn about it."
What psychology professors have yet to learn about prejudice is that their prejudices about it are due to ignorant woke bigotry.
https://jclester.substack.com/p/prejudice-a-libertarian-viewpoint
https://jclester.substack.com/p/woke-a-libertarian-viewpoint
https://jclester.substack.com/p/bigotry-and-libertarianism
> Someone could have proven 1000 years earlier that heavier things don’t fall faster than lighter things, for example
But they do!
I wish I could give this 1000 likes!
"The strictest limits are inside our own minds—the questions we never ask because they seem too obvious, the explanations we accept without sufficient evidence, the studies we never run because they’re too far outside the consensus."
From my own interests, I think that gender may be important to the evolution of language. A few linguists and anthropologists have argued the same, but a big hurdle has been that sex differences in the brain are taboo. For my money, that taboo set us back decades in understanding who we are and where we came from.
The problem of the book is who is interviewed: only people who the author has heard about as journalist (a superficial profession), which means people who say crazy, attention grabbing things. As to Wittgenstein, it’s because his work is word salade that everyone (some people) like him so much.
> As to Wittgenstein, it’s because his work is word salade that everyone (some people) like him so much.
It isn't possible for you to know why people like Wittgenstein, it seems like you can because of the culture you've been raised in.
I fail to understand your comment. But, if anything, I was raised in a culture reasonably close to that of Wittgenstein. But I believe only in science as a method of knowledge, but not in culture.
What methodology so you use to acquire accurate knowledge of the motivations of people you've never met?
You are perfectly entitled to like Wittgenstein. And you are perfectly entitled to think that it is not word salad. But I think it is word salad, which is why I thought that I could infer that some people like it because word salad can be interpreted in every which way, and if you interpret it in a way you find interesting you will like Wittgenstein. Note, however, that I am also entitled to have an opinion about Wittgenstein and no appeal to the culture of Wittgenstein or the culture of Wittgenstein lovers can change that.
You call Wittgenstein "word salad", but then you state your opinions as facts - is this not rather paradoxical?
Do you believe that it may be possible that at runtime, you may sometimes forget that your opinions are facts?
Do you ever wonder what is true?
> Note, however, that I am also entitled to have an opinion about Wittgenstein and no appeal to the culture of Wittgenstein or the culture of Wittgenstein lovers can change that.
Are you claiming that *it is not possible* for your opinion of Wittgenstein to be changed by external forces?
What I think needs to be recognized is that often, due to frustrations with the limits of science (as well as its previous achievements), believers in scientific orthodoxy go rogue in order to gain influence (or funding), notoriety, or because they just can't help themselves.
https://mikeklymkowsky.substack.com/p/orthodox-science-as-a-mostly-good?utm_source=profile&utm_medium=reader2
> believers in scientific orthodoxy go rogue
If you think of science from the perspective of yet another human ideology, it shouldn't be too surprising that the faithful are going to misbehave, hallucinate, etc...it goes with the territory of humans!