After arguing with someone about something (about which they were clearly wrong) I started wondering how I knew that I was right about stuff. How do I know the things I know are true? And why do I believe these things are true, rather than some other things? And why are some people so dead certain of things that are obviously false?
My initial feeling that I had some special, exceptionally clever system for deciding what was true eventually gave way to the realization that we all do it pretty much the same way. This page is a summary of my conclusions.
In general, there is no such thing as "absolute" knowledge. The best we can achieve is what we might call "reasonable certainty". Our concern on this page is to figure out which things we should be "reasonably certain" are true.Consider this list of questions. For the ones where you know the answer, ask yourself how you know; that's the point of this exercise.
I don't expect anyone who reads this to be able to answer more than a few of these from their own experience. (Were you in New York on 9/11? If not, you're taking someone else's word for it as regards the airplanes.) Most of the time, we're dependent on information we got from someone else. But people can be wrong, they can lie, experiments can be botched, and even our own memories can be wrong, as studies of eye witnesses have shown. So, how do we know what's true? If you've never been there, how do you know Philadelphia exists? (And if you have been there, how did you know you really were in Philadelphia? Did you walk there, checking landmarks to be sure of where you were, or check your location with your own sextant and clock after you arrived? Or did you just take someone else's word for it?)
When we don't know the answer to a question, we look for someone who does. In arguments on the Internet, that's commonly reviled as appeal to authority. In fact, unless you're an incredible polymath, you probably answer nearly all questions of fact which you encounter by "appeal to authority". What year is it? Check the calendar. What time is it? Check your phone, which (you've been told) receives the time from a central source, which (you've been told) is accurate.
This is inevitable, and has been true for thousands of years (except the cell phone part). Finding genuinely new facts on your own, and confirming that things we think we know really are true using your own resources, is very difficult, and there isn't time to do more than a small amount of it. So, nearly all the time, we depend on appeal to authority.That brings us to the central question, which determines what we think is true, and leads directly to the problem of conspiracy theories and Internet-based stupidity: Who is an authority? How do we know who to believe?
A "conspiracy theory" is an assertion which is generally considered false by "mainstream" information sources, but which is asserted by a minority to describe the true state of affairs. In order for it to be true, the "mainstream" sources must be conspiring to tell us otherwise, either intentionally or through shared stupidity.
We'll have more to say about them later, but for now we need one fact about conspiracy theories: They are irrefutable. By their nature, it is impossible to prove them wrong. (By Popper's criteria, therefore, they're not valid theories, as they are not falsifiable.) The theory just bends to accommodate new evidence. Do you think we went to the Moon because you saw it on live TV? The TV feed was faked. Do you know Buzz Aldrin personally, and he told you he went to the Moon? He lied to you. Are you Buzz Aldrin? If you really believe you were on the Moon, then you've had false memories implanted in you by the CIA, through a course of drugs and hypnosis, so you think you went to the Moon but you really didn't.
Before the Internet, it was rare to find someone who accepted
conspiracy theories and rejected mainstream beliefs. In the
Internet era, it has become depressingly common. On the remainder
of this page, we'll be considering why that's true.
We'll also be looking at the question of how we can be so sure most
conspiracy theories are incorrect, given that they can't be proved
Long ago, when there were fewer people and they lived in isolated communities (or tribes, or whatever), it was straightforward to assemble a mental model of the sources of knowledge. People who were reliable would become known for that. On the other hand, people who were unreliable, or made up stories, and often were proved wrong, would soon be known as well. When new information arrived, one could test it against what one already knew, and evaluate the source in the light of what one knew about everyone in the group, and get a pretty good idea of whether it was likely to be true or not.
In the pre-Internet but post-industrial era, our information sources were still limited. If you wanted to learn about something technical (such as the theory of relativity, or how a Saturn V rocket worked) you could look in an encyclopedia (if you owned one), you could visit the library, or you could read about it in a magazine such as Science News or read about it in a newspaper. If you were in school, you could ask a teacher. These were all what we might call very mainstream sources -- it was unlikely you'd ever encounter an "alternative" point of view this way. Television news was generally restricted to the three major networks and PBS, and they were nearly as "mainstream" as the encyclopedia or your high school physics teacher. (We'll have more to say about Fox later.)
In that world, if you happened to run across an "alternative" point of view, it would fail to fit in with what you knew, it would fail to fit with information you could find in sources you "knew" were accurate, and if you asked anyone whose opinion you respected about it, chances are they'd tell you it was wrong. Consequently, you would tend to reject it as being unlikely to be true. I recall running across a book in the library which claimed to refute Special Relativity, back long before the Internet existed. I didn't know enough to judge it at the time, but one thing stood out for me: It didn't fit with anything else I read at that time, and it absolutely didn't fit with anything I subsequently learned about relativity. So, I eventually wrote it off as a rather strange bit of crackpottery.
Judging new information based on whether it fits with what you already know is confirmation bias. In a primitive world it's an extremely useful tool for weeding out nonsense.
Perhaps the best known statement of the principle of confirmation bias as a useful tool is "Extraordinary claims require extraordinary proof". That is confirmation bias in a nutshell. Something "extraordinary" is something "out of the ordinary". In other words, if a claim is not what we expect, then it is probably false.
In the presence of the Internet, there are problems with confirmation bias.
On the Internet, the problem of sorting out truth from nonsense is much harder. There are a multitude of sources, and we have no direct knowledge of any of them. Furthermore, we can find ourselves discussing things with individuals we'd never ever have encountered in the days before the Internet. We meet them with no reputation, no prior knowledge, and no way of judging them beyond their own words.
In this milieu, if we start out suspecting something to be true, it's usually not hard to find sources on the Internet which confirm our suspicions, whether or not the thing is true. If we have no objective criteria for judging the truth of what we see, then by selecting the sources which "fit" with what we already (think we) know, we can reinforce our initial view of things, which in turn makes it more likely that we'll select sources which agree with that view in the future. And in short order we find ourselves posting messages on the Koreshan website, discussing the fact that astronomers are all hiding the real situation from us, which is that we live on the inner surface of an inside-out planet.
Thus, confirmation bias, which is extremely useful in a simple world, turns out to be a recipe for disaster in a "fully connected" world. Unless you're lucky enough to start out with a "sensible" collection of beliefs into which to fit newly acquired knowledge, access to the Internet is as likely to lead you astray as to teach you anything which is correct.
As we observed earlier, confirmation bias is what nearly all of us use to filter truth from garbage (whether we realize it or not). Only things that get past our "bias filters" are considered for serious investigation. We have to do it this way; there's too much information, and too much nonsense, to treat all of it seriously. The Koreshans are going about acquiring knowledge the same way the rest of us are. It's the ease with which one can fall into the trap of using the Internet's richness to reinforce incorrect ideas that leads them on, not some innate personal failing.
So, how do we know we're right? How do we know that the Koreshans aren't the ones who have it right, and all the rest of us are confused? For that matter, how do we know Philadelphia really exists?
There is an answer to this. It should be taught in school alongside the scientific method (but it's not, last I heard). It's the defense against Internet craziness, and it's what we'll consider next.
I've often seen Occam's Razor presented as a "rule of thumb", and, as I understand it, its original description involved choosing less complex solutions on the grounds of esthetics.
It's generally stated as something along the lines of "One should not unnecessarily multiply assumptions".
In fact, while this may be how Occam stated it, there is a sound mathematical principle at work here, and Occam's razor can be restated as a law rather than a rule of thumb.
The probability of several independent events all occurring is the product of their individual probabilities.
Thus, if a statement depends on some collection of assumptions, the probability that the statement is correct is the product of the probabilities that each of the assumptions is correct. While we can't generally know the exact probability that any particular assumption is correct, we can, at least, be quite sure that the probability of each assumption will be between 0 and 1, and we may be able to guess whether it's closer to 1 or 0.
Consequently, if we have two statements, and we can figure out that one of them requires many more (independent) assumptions to be true than the other, then we can conclude that the one with fewer assumptions is more probably correct. And if the difference in assumptions is extreme, we can conclude that the statement with fewer assumptions is far more likely to be true. This is not just a matter of esthetics; it means something specific. If we're faced by a series of decisions, and in each decision, we are given two conflicting statements and need to guess which is correct, and, when we have nothing else to go on, we always choose the statement which embodies fewer (improbable) assumptions, we'll find that most of the time we made the right choice. And there is no issue of "esthetics" involved.
We'll come back to Occam's razor after we spend some time looking at the scientific method, and how it's actually used to sort out truth from confusion.
In the pre-scientific world, a great deal of what was "known" (beyond that which was tested in daily existence) was believed to be true only because some "accepted authority" had said so. This was totally unchecked "appeal to authority". If Aristotle (for example) said it, it must be true. (So, for example, the function of the lungs was to cool the body -- Aristotle said so, and he was usually right about stuff.)
In the absence of any tool beyond confirmation bias, this is a rational way to proceed: Look for someone we know has been a reliable source in the past. They are more likely to be correct than someone who has no track record.
On the Internet, where determining whether someone is "generally reliable" or not is far more difficult, the "Joe said so" principle is even shakier than it was in the pre-scientific world.
The discovery of the scientific method was revolutionary. It replaced the unreliable "Joe said so" with a way of determining, to high probability, what is actually true.
When we're introduced to the scientific method in school, we're typically told that it involves observing, then making a theory, making predictions from the theory, and then testing the predictions. If the predictions are not born out, then we reject the theory and start over. We're also told that testing the predictions and rejecting the theory if it's proved false is what distinguishes scientific societies from pre-scientific societies.
This is true enough, and it already puts conspiracy theories in a shaky position, since by their nature they typically cannot be used to make testable predictions. However, there is an additional, extremely important step, which is replication. When one person has tested something and confirmed it, we have reason to believe it might be true. But until other people have also tested it and confirmed the original result, it is, at best, a tentative conclusion.
It is replication which distinguishes a confirmed probably-true fact from something which we know to be true because "Joe said so". The reason lies with Occam's razor. To see this, assume we have a result -- say, "antarctic fish blood freezes below the freezing point of pure water". Now, consider the assertion that the result is false. What assumptions must hold in order for the result to be false?
If one researcher has done one experiment which supported the result, then for the result to be false, just one experiment must have been botched. Someone misread a thermometer, or a frustrated grad student made up some numbers instead of making real measurements, or any number of other things went wrong.
On the other hand, if the experiment has been replicated by five different labs, then in order for the result to be wrong, every one of those experiments would have to have been botched. If it's been replicated, then we need to make many more assumptions to arrive at the conclusion that the result is wrong.
Replication is needed to rule out errors on the part of the researches, and also to rule out fraud. The honor system never works very well when humans are involved (consider the lock on your front door).
Much is made of peer reviewed journals in discussions of modern science. In fact, while peer review assures that bad articles are less likely to be printed, and feedback from peer review may cause articles to be improved before being published, and while peer review saves journal editors a bundle of work (without it, they'd need to review every article carefully themselves), peer review cannot prevent fraud. A well done article with cooked data will not get caught by peer review, unless the data is so badly cooked that it's obviously impossible.
The real guards against fraud are publicity (when it's caught) and replication (which can lead to the realization that a particular result must be wrong). Papers reporting replications of earlier results can be published in Arxiv just as easily as they can be published in peer reviewed literature, and they'll have just as much validity.
Experiments done in a laboratory are valuable but ultimately they only tell us what happens during experiments. In some fields, such as physics, that's all we need to know. However, in the life sciences in general, and in the field of medicine in particular, there are too many variables in the real world to allow exact modeling of real situations in a laboratory setting. Furthermore, there are a lot of questions involving impact of lifestyles on health which can't be answered with simple laboratory experiments.
For such things, researchers do studies. They examine matched groups of people between which only one or a few variables are changed, or they follow a group of people over a period of time during which various procedures may be done, and they observe the results.
Studies, just like other experiments, are most convincing when they're replicated. However, some studies -- in particular, large longitudinal studies -- can be very convincing even when not replicated. This is because the actual "experiment" consists of analyzing the data gathered in the study and, if the raw data is published (as it it normally is), any other researcher can redo the data analysis, and can look for other causes or effects which might explain the data, which effectively "replicates" the original researcher's conclusion.
It is worth mentioning in passing that "alternative medicine" tends to be severely lacking in studies which support the claims of its practitioners, and in fact advocates of alternative medicine are often very critical both of the use of studies and of the mainstream journals in which results are commonly published. We may have more to say about this later.
Lies are anathema to mainstream science. When a researcher is caught lying about his results, the major journals typically retract all of his papers, and every paper on which he collaborated is carefully examined to see if he had any influence over the raw data. If he did, those papers may be retracted, too, because he's no longer considered reliable.
Justice can be draconian. A researcher who lies about his data, and is caught, has probably just ended his career.
This is not true in the world of conspiracy theories, but it ought to be. As soon as someone who is expounding an "alternative" theory is caught in an outright lie, we know that they cannot be trusted. From that point on, everything they say should be viewed with suspicion. Unless we know it to be true from third party sources, we should assume anything else they say is also a lie.
This is harsh. However, it is completely natural; it is, in fact, the flip side of confirmation bias: When you know someone is unreliable, do not accept what they say at face value. Confirmation bias, our most basic tool, though hazardous in the world of the Internet, remains valuable.
We've made the point that conspiracy theories are not falsifiable. There are other hallmarks of such theories, and there is also the very interesting question of where they come from.
Even more than the lack of falsifiability, the primary hallmark of a conspiracy theory is that it includes the assertion that mainstream sources cannot be trusted. So, you can't refute it just by looking in the Encyclopedia Britannica or checking a copy of Science magazine, because they are part of the conspiracy. The term for this is "poisoning the well" -- your main source of information which you might use to refute the theory has been "poisoned".
In some cases, we're told that all conventional sources are confused, misled, or ignorant; in other cases we're told that the conventional sources are actively suppressing the "new knowledge". But one way or another, the point is made that conventional sources are not to be trusted. This must be the case; otherwise the theory would be shown wrong from the get-go.
As we said earlier, conspiracy theories are irrefutable. In the
crudest version, any attempt at proving a conspiracy theory wrong by
introducing a new fact can be countered by enlarging the alleged
conspiracy to include the source of the new fact. Doing so
increases the number of assumptions which must be true in order for the
theory to be correct, and so makes it even less probable when
viewed with Occam's razor.
This wonderful term -- "Gish Gallop" -- comes to you from TVTropes and puts a name to a familiar phenomenon. In an argument about a conspiracy theory, it's common to encounter a series of objections thrown up against the conventional viewpoint. They are generally simple objections to raise, but each one would take significant effort to prove incorrect. Furthermore, it's common to encounter objections that only someone with a technical background would recognize as being specious. This approach is intended to overwhelm opposition, and to convince anyone who is not a specialist. It can be very effective. You won't usually see it used to support conventional science -- most conclusions of mainstream science are supported by relatively limited numbers of solid tests rather than large numbers of trivial points.
As an example, if you watch one of the Moon landing hoax videos, you may be told that:
The list goes on. These are some of the points I can recall off the top of my head. Every one of them can be refuted easily. However, by "easily" I mean it would take at least a few sentences of text, and ten or twenty minutes of research. So, to hit all the entries in this Gish gallop I'd need to invest several hours, and in that time someone who was determined to find flaws in the Moon missions could come up with yet more points that would take time to refute.
When confronted with something like this, an effective initial approach is to glance down the list and see if there are any obvious lies. (A lie is not just an error -- a lie is a false statement we can be sure the author knew was false.) If there aren't, then maybe it's something to take seriously. If there are, on the other hand, maybe we should throw it all in recycling and go read Knights of the Dinner Table.
The first objection, that the stars should have been visible in the photos, may seem totally startling (and convincing) when first encountered. It can be refuted but it takes some work. In a page elsewhere on this site I show the results of experiments with a camera at night which indicated that the photos should appear very much as they do (the issue is dynamic range) but it's a confusing enough issue that the authors of the 'hoax' video could have just been mistaken about it. So we won't call it a "lie".
However, speaking as an amateur photographer and artist, the second point -- shadows cast by sunlight are parallel -- did indeed jump out at me as being off-the-wall bogus. The examples shown in the video were carefully selected to show parallel shadows, but anyone who has learned perspective drawing knows the claim is false; shadows in a photograph are rarely parallel. If the producers of the video had any grasp of the principles of graphics (and they claimed to be graphics experts) they knew it too. Therefore, that is a lie, not just an error.
The fourth and fifth points also turn out to be trivially false, and the authors should certainly have known they were false, so I'd call them lies as well. UV filters have been available for cameras since at least the 1950's, so the UV would not have fogged the film (and Hasselblad, which made the camera, knows all about stuff like this). So point 4 is wrong, and the authors should have known it. Point 5 is a flat-out lie, as a few minutes of research on the Internet revealed -- the camera was designed to be clipped to the astronaut's suit or to be held in a gloved hand and could be used either way.
We conclude that somebody seeded this list with lies, and so, using our good friend Confirmation Bias, we should assume that anything on it which we can't independently verify is also false because it comes to us from an unreliable source.
Where do conspiracy theories come from, and who spreads them?
Most people who spread them are well meaning but wrong (and have no understanding of Occam's Razor). The Moon landing hoax conspiracy theory probably sucks a lot of people in because of the lack of stars in the photos. It's extremely surprising and the reasons for it are beyond the photographic understanding of most people. Once someone has the idea that it might have been faked, it's easy to find lots of "information" on the Internet to support that notion.
But someone took the theory to the next level. The Gish gallop of Moon landing hoax nonsense I showed above contained a number of outright lies, which would have been recognized as lies by anyone with the expertise to put together such a list to start with. Therefore, someone intentionally created a series of misleading arguments to convince people the Moon landings were false. Since they were willing to lie to support the claim, we're left with the suspicion that, not only are they unscrupulous, they also know perfectly well that the hoax claims they're spreading are false.
It may not be immediately clear why anyone would do that, but I can see signs of it in all the major conspiracy theories I've encountered. In back of the theories, there are malicious people who are willing to just make stuff up to support the theory. Most of the people who repeat the theories are merely misguided and ignorant, but it takes a certain amount of expertise and a lot of ill will to create the arguments which support them.
One common argument in favor of conspiracy theories is that the people spreading them have nothing to gain. They are putting in their time and labor to spread "the truth", and gaining nothing from it, so they must be fundamentally honest and well meaning. There is a flaw in this, however, which is that those who spread the theories often do gain something from them. The people selling Moon landing hoax videos are selling them, not giving them away for free on Youtube. The man who helped some burn victims out of the elevator in the basement of the World Trade Center shortly before it collapsed, and who subsequently told a bizarre tale of explosions and inexplicable fires, later went on a lecture tour charging, IIRC, something well over a thousand dollars for each appearance. Financial incentives to lie should not be ignored.
The people I knew who handed out videos and literature at Truth Out rallies gained nothing from it, and were apparently completely sincere, but they were also not the originators of the ideas about 9/11 which were they were spreading. On the other hand, Thierry Meyssan, who claimed the Pentagon was hit by a truck bomb, not a plane, sold a huge pile of books asserting there was no airplane (his book has been translated into 25 languages). Did he have a financial incentive? Obviously! When the "truck bomb" version of his theory was disproved (by, among other things, recovery of a bunch of airplane pieces) he didn't back down, but changed the claim, to say there was a plane alright but it was a Global Hawk drone, not a Boeing. And so the theory that there was no passenger plane involved proved unkillable (just as all conspiracy theories do).
This is perhaps the worst thing the Internet does: It gives the authors of conspiracy theories -- who are not at all well-meaning -- a far larger audience than they would otherwise have.
We shall now come back to the list of questions we started with, and consider how we know the answers to them, using Occam's razor as needed to choose between alternatives.
We only know that Lincoln existed because we are told he existed. Nobody alive today actually remembers him. So how do we know the sources which tell us he existed are telling the truth? Let's look at the assumptions which are involved.To conclude that Lincoln did exist, we need to assume that the sources which tell us that historical records (documents in libraries, government archives, and so forth) exist that discuss his existence are what they seem, and that anecdotes we may read of people who knew Lincoln, or, more recently, knew old people who had known Lincoln, are generally correct. Furthermore, we have no need to assume these sources are all correct, nor that any of them are completely correct -- all we need to do is assume that some of them are generally correct in that they report on a real historical personage.
To answer this, we once again need to count up our assumptions.
If we are to conclude that Philadelphia does not exist, then we must assume that there is an enormous amount of infrastructure, in books, on television, on airport marquees, and even on the Internet, which has just one purpose: to present the illusion that there's a city by that name in eastern Pennsylvania.
And all of that infrastructure was constructed faultlessly, and with the cooperation of thousands of people, and none of them spilled the beans to us about the fraud. We can say, very roughly, that we are making one assumption per person involved in setting this up. That is a lot of assumptions. All those people must have behaved in a way that appears to us to make no sense, just to keep up the facade; any of them could have broken the illusion, but none did. So we need to multiply together the probabilities that all those thousands of people would have behaved in this one strange way. It seems unlikely that one person would be willing to do that; what's the probability that thousands of them would? Any probability that's significantly smaller than 1, when multiplied by itself several thousand times, will produce a result that's nearly zero. And so we can conclude that the probability that Philadelphia doesn't exist is nearly zero.