Atheism and the Strange Library

Atheist YouTuber and “Street Epistemologist” Douglas Letkeman invited me to do a live conversation online this past Monday night. We talked for an hour and 45 minutes! Over 100 people joined us in the live chat, including half a dozen of my logic students. The conversation centered on what Doug calls “the Library thought experiment”– an exercise aimed at sniffing out inconsistencies in a person’s epistemology, that is, in the way they respond to evidence.

The Strange Library

The thought experiment invites you to imagine a strange library, where you work shelving new books. A new book arrives, and you are tasked with shelving it, but you may only look at a single page. The title, author, etc. are all hidden. On the single page you read, there is an account of a man who runs off a cliff and flies through the air like Superman. So where do you place the book? In “history” or in “fiction?” You will probably assign it to the fiction section, presumably because you don’t believe this event actually happened.

book, library, evidence

Now suppose you read another random page, and the book claims that more than 500 witnesses saw the flying man fly. Still fiction? Suppose a third page reveals a “creed” that was recited by believers in the flying man, written soon after his amazing flight. Are you now inclined to think this is a true story? A fourth page asserts that many of the witnesses died because of their belief in the flying man. Is this enough to convince you that the flight of the flying man was an historical event?

You see the parallel. The account and support for the flying man is supposed to be analogous to the account and support for the resurrection of Jesus. Doug’s view is that since they are analogous, you should treat them the same: if you believe in the resurrection, you should also believe in the flying man. But Doug also assumes that most of us would find the flying man story to be rather unbelievable. And if this is so, then you should also be skeptical of the resurrection of Jesus.

The Video

In the video, we talk about the thought experiment and these parallels in great depth. If you don’t have time to watch it, here’s the executive summary. As Doug added on each bit of evidence for the flying man, I said that my confidence in the truth of the account shifted toward belief. However, after 5 or 6 items of evidence for the flying man, I didn’t yet feel that the case was strong enough to warrant belief. Similarly, if I had exactly the same 5 or 6 fragments of evidence for the resurrection, I would probably not believe it happened. I would be right on the threshold between “undecided” and “belief.” I would need additional evidence to propel me through the door.

Thankfully, the case for the resurrection of Jesus contains that additional evidence. One of the additional items is the coherence of the narrative surrounding the resurrection–there is a clear purpose for the event and it fits into a coherent, broader story. Another is the presence of genre clues embedded in the text. The authors of the gospels wrote in the style of ancient biography, not poetry or myth. We could go on: the empty tomb, the first witnesses being women, etc.

The Cumulative Case

Doug also wanted this to be an argument against the “cumulative case” defense of Christianity. The cumulative case approach goes something like this: no single argument or bit of evidence for Christianity is sufficient to warrant belief, but all of them together are sufficient. I think this makes good sense. But Doug wanted to use the Library story to somehow undermine it. I never quite understood how that was supposed to work. The Library case just shows that you shouldn’t believe an unlikely story based on one bit of evidence. But what if you have many bits of evidence?

library, evidence

To show how the cumulative case strategy works, consider a murder investigation. A detective may begin with several suspects and a few items of evidence. One person saw Mr. Smith near the crime scene on the night of the murder. Should the detective believe that Smith is guilty based on this one bit of evidence? Of course not. But over the course of the investigation, the detective uncovers more and more evidence that Smith is the murderer. It becomes increasingly unlikely that all these bits of evidence are coincidence–Smith looks more and more guilty. Eventually, there is sufficient evidence to arrest Smith and prove his guilt. No single item is sufficient, but the total package is. The detective is using a cumulative case approach.

An Objection

One objection I’ve heard is this. If each individual item is weak or bad evidence, then no matter how many you add up, it won’t be sufficient for belief. Well, if the items are just weak on their own, then this is exactly what the cumulative case says. Any single column of a cathedral is too weak to hold up the roof. But this doesn’t mean that a collection of columns couldn’t do the job. This is, in fact, how architecture works.

library, evidence

But what if each item is bad evidence? A fairly standard definition of evidence is “anything that raises the probability of the hypothesis.” So, if by ‘bad’ we mean that the item does not raise the probability that the resurrection account is true, then I agree. But each of the items I discussed with Doug would, if true, raise the probability that the resurrection occurred. One may argue that one or more of the evidential claims is itself false, but that is a completely separate discussion. Needless to say, one can’t simply dismiss them as false without argument. Near universal agreement among liberal and conservative scholars alike supports many of them, such as the fact of the empty tomb, or that the disciples claimed to have seen him alive.


So, I don’t see how the Library thought experiment undermines the cumulative case strategy. In fact, it seems to illustrate exactly how it works. Start with one item of evidence, and continue to add more until you have enough for justification. It isn’t easy to say how many items that would take (a discussion for another time), but surely it is finite. My faith, and the faith of many people I know, rests securely on a collection of sturdy “columns.”

3 thoughts on “Atheism and the Strange Library

  1. Haven’t watched the video, here’s some hot takes from the executive summary:

    1) Was half-afraid, given the title, that this was going to be some take on the “Library of Babel.” Somewhat relieved that it isn’t.

    2) Why is Douglas Letkeman taking an hour and a half out of his day to talk about this? Is his atheism a religion that he feels a need to convert others to? (Hehe, a little atheist-on-atheist snark. Heretics are always more subjectively objectionable than infidels.)

    3) This piece gave me a serious case of flashback, to all the kids that read superhero comics, and then whaled away on a power transformer with a fire ax, in the hopes they would get super-powers. If your library was full of nothing but superhero comics, how would you go about determining where on the fiction/non-fiction spectrum various volumes would fall? (And I’m not sure that there’s a clear distinction between fiction and non-fiction – how about books on cryptozoology?)

    4) Instead of trying to draw conclusions about the outside world from the contents of the library, it might be easier to draw conclusions about the nature of the library itself. Who is curating the library’s collection?

    5) In the Western tradition, Herodotus is known as both “the Father of History” and “the Father of Lies.” This is not a coincidence, there is no such thing as a “true” history. Granted, some histories are more veridical than others. A lot of the value of history is that you get to hear the “voices” of people from the past, in their own words. But if you want to know the truth of what actually happened… First, autobiographies are the biggest lies, yet, in some ways, also the most truthful. Second, it helps to see what a figure’s enemies have to say about them. Not really available in Jesus’s case: he simply doesn’t show up in non-Christian sources until the 2nd Century. Thirdly, if possible, it’s always helpful to see what other fields, like archaeology, have to say. Archaeology did confirm the historical existence of Troy (although we’re still trying to figure out *which* Troy the Illiad cycle was about… there’s quite a few layers of Troy, built on top of each other). But archaeology has also uncovered evidence of plagues and famines that were not recorded in our histories, since they had only minimal effect on court life, where the chronicling occurred. So given ‘histories’ of superheros, might at least see if anybody left behind commemorative statues in town squares, named streets after them, etc.

    5) I’m not exactly up to speed on the different genres of classical writing. But wasn’t “biography” a Greco-Roman thing? Which makes sense, given that the earliest gospels we have are from the Greek. There are rumors of an earlier “Gospel of the Hebrews,” of course… I recently developed a hypothesis that in the immediate wake of Jesus’s crucifixion, his inner circle (especially Peter) spun a resurrection story based on a “super-Elijah” take on Jesus’s status as a prophet and a messiah. How would the Hebrew literature of Elijah be translated into the genres of Greco-Roman literature? Coming the other way, I know that a lot of supernatural stuff came into Herodotus’ stuff. He generally cited sources who were telling him the story second or third hand. What’s the most extreme supernatural stuff that went into Greco-Roman biographies? (And would Herodotus’s histories even be considered “biographies”?)

    That’s just my hot takes, haven’t come to an firm conclusions about this line of argument yet.

    1. Wow “Eddie,” your barrage of comments is a little too much for me. I think the thought experiment makes a simple epistemic analogy between the resurrection and any other claim of a highly improbable event with limited evidence. There are problems with the experiment set-up, but nothing insurmountable. As far as your theory about the resurrection . . . I think you’re better off leaving the heavy lifting to the experts.

      1. Wow, in retrospect, my takes were even hotter than I thought. My apologies. Hm. So, if I understand the argument correctly, it’s something like this: we have alleged observations of an event with a low initial likelihood, something that is highly improbable given what we know about how the universe acts. Something like a man resurrecting, or a man flying (without the aid of an airplane). We might also add in something like the EM drive, aka the “Impossible Drive.” All three events are virtually impossible given what we understand of how physics works, and are only possible if our understanding of physics is wrong. (Our understanding of how physics works being wrong wouldn’t be that surprising. Our understanding of it being wrong in the kind of way that would permit these three events would be… much more surprising.)

        In all three cases, we might have writings in that library that try to incorporate these alleged events into a narrative that attempts to explain them. “God did it,” “The man was actually a human-appearing alien who was powered by Earth’s yellow sun,” and, I’m not sure about the last one, “photons have mass,” maybe. That would go along with your point about “the coherence of the narrative surrounding the resurrection–there is a clear purpose for the event and it fits into a coherent, broader story.”

        There are other pieces of evidence available that tend to confirm that the events actually happened. With the EM drive, there are some tests that suggest the drive does actually exert some thrust (at least on Earth, in a vacuum chamber, etc.). With the flying man story, we might have the recorded statements of a certain intrepid and spunky reporter, etc. With the resurrection, we have the testimony of several witnesses. (Recorded several decades after the fact, but possibly within the lifetimes of those witnesses.)

        Hmm. I would agree that, even if there is no single “smoking gun” piece of evidence that would establish “beyond a reasonable doubt” that one of those three events happened, you might have a constellation of evidence that could collectively do so.

        If that *is* the claim, then my main concern would be with this one feature of Bayes’ Theorem. The Theorem is often used to model how scientific theory confirmation works, and in its naive forms, has the problem that it assumes that different pieces of evidence are statistically independent. In practice, different pieces of evidence usually *are* statistically related, and ignoring this can cause naive Bayes’ Theorem to assign a probability to a hypothesis that is greater than 1. (Dealing problem with this in a formal way would require determining a correction formula for each potentially related constellation of pieces of evidence. The number of required correction formulae increases factorially with the number of pieces of evidence…. which is not the kind of thing that makes a mathematician or a computer happy.)

        One way that multiple pieces of evidence can be statistically related is if there is an alternative hypothesis that explains them or explains them away. Bayes theorem would assume that this alternative hypothesis has a certain likelihood of being true. In order to establish one of the three hypotheses as being true ‘beyond a reasonable doubt,’ you would presumably have to show that the probability of the combination of the various alternative hypothesis is less than the probability of our understanding of how physics works being wrong in a way that would permit the alleged observed events to happen.

        Or, to use the murder mystery analogy, a possible alternative hypothesis is that the murder was actually done by somebody else, who left clues to frame an innocent party.

        In the case of the EM drive, Wikipedia lists 5 different hypotheses that attempt to explain the observed thrust that the “Impossible Drive” produced in certain experiments. Only 1 of those 5 hypotheses involve violation of known physical laws. I won’t, on that basis alone, claim that the “photons have mass” hypothesis is conclusively disconfirmed, but… Well, if you’re interested, read the Wikipedia article for yourself.

        In the case of the “flying man,” having the background knowledge to realize that the ‘library’ you are in is actually a comic book shop, and that the intrepid reporter is just as fictional as the flying man, would, I believe, significantly reduce the subjective probability that there was a humanoid alien from Krypton who was raised as a human in Kansas.

        I would agree that the evidence you provide does strongly disconfirm the alternative hypothesis that the Gospel is mere fiction, intended to be read as such. I would agree that there is sufficient evidence to establish beyond reasonable doubt the historical existence of Jesus, and his execution by the Roman state. I would agree that the narrative of the Resurrection does not show signs of developing like a “naturally evolved” myth, although I am less certain of that conclusion.

        My own alternative (working) hypothesis is that the Resurrection narrative was a piece of propaganda deliberately constructed by the inner circle of Jesus’s movement, in the immediate aftermath of Jesus’s crucifixion, with Peter as the prime suspect for being the “ring-leader” of the conspiracy. The motive of this conspiracy was to prevent the destruction of the movement after the execution of its Messiah, and to keep the movement focused on the goals laid out by its founder. A lie, perhaps, but I would classify it as a well-intentioned one, and perhaps even a justified one.

        I will admit that this is a *literal* conspiracy theory. The usual problem with conspiracy theories is that they postulate conspiracies too large to effectively keep the secret. But I think this alleged conspiracy is small enough (I estimate that it would probably have on the order of about 20 people in on the gag. A quick re-read of the synoptic Gospels, or even just Mark, would probably be enough to get an estimate) that I think they would have been able to keep mum *enough* to keep Jesus’s sect together and on mission. Plus, if there were members of this small group who later defected and spilled the beans, it’s likely that history would not have recorded it. “Don’t worry, the Messiah’s coming back!” is a heckuva better of a story than “Then the Messiah was executed, the end.”

        I do not necessarily endorse this hypothesis, but I believe that it would explain – or explain away – much of the available evidence for the Resurrection. I would challenge you to think of ways to disconfirm it.

        One thing that would tend to confirm the Resurrection hypothesis in the face of this alternative hypothesis is if we had independent evidence for the existence of God. That would provide both “means” and “opportunity.” Leaving “motive,” which I think you have reasonably covered… But I would warn you that it is not possible for science to establish the existence of an all-powerful God. At most, science might be able to establish the existence of god-like entity, and then establish progressively higher lower bounds on that entity’s abilities. (Okay, we’ve established that the Big Guy can do resurrections – but can he make planets?) “Fortunately,” an all-powerful God is not necessary for the Resurrection hypothesis to work – a mere Odin or Zeus would be able to pull a Resurrection off, I believe… Or even a an entity with the personality of YHWH but only an Odin or Zeus level of ability.

        I hope that this reply doesn’t annoy you like the last one did, and I hope it was at least vaguely on topic!

Leave a Reply

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