Intro/FAQ ⎜ Episode 1 ⎜ Episode 2 ⎜ Episode 3 ⎜ Episode 4 ⎜ Episode 5 ⎜ Episode 6 ⎜ Episode 7 ⎜ Episode 8 ⎜ Episode 9 ⎜ Episode 10 ⎜ Episode 11 ⎜ Episode 12 ⎜ Episode 13 ⎜ Episode 14 ⎜ Episode 15 ⎜ Episode 16 ⎜ Episode 17 ⎜ Episode 18 ⎜ Episode 19 ⎜ Episode 20 ⎜ Episode 21 ⎜ Episode 22 ⎜ Episode 23

[Editor’s Note: This is the seventeenth in a series of 23 essays summarizing and evaluating Book of Mormon-related evidence from a Bayesian statistical perspective. See the FAQ at the end of the introductory episode for details on methodology.]

The TLDR

It seems unlikely that Joseph could keep the Book of Mormon’s complex geographical details straight without committing frequent and obvious errors, particularly while dictating the text in a single draft.

The Book of Mormon details a complex world, referencing over 100 different cities and geographical features tied together in a dense web of over 150 unique geographic relationships. Throughout the book, there are only two potential spots that could be construed as geographic errors. Based on psychological studies of memory, we should expect something closer to 28 such errors. Though we can’t incorporate this evidence into our overall probability estimate for an authentic Book of Mormon (due to a lack of independence with other types of evidence), I estimate the probability of producing a geography that consistent to be p = .000079. Critics are quick to claim that Joseph must have had a savant-level memory, but they’re going to have to work harder to build their case (using evidence outside the dictation of scripture) that Joseph was capable of pulling off this kind of memory-related feat.

Evidence Score = 4 (i.e., if this evidence was incorporated in our overall estimate, it would increase the probability of an authentic Book of Mormon by four orders of magnitude)

The Narrative

When last we left you, our ardent skeptic, you had reluctantly set aside the book that still lay on your rough-hewn wooden table. The hunger that drove you to devour page after page was overtaken by that old-fashioned hunger of the body, and you take the time to quickly prepare a mid-day meal. But as you gnaw on a humble portion of bread and cheese, it’s the book that keeps gnawing on you.

In that book, you’re stuck in what seems like an endless war between peoples that had been blood enemies for hundreds of years, and the narrator seemed to take an unhealthy interest in describing battle after battle. Though you think such military obsessions as more than a little odd for an ostensibly religious text, you write it off as the vain imaginations of the foolhardy youth who surely authored it. Still, you have to give that author some credit. These battles involved city after city after city changing hands multiple times in a dizzyingly complex arrangement that remained remarkably consistent throughout. You wouldn’t be surprised to learn that cities and lands last mentioned dozens of pages before held true to the internal structure that seemed to bind them.

You think back to your own somewhat awkward attempts at authoring fiction as a younger man. Such efforts had given you an immediate and forceful respect for the likes of Cooper and Dickens—keeping a story straight had been quite a bit more difficult than you’d imagined. And if the young man who had given you the book had been telling you the truth, the book’s author would’ve had to keep these plates spinning while dictating the entirety of the text. It seems unlikely that an author dictating such a complex text could keep all these geographical details straight without committing frequent and obvious continuity errors.

The Introduction

Despite the seemingly everlasting debates over Book of Mormon geography, one overriding fact generally receives short shrift—that it’s possible to build a coherent internal model of the book’s geography at all. The book’s proponents often point to the book’s somewhat staggering complexity, and that complexity extends to the dozens of physical locations and hundreds of geographical references in the text. To keep all those details straight would be impressive even in professionally edited books relying on comprehensive maps. It would be even more impressive in unedited first drafts, and more impressive still in a dictated manuscript.

Doing so would seem to have required an impressive feat of memory, similar to the feats we’ve previously discussed regarding Joseph’s dictation of the text. That immediately presents a problem for us, since that means the probability of maintaining a consistent internal geography wouldn’t be independent with, say, being able to resume dictation without prompting, or following through on describing the reigns of the Jaredite kings in perfect reverse order. Though being able to pull off all those feats of memory would probably be even less likely than pulling off just one of them, we’ll be playing it safe by not incorporating this evidence into our overall probability estimate.

What this episode will instead be is an opportunity to dig more deeply into the Book of Mormon’s fascinating geography, and to see if that internal geography is indeed unexpectedly consistent in isolation. To do that, we’ll need to first establish whether those geographic details actually are as consistent as claimed. Then we’ll need to try to figure out just how many errors we’d expect to show up in a work as complex as the Book of Mormon.

The Analysis

The Evidence

What sort of geographical complexity are we actually talking about here? It depends a bit on who you ask. If we’re going by the raw number of physical locations referenced in the book, we could use this exceptionally detailed database of Book of Mormon locations. My count, based on a review of that database, is 102 (see the Appendix for a listing and description)—a number that’s surprisingly comparable to Tolkien’s Middle Earth, which over the course of his collected works totals about 130 (keep in mind, of course, that Tolkien had decades to craft his universe, and that sort of complexity is far from a given in fictional worlds—my own humble set of amateur fantasy novels features a grand total of 10 locations over the course of 180,000 words). We could also go by the number of times those locations are referenced in the book, with estimates of between 500 and 600 such references.

But for me the number that matters is the number of geographical relationships that the book establishes. Merely stating a place or referring to it doesn’t necessarily mean that a writer would be at risk of committing a continuity error. What produces that risk is establishing a connection between one place and another. Once you say that a place is nearly bordering, or north of, or across the river from another, then you’re committed to maintaining that connection throughout the rest of the book, and you place yourself at risk of contradicting that established connection. Now, many of the places in the Book of Mormon are merely mentioned in isolation, without connecting them to any other locations, such as the list of cities destroyed in the calamities of 3 Nephi, or many of the locations in Jaredite lands. But quite a few locations are very tightly knit together with various others in a web of mutual geographical connections, connections that could be easily broken. Based on the geographical references listed in the database I linked above, I count at least 151 unique geographical relationships (note that there may be some that I’ve missed, particularly ones implied by the movement of armies rather than directly referencing a connection between cities). I’ve plotted those relationships in the figure below, with numbers corresponding to the location numbers in the appendix. Note that this only plots the relationships themselves, and doesn’t imply anything about relative position (particularly for the ones without any connections, which are plotted randomly).

So the book’s geography is indeed complex. But is it as consistent as some claim? Thinking through that question, I figure that the best people to ask would be the book’s critics, who generally never miss a chance to poke holes in the book’s narrative. After spending what may have been an unreasonable amount of time combing the internet for alleged inconsistencies, I could only find one, based on this list. Yet of the 17 (supposed) inconsistencies listed, only two of them have anything to do with geography, and they’re really referring to the same issue—the fact that there seems to be two different cities named Mulek—one a Nephite city by the East Sea, and one a stronghold of the Lamanites in the Land of Nephi. I also, in preparing the analyses for this episode, found what appeared to be a second inconsistency: at one point in the war chapters it discusses a Lamanite army crossing the “head of Sidon,” presumably meaning the headwaters of the Sidon River, in order to attack Nephihah. The problem is that Nephihah is, as with Mulek, on the coast of the East Sea, quite a fair distance from the headwaters of the Sidon, which can be found near the narrow strip of wilderness separating the Nephite and Lamanite lands. (Note that this problem goes away if “head” is a reference to the river’s mouth rather than its headwaters.)

So while there could be very reasonable explanations for these apparent inconsistencies, it wouldn’t be fair to call it pristine—there are some instances that could be construed as errors or inconsistencies in the geography of the text, though these instances are decidedly vague, and, importantly, decidedly infrequent. It’s very possible to build a detailed internal geography of the Book of Mormon’s lands and cities, one detailed enough to rule out a number of the potential external geographies that have been proposed over the years.

What theories are available to us to explain this level of consistency?

The Hypotheses

The geographical consistency of the Book of Mormon is produced from authentically ancient authors familiar with their own surrounding geography—Under this theory, the book is as consistent as it is because the book is authentic, and the book’s ancient authors knew their own geography well. Inconsistencies, on the other hand, would largely be due to the fact that we, as modern readers, don’t fully understand that geography, and don’t have the context necessary to keep it as straight as we might prefer. We also wouldn’t necessarily expect even an authentic book to be fully consistent—scribal errors or misinterpretation on the part of the book’s ancient editors could easily lead to an apparent inconsistency.

The book is consistent because Joseph’s memory was sufficient to track all of its locations as he produced the dictated manuscript—This hypothesis would assume that the book was produced by Joseph, and that he possessed enough of a prodigious memory to not only be able to start dictating from where he last left off without being prompted, but to hold a complex internal geography in his head, deftly referencing it without making egregious continuity errors.

Prior Probabilities

Since we’re not going to be incorporating this evidence into our overall analysis, we’ll be skipping establishing prior probabilities, though it’s worth a reminder of where we are and where we’ve been on our journey this far:

Consequent Probabilities

CH—Consequent Probability of Ancient Authorship—If the events and locations of the Book of Mormon were recorded by ancient authors (and subsequently dictated to modern scribes), would we expect to find two potential geographical errors such as the ones noted above? Well, we know even historical observers, recording real events, can get things wrong. We also know that scribes, particularly those who recorded the words of the Book of Mormon, could and did make typographical errors that substantially altered the text, including some that have been retained in the text we have today. It’s also possible for us to misconstrue the text in a way that makes it seem as if an error is there—for instance, it’s not impossible for there to be two distinct cities named Mulek, particularly since Zeniff’s (largely Mulekite) migration into Lamanite territory could have left its mark in terms of place names (though it’s probably more likely that Mormon meant to write that Mulek was in “the land of the Nephites” instead of the “land of Nephi”).

In that context, the fact that there’s only two potential inconsistencies is kind of a minor miracle in its own right, even with an authentic Book of Mormon, and the two that we do have wouldn’t have been at all unexpected. I have no problem assigning this evidence a consequent probability of p = 1.

CA—Consequent Probability of Modern Authorship—But how many errors would we expect if the Book of Mormon was the fictional work of a modern author, and one who was dictating the text without notes and without revisions? We take consistency for granted in the books we read, but that consistency is the product of fastidious editing and notetaking. It is, of course, possible for careless revisions to introduce consistency errors as much as remove them, but modern publishers have to work very hard to make sure such errors are removed from the text before it hits the shelves. Tolkien himself made continuity an obsession, relentlessly revising and relying on an extensive set of notes and maps to keep his fictional world straight. The odds would not have been in Joseph’s favor as he tried to do the same for the world of the Nephites and Lamanites.

Just to make sure I wasn’t completely off my rocker, though, I consulted with a friend of mine who happens to be a prolific and popular author of mainstream fiction—he’s published some 17 novels over the course of the last decade. I asked him two questions: 1) about how many continuity errors would you expect to find in the first draft of a novel?, and 2) in the context of the theory that Joseph dictated the text in a single draft, how consistent is the Book of Mormon? His answers were a bit terse, but telling:

1. In a first draft? Tons.
   
2. Wildly, astonishingly consistent.
   

But we want to get something a bit more precise than “tons,” and making that sort of estimate is harder than it sounds. It seems pretty clear, though, that it’s an issue of memory—every time Joseph would have attempted to recall where a location was, he would have a chance to forget, or at least to remember it incorrectly. Just how many chances he would have to forget is difficult to say, but we can make a conservative estimate using the number of geographic relationships noted above. We could assume that for each relationship, he would have at least one chance, on average, to mess that relationship up, leaving us with an estimated number of 151 chances to make an error in recalling geographic details.

We’ve dealt with issues of memory before, in the context of Joseph remembering where he left off in the previous day’s dictation. We could be tempted to use the same probability estimates as we did there. But it’s worth seeing if we can find a better fit in the literature for recalling details about a complex array of different items with varying characteristics. There still doesn’t seem to be any studies that have looked at that specific sort of task, but we might be able to get close.

Though I considered several types of memory tasks, only one had the sort of characteristics that might correspond to trying to remember the relative position of a city or landmark in an internal geography. For one, we want the task to be a test of long-term memory rather than working or short-term memory—Joseph would have had to remember these locations over a series of days, or even months. Second, we want the task to involve recalling items from a potential set of similar confounders or distractors, since Joseph would have had to accurately recall that, say, Manti, rather than Mulek, was located in the middle of the narrow strip of wilderness. Such a task would involve distinguishing between similar items rather than merely recalling an item’s characteristics.

This brought me to the appropriately named Plagiarism Memory task. In this task, a person is brought into a room with a confederate, and both of them are asked to provide six creative solutions to a particular problem (e.g., an environment-related policy issue). Then, a day later, the participants returned, and were asked to recall their own ideas, as well as the confederate’s ideas, and then to generate new ideas listed by neither them nor the confederate. This might seem like a relatively simple task, but it’s harder than it sounds. Often people would recall the ideas, but would get the source of the idea wrong, attributing their partners ideas to themselves and vice versa. They also had trouble coming up with genuinely new ideas, and, more often than not, unconsciously stole from their partner (or from themselves!) when trying to generate new solutions.

Though it’s not a perfect fit for the situation Joseph would have found himself in when maintaining an internal geography, there are some definite similarities—Joseph would’ve had to recall things he said a day or more prior, and he would have to make sure not to confuse them with yet other things that he’d said. And forgetting who came up a particular item should be pretty comparable to forgetting whether one city is north or south of another, or whether it’s near the coast or near the wilderness. In the study, the error rate on the recall task was somewhere between 18.3% and 33%, depending on the experimental condition, and Joseph’s error rate should have been somewhat similar. For the sake of a fortiori reasoning, we’ll use the bottom of that range. When applied to the 151 geographic relationships in the text, that means he should have messed up somewhere on the order of 28 times, substantially more than the 2 we’ve been able to identify.

Since the 18.3% group also had a standard deviation of 4.3%, that lets us estimate the probability of observing 2 errors (1.3% of 151) using Z scores. That level of consistency would have put him 3.95 standard deviations away from the mean, with an associated probability of p = .000079. We’ll go ahead and use that for our estimate.

Posterior Probability

We’ll just be calculating an evidence score for this analysis, as outlined below.

CH = Consequent Probability of the Hypothesis (the probability of observing a consistent internal geography if the book was written by authentically ancient authors, or p = 1)

CA = Consequent Probability of the Alternate Hypothesis (the probability of observing a consistent internal geography if the book was written via dictation in the modern era, or p = .000079)

Lmag = Likelihood Magnitude (an estimate of the number of orders of magnitude that the probability will shift, due to the evidence)

Lmag = log₁₀(CH/CA)

Lmag = log₁₀(1/.000079)

Lmag = log₁₀(12658.23)

Lmag = 4

Conclusion

Though it wouldn’t have made much of a statistical splash in the grand scheme of things, my analysis suggests that the Book of Mormon’s internal geography is indeed unexpectedly consistent. With a complexity that rivals that of fiction’s greatest worldbuilding, we would absolutely expect Joseph to have produced more geography-related continuity errors than the paltry set on offer.

This, of course, would be much less unexpected if Joseph had the type of savant-level memory that would’ve been needed to produce the dictation evidence we have, or the narrative consistency we see elsewhere in the Book of Mormon. Critics are usually quick to note that there are a number of people with savant-level memory skills who can do amazing things, like recite pi to thousands of decimal places, or tell you the day of the week from calendar dates hundreds of years ago (though it turns out this one isn’t too bad if you memorize a few key facts). But one question is fair to ask in that context: outside of his dictation of scripture, did Joseph give any indication that he had this sort of savant-level memory elsewhere in his life? It’s easy to assume that he must have given what we have with the Book of Mormon, but it’s hard to build a case that stands in isolation. A few quick pieces of contradictory evidence even come to mind, such as Joseph’s failing to remember that Jerusalem had walls, or forgetting whether he said he was 14 or 15 when writing down the story of the first vision.

If the critics would like me to believe that Joseph was capable of dictating an extensive and consistent internal geography, they have their work cut out for them tracking down similar instances of consistent recall in his personal and professional life. Until then, I can more fully appreciate the world of Nephi, Alma, and Moroni, treating the stories that take place there with a considerable and well-earned credulity.

Skeptic’s Corner

One might wonder what the point of a skeptic’s corner would be when this one is essentially a gimme, but we’ll go ahead and think through some of this anyway. It goes without saying that it would be nice to have more direct studies of inconsistencies in first drafts or more robust inventories of opportunities for contradictions in the Book of Mormon. More applicable tests of memory would be pretty useful too, including ones that test how and whether memory on the two different dictation-related tasks we’ve taken a look at here (including the resumption of dictation without prompting) are independent of each other.

It might also be interesting to think through some different alternative assumptions. Instead of assuming that Joseph relayed the book in a single draft, we could take a look at how many drafts it would take to get a book as consistent as the Book of Mormon. My guess is it would generally be more than two—Brandon Sanderson, for instance, likes to do six or seven, not including proofreading by editors. You could then take a stab at how long multiple drafts of the Book of Mormon would take to put together, which might lead us to question just how much paper Joseph would’ve had access to, let alone gold.

Next Time, in Episode 18:

When next we meet, we’ll take a look at the perfections and imperfections of prophets, both living and dead, and determine what kind of standards might be reasonable to apply to the prophetic mantle.

Questions, ideas, and some fraction of a thousand cuts can contribute to the death of BayesianBoM@gmail.com or submitted as comments below.

Appendix