Pied Piper and William Manchester (and AI)

In his book World Lit Only By Fire, William Manchester makes claims that the Pied Piper of folklore was a real person:

a psychopath and a pederast who on June 24, 1484, spirited away 130 children in the Saxon village of Hammel and used them in unspeakable ways. Accounts of the aftermath vary. According to some, the victims were never seen again; others told of dimebodied little bodies found scattered in the forest underbrush or festooning the branches of trees.

This got my attention to say the least when I read the book around age 15. This stop motion film about the Pied Piper creeped me out/held me enthralled around that time.

I went looking for more information.

And… this is totally made up. “Accounts of the aftermath” don’t “vary,” they don’t even exist!

Where did Manchester get this? Was it pure fiction? Can we trust anything Manchester says on anything else? Or in his other books, on Churchill or on his own experiences in World War 2?

As the Internet emerged I was able to learn more about what primary sources exist about the Hamelin legend. Nothing like what Manchester describes has been found by anyone. Other curious people have been baffled and infuriated by what Manchester did in his (very popular) book. A theme of Helytimes is going back to the source. Who saw it? Who was there? How do we know? What do we know?

Now a new tool has emerged: LLMs. I asked one, Perplexity, about the Hamelin story, and it offered to write me a cited PDF on what we might now. I wouldn’t trust anything it says without double-checking, but this is pretty impressive, and it presents what I believe is a plausible theory, that a sort of recruiter led away young people (not necessarily children) to settle in places north of Berlin or in (what’s now) northwest Poland.

The ability of the LLMs to generate pretty good answers to historical research questions is remarkable, and will certainly change my life as a person who likes to chase down answers. Here’s another example. I was wondering if there were any primary sources for the quote of Napoleon about “I found the crown of France lying in the gutter.” The LLM can do something I can’t do very easily: read French language sources.

Here’s one about the history of the Chateau at Annecy, which is way better than the current English language Wikipedia page:

Now, I’m aware that among the youth AI is suspect. Possibly the AI is like a piper leading me, an innocent kinder, to a land where my brain won’t function anymore because I’ve delegated too much. But it seems more likely this is just the latest tool than can take us farther faster. The main argument I can come up with against it is the Shelby Foote one – that in doing one’s own legwork you inevitably find other interesting material you didn’t know you were looking for. That’s true, but on the other hand, won’t that still be the case even if I’m having an LLM doing some of the work? Might it not in fact more speedily direct me to other material?

(The energy use for this kind of query does not seem to be significant. Perplexity estimates its equivalent to running a laptop for 25 minutes.)

The LLMs may make mistakes, or have biases, or whatnot, but remember I got here because a well-credentialed human either invented a lie, took a troubling level of literary license, or was delusional.

The implications of this are pretty staggering. The kind of work I did in college can be pretty well replicated in minutes. That could be bleak, if you’re attached to or dependent on the old ways. There is a way to view it optimistically, we can free ourselves from a lot of time consuming drudgery, and uncork massive new potential.