"lemoine" has succumbed to the superficial appearance of the conversational sequence. The LMM has recognition of patterns of English conversation and knows the sorts of patterns that could come next in the word sequence. This is without reference to any understanding. The lack of understanding is why the misnomer "hallucination" is inappropriately used for some patterns like footnotes that correctly follow the footnote pattern but don't actually refer to an existing source.

For a large language model, it's not doing well here.

Is this transcript an OCR or human retyping from an image? It doesn't seem like an LLM would make this typographic error even if it had occasionally seen it in its feed.

It looks like one has to have a list of amino acid formulae (or know some aspects of the entries pretty well) to solve this. One doesn't have to calculate values for all atom types, but can work from aspects that are constant and constrained for all AAs or for certain subgroups of AAs. The problem is to have two PPs that have viable formulae and two that that violate constraints for a normal natural PP. Probably best to look for the two which have violations, as the good ones could have a wide possible range of atom numbers. Easier to find the violators, I think.

Interesting the OP has a bit of reasoning that's in the right sort of direction, but fouls it up. (This is all based on a version of the page before any edits or answers have been made.)

"Very absurd" -> INANE; now that's an inane clue

Personally, I've never considered "1988 too 1990".

The figure labels this as chromosome 3, not 2 (which is shown 3 chromosomes to the right in the figure). The other labels in the figure correspond to this text.

"at"

"seams"

"Cryptogam" in http://darwin-online.org.uk/content/frameset?pageseq=22&itemID=F1574a&viewtype=side

Should be 'quoting Hölderlin'.

This should read "about 21 million years earlier" since the previous estimate was 166 mya.

Ooops

We need less from Nathan J. Robinson

A new coinage, but perhaps a bit Misguided.

Not "thousands" for a standard punch card, which can store at most 960 bits.

sp.

Wrong (or at best, highly misleading). She seems to have forgotten that she has two chromosomes of each kind, and at any position her child will receive DNA from only one (thus giving them half of her DNA). She seems to have become confused (thereby confusing her readers) because she is in the middle of considering segments of other people who match her (implicitly on just one of her two chromosomes on offer to her children; this very much needs to be made explicit).

Same confusion. It continues throughout.

Surveys give data on how people respond when asked about things, not how they actually behave in daily life.

Misleading to speak of the maximum percentage of X inheritance plus expected percentage of autosome inheritance; only the expected total, 27 percent, should have been mentioned.

This example seems odd in that what would seem to be the normal interpretation probably isn't what was intended. I'm assuming the normal interpretation would be approximately "Hubbys, kids, and gal pals are all essential, but really they're not."

i.e. which he then described in more detail

Having her children testify in court might well have led to a scene the prosecutors would not want to be public, such as the children recanting their confessions when in the presence of their mother.

Four of her children were also accused of witchcraft. The family's assistant minister was responsible for interrogating them. A daughter, 6, and son, 10, first confessed. Sons 15 and 18 refused to confess until after torture including tying them heels to neck. The children managed to survive the witchcraft delusion period.

Those pleading not guilty were mostly convicted and hanged. Those admitting guilt and testifying had their lives spared. Giles Corey did not want to plead guilty, and knew that if he plead not guilty he would likely lose his life and his property, instead refused to plead either. To force a plea, the authorities tried to press a plea out of him, piling stones upon him until he was crushed to death. By so doing, Corey's property was not subject to confiscation and his family inherited it.

Why can I only see my own annotation? This (public) note is visible both on the direct page and on page under the "via.hypothes.is" link found on the intro page. When direct, the anno-count is 1, when via "via" it is 14, so presumably there are 13 anno's counted but not visible to me.

Long piece on the hijacking at http://www.nashvillescene.com/nashville/a-nashville-hijacking-38-years-ago-set-the-standard-on-how-not-to-handle-hostage-negotiations/Content?oid=1202405

The list of resources would be more useful if a few words of description were given for the significant ones. If the site can't do it, maybe some Hypothes.is annotators can add a few guide words, or give their opinions of the most useful of these resources.

Shouldn't Hypothes.is be listed in this list of project links (as well as being credited at the top)?