Deductions vs. Speculation

[camgarsky4]

Wanted to share my view on the differences between fact and logic based deductions vs. speculation based on loosely fitting or unsubstantiated facts.

Screenshot 2026-06-24 135408.png

Below is an example of each.

Facts:
[*]Prince Albert coats were designed for outdoor wear and social occasions. Not worn casually around the house.
[*]Andrew's prince albert coat was kept in the dining room closet (2-3 feet behind Andrew's head)
[*]Upon arriving home from his walk, Andrew went straight to the dining room.
[*]The murderer stood basically in the dining room-sitting room doorway when he struck. (standing1-2 feet from the closet).
[*]A Prince Albert Coat was found wadded up and propping up the head of Andrew Borden
[*]Zero suspects were identified because of blood splatter being found or seen of their clothing.

Deductions:
Upon coming indoors, Andrew took off his coat and hung it in the dining room closet.
The killer used the Prince Albert coat to shield their body from blood splatter.

In this case, there is no conjecture or hypothesis used to form the deductions. They are direct, logical derivatives of the established facts.
All the facts are directly related to the Prince Albert Coat and Andrew. But being a sound deduction does not mean it is accurate. It is just the most plausible explanation for the facts as we know them.


Another example

Facts
[*]John Morse visited the Swansea upper farm the day prior to the murders
[*]Per a letter Morse possessed, Andrew asked if he knew someone to manage the upper farm
[*]Morse owned a farm in Iowa and had previously engaged in some form of business which resulted in horse sales

Rumors:
Victoria Lincoln's grandfather supposedly told someone who told Victoria that Andrew was going to sell some Swansea land.

Speculation
Andrew offered to sell Morse the upper farm for livestock. (speculation goes on with a few theory options...including AJB reneging & Morse murdering him or keeping the sale a secret from Lizzie, so Lizzie murdered him)

In this example, the facts have a 'farming' flavor to them and therefore were forced into being related and then a rumor was used to coalesce into a speculation based theory.

[Lorcan]

This is basically what I am building in software with a few database tools that help with data analysis.

1. Every line of testimony has a unique ID number.
2. A claim is an atomic statement that can't be broken into smaller pieces. Every claim has a unique ID number.
3. Every atomic statement that can be used as EVIDENCE_FOR or EVIDENCE_AGAINST also has a unique ID number.
4. Every Evidence statement has a unique ID number.
5. A dependency chain can be constructed around an INQUIRY (such as The Barn Alibi) which would contain many claims and lines of evidence.
6. All claims and evidence statements would have a timespan and location ID.
7. A dependency graph can then be created.

In my KGSherlock (knowledge graph sherlock) software, it would work like this:

Each of your facts, deductions, rumors, and speculation would be broken into atomic English sentences and assigned a unique ID number.

I want to issue an EVIDENCE_AGAINST statement against one of your facts, let's assign unique ID number 0000006 to the statement "Zero suspects were identified because of blood splatter being found or seen of their clothing."

My EVIDENCE_AGAINST would have the unique ID 0000006A001 and the text would read "A blood droplet that impacted the outside of Lizzie Borden's undergarment was identified. I argue this is more consistent with splatter from a victim than menstrual blood, since it impacted from the outside."

Now, people can dispute me by filing an EVIDENCE_CLAIM against my EVIDENCE_CLAIM, etc.

Because the data is structured in standard database format, graph algorithms, visual reports, gantt charts, flowcharts, data exports, etc. can all be done with it.

[Lorcan]

Once data lineage and dependencies are established, you can filter on "only show me claims tied to sworn testimony" and "show me all the claims tagged with the category of finance or real estate" or "sort all claims by timespan begin time and visualize a gantt chart including the people [Lizzie Borden, Bridget Sullivan, Abby Borden].

Things like this are all features I want to learn how to build.

[camgarsky4]

My goodness...you are really a genius. I had forgot about the spot on the undergarment so your system would have let me know and I could then tighten up the wording of facts and deductions. I love it!

I think the spot came from Abby's death (assuming Lizzie is killer) because of straddling part of Abby. So I could have tailored that info into how I worded the facts for Andrews death. That means for better communication amongst the 'sleuths'!

[camgarsky4]

Facts:
[*]Prince Albert coats were designed for outdoor wear and social occasions. Not worn casually around the house.
[*]Andrew's prince albert coat was kept in the dining room closet (2-3 feet behind Andrew's head)
[*]Upon arriving home from his walk, Andrew went straight to the dining room.
[*]The murderer stood basically in the dining room-sitting room doorway when he struck. (standing1-2 feet from the closet).
[*]A Prince Albert Coat was found wadded up and propping up the head of Andrew Borden
[*]Lizzie had a single droplet on her undergarment. No other suspect was identified with blood splatter or stains.
[*]The killer of Abby most likely straddled her as the final blows were struck.

Deductions:
If from one of the victims, Lizzie got the droplet of blood during the attack on Abby.
Upon coming indoors, Andrew took off his coat and hung it in the dining room closet.
The killer used the Prince Albert coat to shield their body from blood splatter.

In this case, there is no conjecture or hypothesis used to form the deductions. They are direct, logical derivatives of the established facts.
All the facts are directly related to the Prince Albert Coat and Andrew. But being a sound deduction does not mean it is accurate. It is just the most plausible explanation for the facts as we know them.

[Inspector]

I agree with the conclusion of the Prince Albert jacket as being plausible, yet not necessarily factual. One could even rate the scenario as probable and still have sound reasoning to support it.
Of course it can not be considered a concluded fact that this was the route taken by the PA jacket, because there are other possible reasons for how the jacket found its way under the victim.


What an endeavor you have embarked upon Lorcan, and anyone who has studied the Borden case for a while should easily see how your database could help hone in, and weed out, with the end results hopefully getting closer to actual facts .
Another attribute I can invision, is the unveiling of speculation, so it can’t be disguised or hide itself as fact as often as it does , and that could be helpful for those who have presuppositions that blur .
The old saying that numbers don’t lie.

[TeenaBee]

This is so fascinating! I definitely appreciate the parsing of deduction and speculation, that makes sense to me, I can clearly see the difference. I love the mathematical approach, and how that makes speculation more evident. I do think you can come up with more reasonable and absolutely plausible conclusions that way. I can see how it can help mitigate the emotional attachment one can develop toward seeing things a certain way...

Do you suppose it's truly possible to remove the human element in judging a case like this? You say that through deduction you can reach the "most plausible" explanation for a string of related facts. I still find myself wondering, most plausible to who? (What Victoria Lincoln found plausible still makes me shake my head, but like you said, clear example of speculation rather than deduction). Just as AI needs specific prompts to come up with an answer or to design something, don't we still have to decide which factors to feed into the deduction formula? Or, Lorcan, maybe are you talking about a system that does away with cherry-picking and takes the whole of the court record into account? Not just trial, where evidence was withheld from the jury, but inquest, prelminary hearing, etc. Maybe witness statements, as well. I think that would be pretty amazing...

[Lorcan]

The best way I know of to mitigate the problems with subjectivity when evaluating the evidence is to not try to calculate the "truth" but to provide a data structure where the entire data lineage and dependency chain is visible and every link of that chain can be seen, challenged, and evaluated, recursively, by everyone.

What I'm aiming for is the optimal data structure where the sources of claims are transparent and the data structure file format is portable to different analysis tools, ranging from beautiful, human-readable, infographic reports all the way to some use cases that would require cryptographically hashed write-only ledgers (anti-tampering).

What you'll hopefully end up seeing is that the Borden Case is messy enough to be a really effective training ground to develop version 1 of that data model. Once the model is robust enough, then all kinds of tools and use cases become available, centered around helping both people and AI ground their analysis in what is commonly called a "knowledge graph" that is auditable and open, using a standard file format.

What I plan to do is release some free, open-source, software called KGSherlock and include the Lizzie Borden case as the first example of this data structure. I have a larger plan that is more long-term, but at the core of it are data structures that make claim defensibility transparent.

I'll soon have a plaintext version of every single sworn Q&A from the Inquest, Preliminary Hearing, and Trial (within a couple of weeks, I hope). Sworn testimony is the core reference. The reason for doing this is then I can design the software to add linked enrichment data to the original sworn Q&A rows and everything can be uniquely numbered so database queries can work. I'll be linking additional data to the immutable sworn testimony (except for correcting errors in my work getting the Q&A into this format. I hope to eventually have as accurate a copy of the testimony as possible). Here's a list of things I plan to do:

1. Extract atomic claims linked to numbered rows of testimony. One testimony answer may make many claims.
2. Tagging the testimony with topic labels for later querying and filtering.
3. Building period-appropriate timespans for claimed activities (Being careful to account for the time-difference required to do things in 1892)
4. Tying locations, people, and timespans together
5. Data visualizations that can help with contradiction detection.
6. A system for adding evidence_for and evidence_against to claims, recursively, with safeguards (evidence must link to a source)

I don't have a verifiably safe and easy way to get the entire Borden case data structure to all of you yet, but as the project matures I'll work on it.

[camgarsky4]

To my 'old fashion' method I'll add ASSUMPTIONS to go after facts and before deductions. Also need to add other possible answers.
In the case of the Prince Albert Coat, add that Andrew could have placed the jacket under his head and provide why that wasn't chosen as the best solution.

Yes, since we don't know the answer to open questions, we can just do our best to assemble an answer. Deductions will end up being thru the eyes of the beholder, but they will be build upon facts, not speculation built upon unsubstantiated information.

Example:
Alternative solution: Andrew placed the jacket under his head when he laid down to rest.
Excluded because Andrew knew Morse would soon to arriving to eat and chat, and folding up the jacket to improve his pillow solution is interpreted as an intention to take a nap, not just relaxing in advance of receiving a visitor.
Excluded because the dining room had a lounge and probably had a pillow which AJB could have used instead of his business coat.
Excluded because Andrew's persona appeared to be someone who would be particular in how they treat their belongings and clothing when there were other convenient options.

[Lorcan]

For the software, I would have a media folder for photos, pdfs, etc. and only store the text reference inside the database, but I'd file another evidence_against claim that Andrew placed the coat under his own head, as seen in the photo.

My claim is it would be anatomically impossible for Andrew to have shoved the tail of his coat under his pillow in that way and it is far more consistent with someone using the coat tail to position the pillow and Andrew's head.

Of course, that would leave me open for someone to claim that an innocent official used the coat tail as a shield to adjust the pillow and prop him up for the photo or to examine his wounds.

[Inspector]

Lorcan, I’m glad you’re using your knowledge for other areas . Though the Borden case is a challenging search for whatever one is searching for, your expertise would be far better served in other aspects of life as I’m sure you are aware.
I know you probably desire to help others in many ways, and with your computer, and modern knowledge approaches I think it’s a noble cause.
Perhaps something good can come out of the Borden case after all.

[Lorcan]

Your intuition is correct. The Lizzie Borden case is the first stress test of a much larger idea and long-term plan I have to develop a universal file format and data structure that can sit in data analytics pipelines and other software to act as a claim visibility layer. If AI is given the complete data lineage of a claim, with all the claims and counterclaims surfaced, all the dependencies and disputes, and then it is forced to give a referenced decision tree before making a confident-sounding judgement call, I think the world will be a better place and it doesn't just apply to AI, and it goes far beyond amateur sleuthing.

There are many areas in life where if a malicious actor were asked to produce a complete data lineage and decision tree, it would be a lot more difficult to manipulate people. There are technologies to help to tamper-proof and verify reference data sources while maintaining a level of privacy: zero-knowledge proofs, Blake3 hashing, blockchain, etc.

What's your take on that coat tail being shoved so steeply behind the pillow? I know the image contrast isn't great but I tried to show what I meant with the orange arrow pointing toward the area of the pillow I'm referring to.

[Inspector]

I don’t think the coat was used in any way to help move Andrew, or prop him. I believe it was Bowen who testified that Andrew had slumped down some as I’m sure you recall, so I think the photo shows an undisturbed Andrew.
I think the PA is a mystery, although some say his boots were put on for the photo, I haven’t seen any proof or testimony it was done.
I believe Knowlton presented in his closing that the PA could have been used as a shield, and I think the odds are that he put it in the dining room closet where he normally kept it, although we don’t know for sure if he did.
However, if Andrew did hang it up, he didn’t take it back out, and tuck it under the pillow in my opinion.

[camgarsky4]

The boots going on as prep for the photo is a derivative of modern day sleuths trying to explain how Lizzie could have helped remove his shoes, as she stated, but the shoes are on in the photo. There was no contemporary mention of putting shoes on.

[Inspector]

Thank you for the information on the boots.
I put a bunch of thought into them a while back, but didn’t really gain anything significant. I think I confused myself more than anything.

[Inspector]

To throw more confusion into the mix.
Dr Dolan says he may have moved the PA jacket to get to the inside pocket. This would be before the photo was taken.
(Trial, 854)

[Lorcan]

Hmm, is he saying he pulled on the balled up Prince Albert coat to feel around for the inside pocket - presumably, he checked the outside pockets without unfolding or unrolling it. Here is the testimony.

Q. Now as to the position of the body when you saw it, looking at exhibit 17 (photograph), how does that correspond with your remembrance of the position of the body when you saw it?
A. That is right. The only difference possible, the only thing that I am not quite sure of, is that possibly I may have pulled that coat up a little to get to his inside pocket. I am not quite sure of it.

Q. Which is the coat, sir?
A. This is it: His Prince Albert coat.

MR. KNOWLTON. (To the jury) That is the coat, he says, right over there, over the head of the body.

Q. You pulled that up a little, did you: otherwise is the position the same as you found it?
A. Yes, sir.

MR. KNOWLTON. (To the jury). That is the picture that you have already seen, gentlemen.

Q. What was the head resting on?
A. The head was resting upon a small sofa cushion that had a little white tidy on it. The cushion in turn, I think, rested on his coat, which had been doubled up and put under there, and that, I think, rested upon an afghan or sofa cover, a knitted affair.

Q. The lowest of the three was the doubled up coat?
A. No, sir.

Q. Was the---?
A. Afghan.

Q. Then came the coat?
A. Yes, sir.

Q. And then the sofa cushion?
A. Yes, sir.

[Lorcan]

FYI - another entry for our things were different in 1892 files:

In the Victorian era, "tidy" and "doily" were closely related household items, often made of the same white lace or crochet work, which is why the terms were sometimes used interchangeably.

The Tidy (or Antimacassar)
A "tidy" was a protective cover placed over the backs and arms of chairs or sofas. It was used to keep furniture clean from "Macassar oil," a popular and greasy men's hair styling product of the 19th century. Because these cloths kept the parlor looking neat, they were colloquially called tidies.

The Doily
A "doily" was named after Monsieur Doily, a 17th-century London linen draper who sold lightweight fabrics. Over time, the name came to describe small, ornamental mats placed on tables under teacups, plates, or decor to protect the wood from scratches and spills.

In the context of the 1892 Lizzie Borden trial, when a witness referred to a "tidy," they were talking about one of those decorative, protective lace cloths draped over a parlor chair.

[Inspector]

Lorcan, I think anyone’s guess is as good as the other. It must have been of some change from the original jacket position, but it’s hard to say . In the extreme, he could have pulled up and accidentally pulled the sleeve as he tried to get to the “inside pocket.”
It sounds to me that the jacket itself as a whole, may have been tucked under the pillow further than the photo shows.

[Inspector]

Couple things I’ve gathered from Spencer recently, without going into detail.
1) Lizzie’s inquest was not allowed at trial because there was an arrest warrant for her when the inquest was done.
Ironically—the first warrant wasn’t used, rather a second one was written and used, so her inquest was disallowed due to a warrant that was never used.
Talk about fortunate for Lizzie.

2) The prussic acid testimony wasn’t allowed because the poison could have been used for other things.

These are the most basic ways I know to say it, although there was miriads of testimony, and arguments to read through.
The judges made their statements fairly understandable at the end.


Spencer page. 569. CALB
When the witness, a drug clerk, Eli Bense Isic by name, began his testimony Miss Borden fairly glared at him, leaned forward and stared him squarely in the eye. It was a new departure and possibly he may not have been prepared for it. However, that may be, the clerk, a good natured fellow, blushed and stammered as he hurriedly replied, "I do," before clerk Borden had finished the phraseology of the oath. (BG, June 15, 1893)