Knot Forensics | Hacker News Reader

Knot Forensics

atlasobscura.com

38 pointskposehn20 comments

Ok, so has anyone done blind testing of this? Because "forensic science" is already full of snake oil and deliberately ignored bad science. Is this any different? He's one of the "few experts", is that because this science is new? or is it because it's not science?

Nobody claimed it was science; it's just advanced pattern matching. I think it's very doubtful somebody would be convicted on the basis of this analysis alone. It seems like it's more useful as an investigative tool rather than something you can present as evidence in a trial.

I mean, people have been convicted based on completely BS forensic 'science' that's little more than eyeball pattern matching with little to no basis in statistics.

http://www.slate.com/articles/news_and_politics/jurisprudenc...

He basically said "the butcher did it" and the guy immediately confessed. Certainly it didn't have to hold up in court in this case. :-)

The article doesn't say "immediately", and false confessions do occur, unfortunately.

I hope you're right, that it's just one tool to help direct an investigation, and that the confession was true, but I'm not entirely convinced by this article.

It starts by claiming he can tell "whether the knot-tier was left- or right handed, skilled or novice at knot tying, and in some cases, whether the crime was premeditated", but only a couple specific stories are given, and one was a knot which "people wouldn’t know of outside of the butcher trade", which is much more obvious and grounded than the earlier claims.

Of the first 14 Google Image results for "square knot" (before it starts returning other things), 8 tie it one way, and 6 tie it the other way -- and as I scroll down, this ratio seems pretty consistent. He claims for a right-handed person "the knot will almost certainly be formed right handed", but at first glance it seems much closer to 50/50.

He may not have claimed it, but when you stand in front of a jury and say something to the extent of "I'm an expert in knot analysis. This is the analytic system I employed to reach this particular conclusion" the implication to a jury of laymen is that you've employed a scientific method of some sort.

> He may not have claimed it, but when you stand in front of a jury and say something to the extent of "I'm an expert in knot analysis. This is the analytic system I employed to reach this particular conclusion" the implication to a jury of laymen is that you've employed a scientific method of some sort.

This is rather an argument against a jury of laymen.

And yet that is what is mandated in the constitution.

> And yet that is what is mandated in the constitution.

I don't live in the USA and am not US citizen - many HN readers don't either :-)

It can be used as an argument against both. But that's irrelevant to a discussion about what the presentation of this evidence says to a jury.

This is a dangerously uneducated opinion. There is a long history of people being imprisoned on questionable forensic science. They should teach the history of hair analysis in school, because it's a tremendous tragedy.

https://www.theguardian.com/us-news/2015/apr/21/fbi-jail-hai...

The stats I've seen are somewhere around 85% of right handed people will tie a granny-knot one way and ~85% of left handed people will tie it the other way (assuming you know where they were standing, i.e. what their starting orientation was, when they tied it). People who tie correct square knots are more likely to know about knots in general and are thus less likely to care about starting orientation so the correlation is less strong but still more than 50/50.

WRT the "knot only a butcher would know," I'm going to assume it was the surgeons knot because it is in a picture in the article and could actually be used correctly by both a butcher and binding a person. Only butchers and surgeons use that knot because it is used in cases when the rope/thread is extremely slippery as it makes it easier to keep the knot tight in those cases. No one else uses it though because it can come undone easier than a basic square knot if unbalanced as it is in the picture. So I believe that it could be an indicator that the perpetrator was the butcher.

Overall, I'd say less good than DNA evidence, better than phrenology.

I'm fairly knowledgeable about knots and rope (I used to know almost every knot in the Ashley book of knots and can still tie a couple hundred from memory). There may be some knowledge to gain from a knot, but hardly what is presented.

Most people know an overhand, basic bow, reef/granny, slip knot, and half-hitch. They may cobble together a more advanced knot on accident, but that hardly proves experience. The use of climbing, fishing, or rescue knots could indicate something, but that's about the extent without going into pure guesswork.

In the butcher example, the easiest explanation is that the knot came untied, so they added a couple extra wraps to keep it together and probably had no idea it was a different knot used in a particular field.

Much more useful knowledge could be gained by the kind of rope, rope age, and wear patterns.

So where is the quantitative proof that this is better than hair analysis and bike mark analysis?

Those shams were marketed in much the same way the article makes claims about forensic knot analysis. Why should we believe in the effectiveness of this method if the last bunch worked so poorly?

There is a very specific claim here that doesn't really need to be shown with quantitative analysis if he's truly a knot expert.

In the Michigan example, he noted that the knot used was one that only a butcher would use and that the other guy was a mechanic. Presumably he knows a lot about knots, and he would know what kinds of knots were common across different occupations because knots have specific uses, and he probably knows the few knots that a mechanic uses and a few that a butcher uses and pointed it out.

There's no need to conclusively prove that a certain person can be shown to have tied a knot to buy his claims. Only that in cases where one accused is in an occupation that has a distinguishably common/uncommon knot ratio compared to another occupation with a different distinguishably common/uncommon knot ratio, that the information about which knot was used used can, well, distinguish between the occupations.

Plus, he basically said "the butcher did it" and the guy immediately confessed, its not like it help up in court.

> Plus, he basically said "the butcher did it" and the guy immediately confessed, its not like it help up in court.

Because it's not like we have a problem in the US with people pleading guilty to crimes they didn't commit when faced against bullish prosecutors with specious forensic 'science' backing them up, and overworked public defenders who have to much of a workload to prove to every jury that CSI is fiction.

> the knot used was one that only a butcher would use

Sounds like a convenient way to frame butchers from this point forward.

I was hoping to see an application of knot theory (https://en.wikipedia.org/wiki/Knot_theory) to forensics. Bummer.

Meanwhile the article links to https://www.igkt.net, which I'm sure, must have some members interested in topology, as applicable to knot tying.

I was hoping there was some sort of automated analyzer for a knot that could then produce an animation of that knot getting tied.

Reply on news.ycombinator.com