> fiber analysis is also fake as hell (like saying: these threads came from your carpet, or your shirt).

I'm not a forensic technician, but I've watched probably every episode of every true-crime forensic documentary TV show / YouTube channel / podcast / etc that exists; which probably has some weak but non-negligible correlation with having a gestalt impression of what people do and don't use forensic techniques for in practice. (At least in the "solved" cases, since the point of these shows is "how they solved it.")

In every use of "fiber analysis" I ever saw covered, where the "fiber analysis" was actually used to prove something, the thing they're always noticing is "trilobal fibers" — presumably because, at least for some period in recent history, those type of fibers only ever got used by specific manufacturers for specific use-cases (vehicle manufacturers, for car floor matting, IIRC); and weren't part of the composition of anything sold to individuals at retail (except, perhaps, replacement car floor mats.) So "trilobal fibers on the body" could be taken to imply "this person was at some point in a car footwell/trunk."

I don't ever recall any case that made a "fiber analysis" claim other than that particular one. (Maybe once someone claimed something about what proprietary dyes were used on the fibers via GCMS assay — but that's not really "fiber analysis" per se.)

I don't have a link handy, but there was a huge FBI scandle where their experts were essentially calling things a match to a specific person, etc.

But yes, the science is fine if interpreted correctly. The correct phrasing would be exclude or not exclude a suspect. Or for your example, it could be consistent with someone having been in a vehicle. The problem is when the experts say they must have been in a vehicle, or especially a specific individual vehicle.

That seems something that should be criminal. And realistically always asked how many vehicles with this type of material is there around?

The judge should have thrown it out.

Second thing they teach in forensics is that unless I witness the actual event, it is really hard to tie it to a specific person. I can present dozens and dozens of circumstantial evidence which has sufficient weight for a reasonable person to make that inference, but I cannot attest to that.

Judges should do a lot of things. In reality, their quality and knowledge varies greatly.

Any judge worth their salt would prohibit or censor witness testimony making conclusory statements, especially based on circumstantial evidence. This is Evidence 101. Everybody knows the so-called "hearsay" rule, but almost every rule in evidence revolves around one core precept--no evidence is per se conclusive, and the function of testimony of any sort isn't to submit conclusions to the question at issue.

But implying conclusions is exactly what both the defense and prosecutor are working at; they're constantly playing word games, and some judges would certainly be better than others at seeing through the B.S.

Judges are elected in many areas of the US so they're potentially literally some rando in a robe who wants to put "bad guys" in jail, not some master of critical thinking.

Yeah, and the other way you get judges is via appointment. So there's a chance you have a hack that put in because the executive leader likes their politics.

Thank you for the link! Though I would say it highlights a different problem (investigators' bias) than stated. The Spanish police have rejected findings apparently.

> nobody knows how common it is for someone to have very similar fingerprints to another person

Your last example is definitely a good counterexample to this, but I feel in most of cases, the suspects are very few so fingerprints only need to be able to distinguish one of the others, than say, identify a random dude from billions. So it's still useable.

It's like while SHA-1, or even CRC32, is totally incapable for security, it's still perfectly fine to be used to detect if a known file is corrupted (through natural cause, not deliberately altered) or not.

That's like saying if a witness says "he had brown hair, that's all I can say", and only one of "the suspects" had brown hair, that's reason to convict him.

You are misrepresenting the odds of a group of people having an outlier hair color vs a group of people having one which matches sufficiently well to a piece of evidence which excludes the others.

'I remember a brown car' vs "I remember a brown 1976 Ford coupe with worn tires and a broken side window'.

please don't forget all the databases that can search for fingerprints, and all the suspects brought before the detectives, and at that point if they did not have a good enough alibi, or good enough psyche to withstand interrogation, or ...

That's a fair comparison, although I don't know if the numbers work out to a 1976 Ford Coupe or more like a 2005 Honda Civic. But the other thing is, fingerprint analysis is also usually expressed as more like "a 70% chance of being a 2005 Honda Civic with worn tires and a broken side window", the match is always partial (another thing that science doesn't support, those %'s given).

But would a jury think someone having a 2005 Honda Civic with worn tires and a broken side window is alone reason to convict someone? Probably not? Would they think "A 75% match with [fingerprint equivalent]" is alone enough reason to convict someone? Usually. Because they are very misinformed about what fingerprint "science" can and does do.

That is not how criminal convictions (is supposed to) work. You don't have to present evidence that proves that, out of these three people, this one is the most likely killer. You have to prove that, beyond any reasonable doubt, this person has definitely committed the crime in question. If there is some reasonable possibility that some unknown person could have come off the street and committed the crime while leaving fingerprints that happen to match one of the suspects, then that suspect should not be convicted.

>supposed to "supposed to" is doing a lot of work there. People routinely get railroaded in criminal trials because they are too poor to afford good defense or for even sillier reasons like the prosecutor is better looking or more charismatic than the defense attorney.

Absolutely.

But making claims like "fingerprints are good for choosing which of the suspects is more likely to have done it" is harming the situation you describe, not helping. That is, it puts even more onus on the defense attorney to explain how the process is even supposed to work.

The totality of the evidence must prove beyond a reasonable doubt. But not every piece of evidence must provide beyond a reasonable doubt.

If all you have is a finger print--then yea--the possibility someone else with a close match did creates reasonable doubt.

But if the thief was seen in the area at the time and you have a finger print, well that's less reasonable now.

TL;DR; apply Bayesian probability.

I'm not saying you should be convicted by a single fingerprint.

Forensics evidence is almost always circumstantial.

For your example - If everyone wears wool sweaters except one person wearing poly microfibre, what is the inference?

The first rule that is taught in forensics is that every interaction leaves a trace of some sort. The poly microfibre showed up at the scene. How did it get there is now a different type of forensics and where lawyers step in. The prosecutor will insist because the poly microfibre owner was there, while the defense lawyer will insist, there was a party the night before and everyone was rubbing against each other, the poly microfibre wearer and the wool sweater wearers. It can be anyone.

> Fingerprint analysis is not based on solid science, nobody knows how common it is for someone to have very similar fingerprints to another person, and nobody knows what the right standard for "points of similarity" is.

Weird, I mean surely we have enough data to answer these questions with high certainty? Makes you worry that somebody won’t like the answer…