Sherlock Holmes: What an amateur!

[Fazor]

Well, thought this was an interesting find. (http://news.yahoo.com/s/afp/20081222/od_afp/finlandcrimeoffbeat;_ylt=Ag9SOFKSGK6KoZbKr52sb_4DW 7oF)

Apparently, police in Finland are attempting to link a car theif to the crime via DNA found in a mosquito that was found in the stolen vehicle.

Personally, even if you know the guy did it, the mosquito seems circumstancial at best (particular given that the theifs alibi was that he was hitch-hiking and the owner gave him a ride).

This is the line that really struck me:

"It is not easy to find a small mosquito in a car, this just shows how thorough the crime scene investigation was," he added.

...For a car theft! There's no mention of murder or violence of any sort. That's dedication ... I can't imagine having a DNA test run to prosecute a car theif here in the states.

[Roving Philosopher]

Nonsense! Holmes would have found the mosquito with a mere glance at the car, and known in an instant it was a species found only in a small part of Finland. And rather than deal with DNA, he would have noted that the peculiar cut of the accused's left sock indicates that he is a frequent visitor to that location. :whistle:

[hhEb09'1]

That's dedication ... I can't imagine having a DNA test run to prosecute a car theif here in the states.Dedication to what though? If it got the technician on the nightly news, with teaser "CSI Helsinki" or "CSI Akron," would it be worth it?

[Fazor]

I'm not sure I understand your post hheb; but it's dedication to non-violent crime prosecution. Here, a crime like this simply wouldn't be worth the time, effort, and cost.

[tdvance]

yeah, someone smashed the window out of my car and took the Garmin. The police didn't even show up--just took a report over the phone.

[PetersCreek]

Nonsense! Holmes would have found the mosquito with a mere glance at the car, and known in an instant it was a species found only in a small part of Finland. And rather than deal with DNA, he would have noted that the peculiar cut of the accused's left sock indicates that he is a frequent visitor to that location. :whistle:

I rather think that Holmes wouldn't be so easily distracted from other crucial evidence. Knowing that only female mosquitoes feed on blood and that they fly generally upwind to feed, then downwind to a source of water in which to lay their eggs, Holmes would have reconciled the prevailing winds in relation to the car with the direction from which the thief approached and the proximity and direction of a suitable body of water.

[hhEb09'1]

I'm not sure I understand your post hheb; but it's dedication to non-violent crime prosecution. Here, a crime like this simply wouldn't be worth the time, effort, and cost.I'm saying, it might be dedication to riding the cresting wave and getting your face on the local nightly news :)

[Fazor]

I'm saying, it might be dedication to riding the cresting wave and getting your face on the local nightly news :)

Oh, yeah. :)

[LotusExcelle]

Having been to Finland I can honestly say that any sort of fun they can eek out of, say, trying to track a thief via a mosquito, is welcome and thought of as normal.

No really I loved it there and it really is a different kind of culture - they don't let things slide.

[kleindoofy]

... even if you know the guy did it, the mosquito seems circumstancial at best ...
Some legal systems don't *necessarily* view individual pieces of evidence as being singly damning or delivering. It's the overall picture and impression of the evidence and the trial that count for them. That doesn't mean that they are less discriminate in considering and admitting evidence, but it can move the accent slightly away from the "smoking gun."

I have absolutely no idea if Finland's system is like that, but I guess every bit of evidence helps, even if it causes an itch.

[sarongsong]

...amature!:lol:

[Nicolas]

Why was Holmes a motor rotor?

[mugaliens]

The only thing it proves is that the suspect was bitten by the mosquito. If he'd been smart, he would have said, "What car?"

Mosquitos do get around, you know...

[Whirlpool]

and does mosquito sucks blood more than one person , right? So it's not a hard evidence that will pin down a car thief .

[ParaDoctor]

Holmes may be good at meticulous fact gathering and statistical inference, but his deductive logic stinks:
You will, I am sure, agree with me that if […] page 534 finds us only in the second chapter, the length of the first one must have been really intolerable.("The Valley of Fear", first chapter)

Then again, the first chapter may have ended on page 42 ... :D

[mugaliens]

Then again, the first chapter may have ended on page 42 ... :D

Perhaps Doyle struggled with numbers beyond two...?

[ParaDoctor]

Perhaps Doyle struggled with numbers beyond two...?I was always suspicious of the quacks, but this leaves me speechless. :D

[AGN Fuel]

I'm not sure I understand your post hheb; but it's dedication to non-violent crime prosecution. Here, a crime like this simply wouldn't be worth the time, effort, and cost.


...but when we got to the 'Scene of the Crime', there was five police officers and three police cars, being the biggest crime of the last fifty years, and everybody wanted to get in the newspaper story about it.

[TrAI]

Holmes may be good at meticulous fact gathering and statistical inference, but his deductive logic stinks:
("The Valley of Fear", first chapter)

Then again, the first chapter may have ended on page 42 ... :D

Holmes was just pointing out to Watson that it was unlikely that the number 2 refered to the chapter, if their assumption that the first number refered to the page, the thing about the first chapter is a joke.

But since Holmes would have already come to a conclusion about the class of book the cipher was based on, he would know already how likely it would be for the first two chapters of that class of book to consist of more than 500 pages, and he probably had some knowledge about the common distribution of the lengths of the first two chapters of books in his reality.

Of course, quite a bit of the logic behind the deductions of Sherlock Holmes is a bit obscure, we rarely have full access to the required data to reconstruct the chains of deductions and inferences S.H. uses unless he explains them.

Well, thought this was an interesting find. (http://news.yahoo.com/s/afp/20081222/od_afp/finlandcrimeoffbeat;_ylt=Ag9SOFKSGK6KoZbKr52sb_4DW 7oF)

Apparently, police in Finland are attempting to link a car theif to the crime via DNA found in a mosquito that was found in the stolen vehicle.

Personally, even if you know the guy did it, the mosquito seems circumstancial at best (particular given that the theifs alibi was that he was hitch-hiking and the owner gave him a ride).

This is the line that really struck me:



...For a car theft! There's no mention of murder or violence of any sort. That's dedication ... I can't imagine having a DNA test run to prosecute a car theif here in the states.

What does this have to do with Sherlock Holmes?:confused: I was expecting some discussion about one of his cases or something...

Anyway, as for the mosquito DNA evidence, the prosecution should have to show that the blood in the mosquito that of the person commiting the crime. The presence of DNA evidence on a scene(barring evidence manipulation by the investigators) is only proof of the presence of the person that DNA belong to within a spesific time frame, or the presence of some other person with access to the medium carrying the DNA(that is, it may be planted or deposed by some creature that previously came into contact with a sample).

In this case, it is likely that the DNA came from the person it belongs to, by being sucked up by the mosquito, but since the person admitadly have been on the scene of the crime at a time close to the commision of the crime, the evidence is of limited value, unless they can specify the actual time the mosquito aquired the blood and was deposited on the scene.

It is likely that the prosecution would use this piece of evidence as corroberation to some other evidence, but it is rather worthless unless the time of deposition can be shown to be within the time window of the crime with a high accuracy, but it may be introduced in an attempt to manipulate the people involved in judging the case to think that since the scene was investigated so thoroughly that this was found, the rest of the case must have been conducted with equal thoroughness.

It may be that some people have some missconceptions of the evidential value of DNA. Kind of how fingerprints was represented in the dr. Thorndyke story The red thumbmark, where the statement that fingerprints afford identification that needs no corroberation is taken by the police to mean it is also evidence of guilt when a thumbmark is found on the scene of a crime.

[ParaDoctor]

how likely it would be for the first two chapters of that class of book to consist of more than 500 pages, and he probably had some knowledge about the common distribution of the lengths of the first two chapters of books in his reality.As I said, statistical inference, not deductive logic.

The presence of DNA evidence on a scene(barring evidence manipulation by the investigators) is only proof of the presence of the person that DNA belong to within a spesific time frame, or the presence of some other person with access to the medium carrying the DNA(that is, it may be planted or deposed by some creature that previously came into contact with a sample).
To deductively prove guilt, you'd have to rule out the logical possibility of the defendant's innocence. That's not possible, so courts merely establish guilt "beyond reasonable doubt", if you're lucky. One thing that can never be stressed enough is that "deductive proof" is an extremely stringent requirement. If you're serious about it, then almost no scientific paper presents a valid proof, not even those published in the mathematical journals. Not necessarily because they're wrong, just because they imply some context, rendering the actual argument incomplete, hence invalid. Don't believe me, just recall that Wiles had to retract the first version of his proof of Fermat's last theorem.

[hhEb09'1]

If you're serious about it, then almost no scientific paper presents a valid proof, not even those published in the mathematical journals. Not necessarily because they're wrong, just because they imply some context, rendering the actual argument incomplete, hence invalid. Don't believe me, just recall that Wiles had to retract the first version of his proof of Fermat's last theorem.Wiles retracted the first version because someone had found an error, necessarily because it was wrong. :)

[ParaDoctor]

Wiles retracted the first version because someone had found an error, necessarily because it was wrong. :)Both Faltings (http://www.ams.org/notices/199507/faltings.pdf) and Lopez-Ortiz (http://www.cs.uwaterloo.ca/~alopez-o/math-faq/mathtext/node9.html) talk about a "gap", which is a logical error. So my example stinks. :D
Still, the only "real" proofs are formal ones. A solid scientific paper can be transformed into a valid formal proof, while anything less will disintegrate into a mess of assertions not connected by anything but wishful thinking. It never ceases to amaze me that today's scientific journals still accept papers not accompanied by a formal version. :rolleyes:

[hhEb09'1]

Still, the only "real" proofs are formal ones. A solid scientific paper can be transformed into a valid formal proof, while anything less will disintegrate into a mess of assertions not connected by anything but wishful thinking. It never ceases to amaze me that today's scientific journals still accept papers not accompanied by a formal version.Are you still talking about math journals?

[TrAI]

As I said, statistical inference, not deductive logic.

Well, you said his deductive logic stinks, and used this as an example, but if this is an example of statistical inference under philosophical logics, it seems to me that it wouldn't be a good example of bad deductions. I am sure you could find something better to use as an example, though...

Of course, it may be that you were trying to object to the use of the word deductions about these kinds of reasoning, but that is a different issue than the quality of any deductions of S.H.

Anyway, the word deduction, as used in detective fiction is quite commonly a blanket term for any sort of reasoning from a set of knowns or possible interpretations of known data to conclusions of a sufficient probability for the detective to get on with his or her work. This is not really a wrong usage of the word, as it is rather common for words to have different meanings in different fields or situations. And since it is all leading from some known information, I suppose it could be concidered etymologicaly correct to use the word like that.

To deductively prove guilt, you'd have to rule out the logical possibility of the defendant's innocence. That's not possible, so courts merely establish guilt "beyond reasonable doubt", if you're lucky. One thing that can never be stressed enough is that "deductive proof" is an extremely stringent requirement. If you're serious about it, then almost no scientific paper presents a valid proof, not even those published in the mathematical journals. Not necessarily because they're wrong, just because they imply some context, rendering the actual argument incomplete, hence invalid. Don't believe me, just recall that Wiles had to retract the first version of his proof of Fermat's last theorem.

Hmmm. I was just saying that finding someone's DNA at a site only shows that the DNA has somehow gotten deposited there, it isn't by itself evidence of the guilt of this person, and if the person is known to have been at the location at a time close to the crime, the DNA is not very useful, unless one can accuratly establish the timeframe of the deposition to be within that of the crime.

[lek]

I don't know how this CSI stuff works in real world, but I'd find it rather odd if blood from an insect was the only evidence placing the person in the car. (he did admit beeing in the car, but said he was just a hitchhiker). If the police had his dna reference, then they would most likely have fingerprints as well.
<csi mode> hair from the seats would reveal if he was driving or a passenger, and fingerprints on driving wheel etc </csi mode>

Sounds like attention seeking really :I

[ParaDoctor]

Are you still talking about math journals?Absolutely. Ten seconds of browsing the very first paper randomly picked up from the math section (http://www.doaj.org/doaj?func=subject&cpid=58) of DOAJ.org (http://www.doaj.org/):
Theory and Applications of Categories volume 21 page 79 (http://www.tac.mta.ca/tac/volumes/21/5/21-05.pdf): Proof of Lemma 3.4: "Trivial."
Peachy. :rolleyes:

More?

Scientia Magna volume 2 page 40 (http://www.gallup.unm.edu/~smarandache/ScientiaMagna2no1.pdf): "the function" ... "has a simple pole point at s=1"

DOI 10.2174/1874117700801010035 page 58 (http://dx.doi.org/10.2174/1874117700801010035): "Lemma 4 and Lemma 3 below imply that u is a constant function and lambda=1."

ISSN 1224-1784 volume 13 fascicle 1 page 83 (http://www.anstuocmath.ro/mathematics/pdf9/79_88_GGoga.pdf): "Set Y =" ... "and define" ... "Then Y is a complete metric space"

All these are presented without even mentioning anything proof-like. I have little doubt that these assertions will turn out correct under scrutiny, but that doesn't change the fact that these papers, in the form they are presented, do not constitute valid proofs.

In general, occurences of "obviously", "trivially", "clearly" constitute red flags. Heck, watch out for any paper that does not contain a formal proof, explicitly stating the formal system employed.

(I think we're drifting in the direction of General Science.)

[ParaDoctor]

Well, you said his deductive logic stinks, and used this as an example, but if this is an example of statistical inference under philosophical logics, it seems to me that it wouldn't be a good example of bad deductions. I am sure you could find something better to use as an example, though...Your interpretation of the quoted passage would make it a statistical inference. But as stated, the argument is deductive:

"if" ... [then] ... "must" ...

Replace "must" with "is extremely likely to", and you have a statistical inference.

Of course, it may be that you were trying to object to the use of the word deductions about these kinds of reasoning, but that is a different issue than the quality of any deductions of S.H.Talking about the "quality" of Holmes' deductions implies that they are deductions, which is the very point being argued. ;)

Anyway, the word deduction, as used in detective fiction is quite commonly a blanket term for any sort of reasoning from a set of knowns or possible interpretations of known data to conclusions of a sufficient probability for the detective to get on with his or her work. This is not really a wrong usage of the word, as it is rather common for words to have different meanings in different fields or situations. And since it is all leading from some known information, I suppose it could be concidered etymologicaly correct to use the word like that.That Doyle is not alone in abusing the word "deduction" does not exactly exonerate him.

I was just saying that finding someone's DNA at a site only shows that the DNA has somehow gotten deposited thereThen again, the DNA might have mutated randomly. Improbable, but not ruled out by physics as we know it. So, from a logical point of view, DNA tells us nothing about the defendant's behavior.

[hhEb09'1]

In general, occurences of "obviously", "trivially", "clearly" constitute red flags. And, Duh (http://www.bautforum.com/questions-answers/74122-will-moon-become-planet.html#post1243354)! :)Heck, watch out for any paper that does not contain a formal proof, explicitly stating the formal system employed.
The mind boggles.

[ParaDoctor]

And, Duh (http://www.bautforum.com/questions-answers/74122-will-moon-become-planet.html#post1243354)! :)I was thinking more along the lines of unearthing rectifiable corner cuts, but it's nice to know I'm not the only one in possession of personal experience about the lack of causal relation being being sure and being right. :D

The mind boggles.It should. A typical research paper will contain anywhere between tens and thousands of these apparently harmless "timesavers". Multiply that with the hundreds of thousands of papers published each year. Note that we're talking about the best of the best here. Thinking about how much waste of time, space and money is due to totally inadequate use of logic gives me a headache. :doh:

[hhEb09'1]

It should. A typical research paper will contain anywhere between tens and thousands of these apparently harmless "timesavers". Multiply that with the hundreds of thousands of papers published each year. Note that we're talking about the best of the best here. Thinking about how much waste of time, space and money is due to totally inadequate use of logic gives me a headache. :doh:My reaction was just the opposite. What I meant was, if every paper were to be completely "fleshed out", we'd probably have an increase in total word count of 10,000%.

I don't really see a problem with such shortcuts. Sure, they could be a problem, but that's what peer-review is for--to catch them. If you had to "justify" each step, you couldn't build a house.

[ParaDoctor]

My reaction was just the opposite.So I gathered. ;)

What I meant was, if every paper were to be completely "fleshed out", we'd probably have an increase in total word count of 10,000%.That's why logicians developed a convenient notation, just as the mathematicians did. BTW, how did you arrive at the 10000% estimate?

I don't really see a problem with such shortcuts. Sure, they could be a problem, but that's what peer-review is for--to catch them.No need to delve into the debate about the cons and pros of peer review. To be precise, no need for peer review when you can use a verifier for reviewing formal proofs.

If you had to "justify" each step, you couldn't build a house.:confused: So what are architects and engineers doing all day?

Look at it this way: While a formal argument is more work, the final product is that much more reliable, and quality control is a cinch.

[hhEb09'1]

BTW, how did you arrive at the 10000% estimate?An intentional exaggeration, but I have no argument that the number is not even greater. :)
No need to delve into the debate about the cons and pros of peer review.No, I meant that peer review would (sometimes!) catch those pesky "obviously" obfuscations. To be precise, no need for peer review when you can use a verifier for reviewing formal proofs.Are you meaning, machine verification? Are you trying to get computers involved? So, there wouldn't be any drudgery of reading that 10,000%?
:confused: So what are architects and engineers doing all day?Well, that's what I meant. When a carpenter builds a house to engineering and design specifications, they don't require the formal proof in front of them, just the design. They sometimes depend upon the experience of others that the whole thing is going to work--and they have their own experiences to augment that. But, and here's the big but, engineers and architects do the same thing. It happens at all levels. And computers? that's what OOP (object oriented programming) is all about.

[ParaDoctor]

An intentional exaggeration, but I have no argument that the number is not even greater. :)<cruel>If it's an exaggeration, you started out from a realistic estimate. How did you arrive at it?</cruel>

No, I meant that peer review would (sometimes!) catch those pesky "obviously" obfuscations.And misses all the unspoken assumptions?

Are you meaning, machine verification? Are you trying to get computers involved? So, there wouldn't be any drudgery of reading that 10,000%?You sound surprised. ;)

Well, that's what I meant. When a carpenter builds a house to engineering and design specifications, they don't require the formal proof in front of them, just the design.That's the whole point of a proof, that you can safely rely on the conclusion whenever the preconditions are satisfied.

But, and here's the big but, engineers and architects do the same thing. It happens at all levels.Nope, they produce a theorem to the effect that the design implements the specification, including all relevant laws and regulations. The proof is rather implicit, consisting in the diplomas of the engineers and the reports and certificates produced during planning. Of course they use work of others, but they are constructing proofs of their own. Carpenters just use the resultant theorems.