[Genuine Realist]

I note today that the police reform bill passed by the House bans the use of facial recognition software. Good Lord. Facial recognition software, like every other kind of technology, EXCLUDES - it doesn't include. In combination with other technology, it makes false arrest and inconvenience to the uninvolved LESS, not more, likely - and this is obvious. To the extent that it is crude now, it will be refined at exponential speed in the months and years to come. And this too is obvious.

So how does this relate to Covid-19? When I excluded South Korea from comparison to the US in Covid-19 reaction, one of the reasons is because the uninhibited use of advanced technology - contact tracing through credit card tracking, cctv, GPS patterns, and the like - are absolutely impossible in the current legal and social climate of the US. Worries about 1984 and the like inhibit us. But while we are lamenting the lack of contact tracers and so on, we should remember we are ignoring a vast quantum of information ready at hamd.

That reluctance is absolute nonsense. Information technology is neutral - no matter what a society wants to do, it will do it faster and more effectively the more information it has at its disposal. If it wants to determine which of its citizens is exposed to a dangerous virus, it will do that better. If it wants to determine which of its citizens may have distant Jewish or African ancestry, it can do that as well. It's not the technology that makes a society racist. It's the social values.

Right now, you could use even crude facial recognition technology to identify people in a large crowd who are not wearing masks. (Ironically, mask wearing would make the technology less useful.) You could send them a letter or email politely reminding them that mask wearing is required, etc., etc. And if there is an ordinance, statute, or regulation that criminalizes non-mask wearing? I don't think we extend the use of the technology to issuing an actual citation and fine, because it isn't refined enough to provide probable cause. That's the sort of legal system we have.

But we don't discard the tool, either. It's useful, and it is going to get more useful. We can't march ignore the progress of information technology, as much as some would like to.

[Genuine Realist]

I note today that the police reform bill passed by the House bans the use of facial recognition software. Good Lord. Facial recognition software, like every other kind of technology, EXCLUDES - it doesn't include. In combination with other technology, it makes false arrest and inconvenience to the uninvolved LESS, not more, likely - and this is obvious. To the extent that it is crude now, it will be refined at exponential speed in the months and years to come. And this too is obvious.

So how does this relate to Covid-19? When I excluded South Korea from comparison to the US in Covid-19 reaction, one of the reasons is because the uninhibited use of advanced technology - contact tracing through credit card tracking, cctv, GPS patterns, and the like - are absolutely impossible in the current legal and social climate of the US. Worries about 1984 and the like inhibit us. But while we are lamenting the lack of contact tracers and so on, we should remember we are ignoring a vast quantum of information ready at hamd.

That reluctance is absolute nonsense. Information technology is neutral - no matter what a society wants to do, it will do it faster and more effectively the more information it has at its disposal. If it wants to determine which of its citizens is exposed to a dangerous virus, it will do that better. If it wants to determine which of its citizens may have distant Jewish or African ancestry, it can do that as well. It's not the technology that makes a society racist. It's the social values.

Right now, you could use even crude facial recognition technology to identify people in a large crowd who are not wearing masks. (Ironically, mask wearing would make the technology less useful.) You could send them a letter or email politely reminding them that mask wearing is required, etc., etc. And if there is an ordinance, statute, or regulation that criminalizes non-mask wearing? I don't think we extend the use of the technology to issuing an actual citation and fine, because it isn't refined enough to provide probable cause. That's the sort of legal system we have.

But we don't discard the tool, either. It's useful, and it is going to get more useful. We can't march ignore the progress of information technology, as much as some would like to.

[JustAnotherFan]

Information technology is neutral - no matter what a society wants to do, it will do it faster and more effectively the more information it has at its disposal. If it wants to determine which of its citizens is exposed to a dangerous virus, it will do that better. If it wants to determine which of its citizens may have distant Jewish or African ancestry, it can do that as well. It's not the technology that makes a society racist. It's the social values.

1. Your argument seems to suggest that you think the government is more interested in using technology to stop the pandemic than it is in controlling (policing) Black and brown and poor communities. I disagree.

2. Information technology is not neutral. It can have bias built into it. https://www.npr.org/2020/06/24/882678392...on-mistake

I say let's give everyone masks and stop inviting strangers to congregate indoors.

[JustAnotherFan]

Information technology is neutral - no matter what a society wants to do, it will do it faster and more effectively the more information it has at its disposal. If it wants to determine which of its citizens is exposed to a dangerous virus, it will do that better. If it wants to determine which of its citizens may have distant Jewish or African ancestry, it can do that as well. It's not the technology that makes a society racist. It's the social values.

1. Your argument seems to suggest that you think the government is more interested in using technology to stop the pandemic than it is in controlling (policing) Black and brown and poor communities. I disagree.

2. Information technology is not neutral. It can have bias built into it. https://www.npr.org/2020/06/24/882678392...on-mistake

I say let's give everyone masks and stop inviting strangers to congregate indoors.

[Genuine Realist]

Information technology is neutral - no matter what a society wants to do, it will do it faster and more effectively the more information it has at its disposal. If it wants to determine which of its citizens is exposed to a dangerous virus, it will do that better. If it wants to determine which of its citizens may have distant Jewish or African ancestry, it can do that as well. It's not the technology that makes a society racist. It's the social values.

1. Your argument seems to suggest that you think the government is more interested in using technology to stop the pandemic than it is in controlling (policing) Black and brown and poor communities. I disagree.

2. Information technology is not neutral. It can have bias built into it. https://www.npr.org/2020/06/24/882678392...on-mistake

I say let's give everyone masks and stop inviting strangers to congregate indoors.

It is hardly surprising that facial recognition software would be subject at this stage in its development. The same thing happened with DNA databases early on. That does not alter the fact that it works to narrow the focus of investigations, thereby tending to exclude the innocent.

The balance of your post is political and I will not answer it here. If you want to discuss the realities of urban policing, post a topic on Thunderdome or the revived CEB. http://currenteventsboard.com/.

[Genuine Realist]

Information technology is neutral - no matter what a society wants to do, it will do it faster and more effectively the more information it has at its disposal. If it wants to determine which of its citizens is exposed to a dangerous virus, it will do that better. If it wants to determine which of its citizens may have distant Jewish or African ancestry, it can do that as well. It's not the technology that makes a society racist. It's the social values.

1. Your argument seems to suggest that you think the government is more interested in using technology to stop the pandemic than it is in controlling (policing) Black and brown and poor communities. I disagree.

2. Information technology is not neutral. It can have bias built into it. https://www.npr.org/2020/06/24/882678392...on-mistake

I say let's give everyone masks and stop inviting strangers to congregate indoors.

It is hardly surprising that facial recognition software would be subject at this stage in its development. The same thing happened with DNA databases early on. That does not alter the fact that it works to narrow the focus of investigations, thereby tending to exclude the innocent.

The balance of your post is political and I will not answer it here. If you want to discuss the realities of urban policing, post a topic on Thunderdome or the revived CEB. http://currenteventsboard.com/.

[fullmetal]

GR, it sounds like you haven't spent a lot of time working with facial recognition algorithms.  There are aspects of these algorithms that are highly accurate, and yet they are still capable of producing errors.  I don't think the comparison to DNA technology is apt, but I'll think on that a bit tomorrow.

[fullmetal]

GR, it sounds like you haven't spent a lot of time working with facial recognition algorithms.  There are aspects of these algorithms that are highly accurate, and yet they are still capable of producing errors.  I don't think the comparison to DNA technology is apt, but I'll think on that a bit tomorrow.

[Genuine Realist]

GR, it sounds like you haven't spent a lot of time working with facial recognition algorithms.  There are aspects of these algorithms that are highly accurate, and yet they are still capable of producing errors.  I don't think the comparison to DNA technology is apt, but I'll think on that a bit tomorrow.

One basic truism of hi tech is that any given day is day one. Facial recognition software is sophisticated enough to be useful today, but it is nothing like the sophistication it is going to have in a year, let alone five years.

Do you think facial recognition software sophistication will ever reach a point where it would be sufficient by itself to convict? Ask yourself, whether you - sitting as a juror - would ever accept the determination of a computer program over your own visual judgment. It's so far beyond the realm of possibility now as to be laughable. What you'd use it for is to identify a person of interest and then investigate, to see if you can build a case against him.  If the case is there, the circumstantial evidence will be, too, that being the nature of reality. 

It might surprise you to know that no one goes to trial on DNA or fingerprints alone. It's very powerful evidence, of course, but investigators ALWAYS look for circumstances that tie the defendant to the crime. Making judgments on other human beings is the task of human beings, not biological or technical tools. If society ever devolves (sic) to the point where we delegate to automated processes, the fault will be our societys', and not the tools.

I can think of three beneficial applications for FRS right now, the third of which bears on Covid-19 and brings us back to topic:

(a) With a sufficiently sophisticated program and a robust database, you could minimize the intrusive pat downs at airports. The computer scans you, compares your photo to a database of millions of drivers' license and booking photos, and determines that you do not - or in rare cases, do - fall into some sort of threatening category. Whatever ethnic bias might exist at the start becomes more and more minimal as the program develops greater sophistication. Increased information ALWAYS reduces bias - a truism the ACLU has yet to understand.

(b) The primary goal of inner city law enforcement fo 50 years has been to stop gangbangers killing each other. That's why the NYPD and others went to stop-and-frisk 30 years ago. It worked. Homicide rates dropped dramatically. But it came at the cost of embittering any number of Black males who suffered an unnecessary intrusion. With FRS, a cop could take a phone pic of someone loitering with possibly criminal intent, run it through a database, and find out whether the kid is a gangbanger possibly up to no good, or a college kid slumming. You also eliminate randomness - a cop doing a stop-and-frisk is able to tell the kid why. And finally, when you do have someone with a racial bias, it's going to show up in the pattern of his or her inquiries, and you can get them counseling or put them on an assignment where it doesn't matter so much.

(c ) Finally, getting to Covid-19, and contact tracing, you could ask the infected patient about recent public visits. With places that have digital surveillance - just about everyone - you can feed the patient's picture into the computer, run the digital stream past the id pic, and have the FCS identify the patient at the times and place where he or she actually was. You can then pick out the pics from the stream of persons nearby who may have been infected, run the process in reverse to ID them, and contact them to alert them of their situation.

(This scares you? You think that having the patient reconstruct his movements, laboriously ID persons in the vicinity, and then use shoe leather to contact them is any LESS intrusive? It's just less efficient.)

Increased information ALWAYS facilitates whatever social goals or tasks a society wants to undertake. The tool by itself is neutral. If you're worrried society will use the tool to do harmful things, address the social values. Don't limit the tools.

[Genuine Realist]

GR, it sounds like you haven't spent a lot of time working with facial recognition algorithms.  There are aspects of these algorithms that are highly accurate, and yet they are still capable of producing errors.  I don't think the comparison to DNA technology is apt, but I'll think on that a bit tomorrow.

One basic truism of hi tech is that any given day is day one. Facial recognition software is sophisticated enough to be useful today, but it is nothing like the sophistication it is going to have in a year, let alone five years.

Do you think facial recognition software sophistication will ever reach a point where it would be sufficient by itself to convict? Ask yourself, whether you - sitting as a juror - would ever accept the determination of a computer program over your own visual judgment. It's so far beyond the realm of possibility now as to be laughable. What you'd use it for is to identify a person of interest and then investigate, to see if you can build a case against him.  If the case is there, the circumstantial evidence will be, too, that being the nature of reality. 

It might surprise you to know that no one goes to trial on DNA or fingerprints alone. It's very powerful evidence, of course, but investigators ALWAYS look for circumstances that tie the defendant to the crime. Making judgments on other human beings is the task of human beings, not biological or technical tools. If society ever devolves (sic) to the point where we delegate to automated processes, the fault will be our societys', and not the tools.

I can think of three beneficial applications for FRS right now, the third of which bears on Covid-19 and brings us back to topic:

(a) With a sufficiently sophisticated program and a robust database, you could minimize the intrusive pat downs at airports. The computer scans you, compares your photo to a database of millions of drivers' license and booking photos, and determines that you do not - or in rare cases, do - fall into some sort of threatening category. Whatever ethnic bias might exist at the start becomes more and more minimal as the program develops greater sophistication. Increased information ALWAYS reduces bias - a truism the ACLU has yet to understand.

(b) The primary goal of inner city law enforcement fo 50 years has been to stop gangbangers killing each other. That's why the NYPD and others went to stop-and-frisk 30 years ago. It worked. Homicide rates dropped dramatically. But it came at the cost of embittering any number of Black males who suffered an unnecessary intrusion. With FRS, a cop could take a phone pic of someone loitering with possibly criminal intent, run it through a database, and find out whether the kid is a gangbanger possibly up to no good, or a college kid slumming. You also eliminate randomness - a cop doing a stop-and-frisk is able to tell the kid why. And finally, when you do have someone with a racial bias, it's going to show up in the pattern of his or her inquiries, and you can get them counseling or put them on an assignment where it doesn't matter so much.

(c ) Finally, getting to Covid-19, and contact tracing, you could ask the infected patient about recent public visits. With places that have digital surveillance - just about everyone - you can feed the patient's picture into the computer, run the digital stream past the id pic, and have the FCS identify the patient at the times and place where he or she actually was. You can then pick out the pics from the stream of persons nearby who may have been infected, run the process in reverse to ID them, and contact them to alert them of their situation.

(This scares you? You think that having the patient reconstruct his movements, laboriously ID persons in the vicinity, and then use shoe leather to contact them is any LESS intrusive? It's just less efficient.)

Increased information ALWAYS facilitates whatever social goals or tasks a society wants to undertake. The tool by itself is neutral. If you're worrried society will use the tool to do harmful things, address the social values. Don't limit the tools.

[2006alum]

It might surprise you to know that no one goes to trial on DNA or fingerprints alone. 

If so, I'd be curious for you to explain what this law professor at UC Berkeley got wrong, from "Safety in Numbers - Deciding When DNA Alone is Enough to Convict": 

Cases like Puckett are part of an emerging phenomenon of "pure cold hit" DNA prosecutions in which the entirety of the government's case against the suspect, aside from his prior conviction, is a DNA profile match or a match accompanied only by general evidence, such as age or race, that limits the suspect population only to a small and quantifiable degree. These cases have been made possible by law enforcement's increased reliance on DNA databases to investigate crimes, the newfound ability to develop DNA profiles on old, degraded, and mixed evidence samples, and the modern practice of securing so-called "John Doe" indictments against DNA profiles of yet-unknown suspects."

Technically speaking, in most cases yes, there is some other circumstantial evidence. But it's really disingenuous to suggest that DNA tests aren't being used by prosecutors as silver bullets when they lack any other concrete evidence, and there's so many documented problems with DNA evidence I don't even know where to begin. 

This finding documented in the Atlantic was particularly striking:

As recognition of DNA’s revelatory power seeped into popular culture, courtroom experts started talking about a “CSI effect,” whereby juries, schooled by television police procedurals, needed only to hear those three magic letters—DNA—to arrive at a guilty verdict. In 2008, Donald E. Shelton, a felony trial judge in Michigan, published a study in which 1,027 randomly summoned jurors in the city of Ann Arbor were polled on what they expected prosecutors to present during a criminal trial. Three-quarters of the jurors said they expected DNA evidence in rape cases, and nearly half said they expected it in murder or attempted-murder cases; 22 percent said they expected DNA evidence in every criminal case. Shelton quotes one district attorney as saying, “They expect us to have the most advanced technology possible, and they expect it to look like it does on television.”

[2006alum]

It might surprise you to know that no one goes to trial on DNA or fingerprints alone. 

If so, I'd be curious for you to explain what this law professor at UC Berkeley got wrong, from "Safety in Numbers - Deciding When DNA Alone is Enough to Convict": 

Cases like Puckett are part of an emerging phenomenon of "pure cold hit" DNA prosecutions in which the entirety of the government's case against the suspect, aside from his prior conviction, is a DNA profile match or a match accompanied only by general evidence, such as age or race, that limits the suspect population only to a small and quantifiable degree. These cases have been made possible by law enforcement's increased reliance on DNA databases to investigate crimes, the newfound ability to develop DNA profiles on old, degraded, and mixed evidence samples, and the modern practice of securing so-called "John Doe" indictments against DNA profiles of yet-unknown suspects."

Technically speaking, in most cases yes, there is some other circumstantial evidence. But it's really disingenuous to suggest that DNA tests aren't being used by prosecutors as silver bullets when they lack any other concrete evidence, and there's so many documented problems with DNA evidence I don't even know where to begin. 

This finding documented in the Atlantic was particularly striking:

As recognition of DNA’s revelatory power seeped into popular culture, courtroom experts started talking about a “CSI effect,” whereby juries, schooled by television police procedurals, needed only to hear those three magic letters—DNA—to arrive at a guilty verdict. In 2008, Donald E. Shelton, a felony trial judge in Michigan, published a study in which 1,027 randomly summoned jurors in the city of Ann Arbor were polled on what they expected prosecutors to present during a criminal trial. Three-quarters of the jurors said they expected DNA evidence in rape cases, and nearly half said they expected it in murder or attempted-murder cases; 22 percent said they expected DNA evidence in every criminal case. Shelton quotes one district attorney as saying, “They expect us to have the most advanced technology possible, and they expect it to look like it does on television.”

[fullmetal]

GR: your argument boils down to "errors are acceptable."

They are not.

https://www.npr.org/2020/06/24/882683463...-in-michig
https://www.wired.com/story/amazon-facia...forcement/
https://www.eff.org/pages/face-recognition
https://www.asisonline.org/security-mana...mographic/
https://www.bangkokpost.com/tech/1820554...ition-tech

[fullmetal]

GR: your argument boils down to "errors are acceptable."

They are not.

https://www.npr.org/2020/06/24/882683463...-in-michig
https://www.wired.com/story/amazon-facia...forcement/
https://www.eff.org/pages/face-recognition
https://www.asisonline.org/security-mana...mographic/
https://www.bangkokpost.com/tech/1820554...ition-tech

[Genuine Realist]

GR: your argument boils down to "errors are acceptable."

They are not.

https://www.npr.org/2020/06/24/882683463...-in-michig
https://www.wired.com/story/amazon-facia...forcement/
https://www.eff.org/pages/face-recognition
https://www.asisonline.org/security-mana...mographic/
https://www.bangkokpost.com/tech/1820554...ition-tech

Errors in any new software protocol are inevitable. Greater sophistication leads to more and reliability. You don't get there without use and refinement.

Voice recognition with various automated assistants (Siri, Google Assistant, Alexa) is now outstandingly reliable and getting better. It did not begin at that level of refinement. You can't reach that level of refinement without continued development.

You use the tool as a tool - not a solution. The problem in Michigan was undue reliance by law enforcement on the tool. That happens, no matter what tools you use. Continued education is the answer. 

To assume that the 2020 version of any software technology is definitive, that the problems and crudities with use can't be resolved and eliminated, is plain silly, and belied by every software category you can name - including the platform I"m writing on.

The articles you cite, by the way, are all polemical.

---

[quote="Genuine Realist" pid='284269' dateline='1593361397']
It might surprise you to know that no one goes to trial on DNA or fingerprints alone. 

If so, I'd be curious for you to explain what this law professor at UC Berkeley got wrong, from "Safety in Numbers - Deciding When DNA Alone is Enough to Convict": 

Cases like Puckett are part of an emerging phenomenon of "pure cold hit" DNA prosecutions in which the entirety of the government's case against the suspect, aside from his prior conviction, is a DNA profile match or a match accompanied only by general evidence, such as age or race, that limits the suspect population only to a small and quantifiable degree. These cases have been made possible by law enforcement's increased reliance on DNA databases to investigate crimes, the newfound ability to develop DNA profiles on old, degraded, and mixed evidence samples, and the modern practice of securing so-called "John Doe" indictments against DNA profiles of yet-unknown suspects."

Technically speaking, in most cases yes, there is some other circumstantial evidence. But it's really disingenuous to suggest that DNA tests aren't being used by prosecutors as silver bullets when they lack any other concrete evidence, and there's so many documented problems with DNA evidence I don't even know where to begin. 

This finding documented in the Atlantic was particularly striking:

As recognition of DNA’s revelatory power seeped into popular culture, courtroom experts started talking about a “CSI effect,” whereby juries, schooled by television police procedurals, needed only to hear those three magic letters—DNA—to arrive at a guilty verdict. In 2008, Donald E. Shelton, a felony trial judge in Michigan, published a study in which 1,027 randomly summoned jurors in the city of Ann Arbor were polled on what they expected prosecutors to present during a criminal trial. Three-quarters of the jurors said they expected DNA evidence in rape cases, and nearly half said they expected it in murder or attempted-murder cases; 22 percent said they expected DNA evidence in every criminal case. Shelton quotes one district attorney as saying, “They expect us to have the most advanced technology possible, and they expect it to look like it does on television.”

I think you may have wandered onto my turf, though I don't know what your professional background is.  In any case, the article you cite is apples and oranges.

What the teaching fellow (not professor) is discussing is whether DNA evidence all by itself is sufficient to convict. (I don't know whether the author is even aware of such a case. In the one case she describes, she has to concede that the DNA was the primary evidence - which of course it would be. But I"d like to know what the other evidence was.) That's an interesting theoretical question, to which no clear answer can be given - it depends on too many other factors. In the vast majority of cold cases, the initial cold hit doesn't lead to an immediate arrest, but to an investigation as to whether the subject can circumstantially be linked to the time and place of the crime. If it can, you almost certainly have PC for arrest, and a pretty good prima facie case, as DNA is powerful stuff.

But are you saying because a naive prosecutor and a naive jury may theoretically convict improprerly on that basis alone, that we should not be running DNA samples against databases? I doubt you mean that. In any case, you use DNA and fingerprints as tools, among others, to make or unmake a case. They are very valuable used properly. The fact that they can be used improperly doesn't change that.

FRS is different. It isn't going to be all that useful as proof at trial - the jury will make up its own mind about identification. I've never met a human being who would defer to a computer on that score. But it can be extremely useful in identifying a possible criminal perpetrator, and thereby eliminating the innocent. Look back at that stop-and-frisk situation I mentioned. FRS is not going to resolve social dilemma - preventing crime v. unwarranted intrusion - but it sure is going to  help.

In any case, you can't go backwards with this stuff. And its potential applications are extremely useful.

---

[quote="Genuine Realist" pid='284269' dateline='1593361397']
It might surprise you to know that no one goes to trial on DNA or fingerprints alone. 

If so, I'd be curious for you to explain what this law professor at UC Berkeley got wrong, from "Safety in Numbers - Deciding When DNA Alone is Enough to Convict": 

Cases like Puckett are part of an emerging phenomenon of "pure cold hit" DNA prosecutions in which the entirety of the government's case against the suspect, aside from his prior conviction, is a DNA profile match or a match accompanied only by general evidence, such as age or race, that limits the suspect population only to a small and quantifiable degree. These cases have been made possible by law enforcement's increased reliance on DNA databases to investigate crimes, the newfound ability to develop DNA profiles on old, degraded, and mixed evidence samples, and the modern practice of securing so-called "John Doe" indictments against DNA profiles of yet-unknown suspects."

Technically speaking, in most cases yes, there is some other circumstantial evidence. But it's really disingenuous to suggest that DNA tests aren't being used by prosecutors as silver bullets when they lack any other concrete evidence, and there's so many documented problems with DNA evidence I don't even know where to begin. 

This finding documented in the Atlantic was particularly striking:

As recognition of DNA’s revelatory power seeped into popular culture, courtroom experts started talking about a “CSI effect,” whereby juries, schooled by television police procedurals, needed only to hear those three magic letters—DNA—to arrive at a guilty verdict. In 2008, Donald E. Shelton, a felony trial judge in Michigan, published a study in which 1,027 randomly summoned jurors in the city of Ann Arbor were polled on what they expected prosecutors to present during a criminal trial. Three-quarters of the jurors said they expected DNA evidence in rape cases, and nearly half said they expected it in murder or attempted-murder cases; 22 percent said they expected DNA evidence in every criminal case. Shelton quotes one district attorney as saying, “They expect us to have the most advanced technology possible, and they expect it to look like it does on television.”

I think you may have wandered onto my turf, though I don't know what your professional background is.  In any case, the article you cite is apples and oranges.

What the teaching fellow (not professor) is discussing is whether DNA evidence all by itself is sufficient to convict. (I don't know whether the author is even aware of such a case. In the one case she describes, she has to concede that the DNA was the primary evidence - which of course it would be. But I"d like to know what the other evidence was.) That's an interesting theoretical question, to which no clear answer can be given - it depends on too many other factors. In the vast majority of cold cases, the initial cold hit doesn't lead to an immediate arrest, but to an investigation as to whether the subject can circumstantially be linked to the time and place of the crime. If it can, you almost certainly have PC for arrest, and a pretty good prima facie case, as DNA is powerful stuff.

But are you saying because a naive prosecutor and a naive jury may theoretically convict improprerly on that basis alone, that we should not be running DNA samples against databases? I doubt you mean that. In any case, you use DNA and fingerprints as tools, among others, to make or unmake a case. They are very valuable used properly. The fact that they can be used improperly doesn't change that.

FRS is different. It isn't going to be all that useful as proof at trial - the jury will make up its own mind about identification. I've never met a human being who would defer to a computer on that score. But it can be extremely useful in identifying a possible criminal perpetrator, and thereby eliminating the innocent. Look back at that stop-and-frisk situation I mentioned. FRS is not going to resolve social dilemma - preventing crime v. unwarranted intrusion - but it sure is going to  help.

In any case, you can't go backwards with this stuff. And its potential applications are extremely useful.

---

GR: your argument boils down to "errors are acceptable."

They are not.

https://www.npr.org/2020/06/24/882683463...-in-michig
https://www.wired.com/story/amazon-facia...forcement/
https://www.eff.org/pages/face-recognition
https://www.asisonline.org/security-mana...mographic/
https://www.bangkokpost.com/tech/1820554...ition-tech

Errors are unacceptable? Isn't that straight out of the anti-vaxxer playbook?

 in any new software protocol errors are inevitable.  You rely on common sense (sorely lacking in the Michigan anecdote) to mitigate and minimize. 
Greater sophistication leads to more and reliability. You don't get there without use and refinement.

Voice recognition with various automated assistants (Siri, Google Assistant, Alexa) is now outstandingly reliable and getting better. It did not begin at that level of refinement. You can't reach that level of refinement without continued development.

You use the tool as a tool - not a solution. The problem in Michigan was undue reliance by law enforcement on the tool. That happens, no matter what tools you use. Continued education is the answer. 

To assume that the 2020 version of any software technology is definitive, that the problems and crudities with use can't be resolved and eliminated, is plain silly, and belied by every software category you can name - including the platform I"m writing on.

The articles you cite, by the way, are all polemical.

[Genuine Realist]

GR: your argument boils down to "errors are acceptable."

They are not.

https://www.npr.org/2020/06/24/882683463...-in-michig
https://www.wired.com/story/amazon-facia...forcement/
https://www.eff.org/pages/face-recognition
https://www.asisonline.org/security-mana...mographic/
https://www.bangkokpost.com/tech/1820554...ition-tech

Errors in any new software protocol are inevitable. Greater sophistication leads to more and reliability. You don't get there without use and refinement.

Voice recognition with various automated assistants (Siri, Google Assistant, Alexa) is now outstandingly reliable and getting better. It did not begin at that level of refinement. You can't reach that level of refinement without continued development.

You use the tool as a tool - not a solution. The problem in Michigan was undue reliance by law enforcement on the tool. That happens, no matter what tools you use. Continued education is the answer. 

To assume that the 2020 version of any software technology is definitive, that the problems and crudities with use can't be resolved and eliminated, is plain silly, and belied by every software category you can name - including the platform I"m writing on.

The articles you cite, by the way, are all polemical.

---

[quote="Genuine Realist" pid='284269' dateline='1593361397']
It might surprise you to know that no one goes to trial on DNA or fingerprints alone. 

If so, I'd be curious for you to explain what this law professor at UC Berkeley got wrong, from "Safety in Numbers - Deciding When DNA Alone is Enough to Convict": 

Cases like Puckett are part of an emerging phenomenon of "pure cold hit" DNA prosecutions in which the entirety of the government's case against the suspect, aside from his prior conviction, is a DNA profile match or a match accompanied only by general evidence, such as age or race, that limits the suspect population only to a small and quantifiable degree. These cases have been made possible by law enforcement's increased reliance on DNA databases to investigate crimes, the newfound ability to develop DNA profiles on old, degraded, and mixed evidence samples, and the modern practice of securing so-called "John Doe" indictments against DNA profiles of yet-unknown suspects."

Technically speaking, in most cases yes, there is some other circumstantial evidence. But it's really disingenuous to suggest that DNA tests aren't being used by prosecutors as silver bullets when they lack any other concrete evidence, and there's so many documented problems with DNA evidence I don't even know where to begin. 

This finding documented in the Atlantic was particularly striking:

As recognition of DNA’s revelatory power seeped into popular culture, courtroom experts started talking about a “CSI effect,” whereby juries, schooled by television police procedurals, needed only to hear those three magic letters—DNA—to arrive at a guilty verdict. In 2008, Donald E. Shelton, a felony trial judge in Michigan, published a study in which 1,027 randomly summoned jurors in the city of Ann Arbor were polled on what they expected prosecutors to present during a criminal trial. Three-quarters of the jurors said they expected DNA evidence in rape cases, and nearly half said they expected it in murder or attempted-murder cases; 22 percent said they expected DNA evidence in every criminal case. Shelton quotes one district attorney as saying, “They expect us to have the most advanced technology possible, and they expect it to look like it does on television.”

I think you may have wandered onto my turf, though I don't know what your professional background is.  In any case, the article you cite is apples and oranges.

What the teaching fellow (not professor) is discussing is whether DNA evidence all by itself is sufficient to convict. (I don't know whether the author is even aware of such a case. In the one case she describes, she has to concede that the DNA was the primary evidence - which of course it would be. But I"d like to know what the other evidence was.) That's an interesting theoretical question, to which no clear answer can be given - it depends on too many other factors. In the vast majority of cold cases, the initial cold hit doesn't lead to an immediate arrest, but to an investigation as to whether the subject can circumstantially be linked to the time and place of the crime. If it can, you almost certainly have PC for arrest, and a pretty good prima facie case, as DNA is powerful stuff.

But are you saying because a naive prosecutor and a naive jury may theoretically convict improprerly on that basis alone, that we should not be running DNA samples against databases? I doubt you mean that. In any case, you use DNA and fingerprints as tools, among others, to make or unmake a case. They are very valuable used properly. The fact that they can be used improperly doesn't change that.

FRS is different. It isn't going to be all that useful as proof at trial - the jury will make up its own mind about identification. I've never met a human being who would defer to a computer on that score. But it can be extremely useful in identifying a possible criminal perpetrator, and thereby eliminating the innocent. Look back at that stop-and-frisk situation I mentioned. FRS is not going to resolve social dilemma - preventing crime v. unwarranted intrusion - but it sure is going to  help.

In any case, you can't go backwards with this stuff. And its potential applications are extremely useful.

---

[quote="Genuine Realist" pid='284269' dateline='1593361397']
It might surprise you to know that no one goes to trial on DNA or fingerprints alone. 

If so, I'd be curious for you to explain what this law professor at UC Berkeley got wrong, from "Safety in Numbers - Deciding When DNA Alone is Enough to Convict": 

Cases like Puckett are part of an emerging phenomenon of "pure cold hit" DNA prosecutions in which the entirety of the government's case against the suspect, aside from his prior conviction, is a DNA profile match or a match accompanied only by general evidence, such as age or race, that limits the suspect population only to a small and quantifiable degree. These cases have been made possible by law enforcement's increased reliance on DNA databases to investigate crimes, the newfound ability to develop DNA profiles on old, degraded, and mixed evidence samples, and the modern practice of securing so-called "John Doe" indictments against DNA profiles of yet-unknown suspects."

Technically speaking, in most cases yes, there is some other circumstantial evidence. But it's really disingenuous to suggest that DNA tests aren't being used by prosecutors as silver bullets when they lack any other concrete evidence, and there's so many documented problems with DNA evidence I don't even know where to begin. 

This finding documented in the Atlantic was particularly striking:

As recognition of DNA’s revelatory power seeped into popular culture, courtroom experts started talking about a “CSI effect,” whereby juries, schooled by television police procedurals, needed only to hear those three magic letters—DNA—to arrive at a guilty verdict. In 2008, Donald E. Shelton, a felony trial judge in Michigan, published a study in which 1,027 randomly summoned jurors in the city of Ann Arbor were polled on what they expected prosecutors to present during a criminal trial. Three-quarters of the jurors said they expected DNA evidence in rape cases, and nearly half said they expected it in murder or attempted-murder cases; 22 percent said they expected DNA evidence in every criminal case. Shelton quotes one district attorney as saying, “They expect us to have the most advanced technology possible, and they expect it to look like it does on television.”

I think you may have wandered onto my turf, though I don't know what your professional background is.  In any case, the article you cite is apples and oranges.

What the teaching fellow (not professor) is discussing is whether DNA evidence all by itself is sufficient to convict. (I don't know whether the author is even aware of such a case. In the one case she describes, she has to concede that the DNA was the primary evidence - which of course it would be. But I"d like to know what the other evidence was.) That's an interesting theoretical question, to which no clear answer can be given - it depends on too many other factors. In the vast majority of cold cases, the initial cold hit doesn't lead to an immediate arrest, but to an investigation as to whether the subject can circumstantially be linked to the time and place of the crime. If it can, you almost certainly have PC for arrest, and a pretty good prima facie case, as DNA is powerful stuff.

But are you saying because a naive prosecutor and a naive jury may theoretically convict improprerly on that basis alone, that we should not be running DNA samples against databases? I doubt you mean that. In any case, you use DNA and fingerprints as tools, among others, to make or unmake a case. They are very valuable used properly. The fact that they can be used improperly doesn't change that.

FRS is different. It isn't going to be all that useful as proof at trial - the jury will make up its own mind about identification. I've never met a human being who would defer to a computer on that score. But it can be extremely useful in identifying a possible criminal perpetrator, and thereby eliminating the innocent. Look back at that stop-and-frisk situation I mentioned. FRS is not going to resolve social dilemma - preventing crime v. unwarranted intrusion - but it sure is going to  help.

In any case, you can't go backwards with this stuff. And its potential applications are extremely useful.

---

GR: your argument boils down to "errors are acceptable."

They are not.

https://www.npr.org/2020/06/24/882683463...-in-michig
https://www.wired.com/story/amazon-facia...forcement/
https://www.eff.org/pages/face-recognition
https://www.asisonline.org/security-mana...mographic/
https://www.bangkokpost.com/tech/1820554...ition-tech

Errors are unacceptable? Isn't that straight out of the anti-vaxxer playbook?

 in any new software protocol errors are inevitable.  You rely on common sense (sorely lacking in the Michigan anecdote) to mitigate and minimize. 
Greater sophistication leads to more and reliability. You don't get there without use and refinement.

Voice recognition with various automated assistants (Siri, Google Assistant, Alexa) is now outstandingly reliable and getting better. It did not begin at that level of refinement. You can't reach that level of refinement without continued development.

You use the tool as a tool - not a solution. The problem in Michigan was undue reliance by law enforcement on the tool. That happens, no matter what tools you use. Continued education is the answer. 

To assume that the 2020 version of any software technology is definitive, that the problems and crudities with use can't be resolved and eliminated, is plain silly, and belied by every software category you can name - including the platform I"m writing on.

The articles you cite, by the way, are all polemical.

[2006alum]

I think you may have wandered onto my turf, though I don't know what your professional background is.  In any case, the article you cite is apples and oranges.

You're right, you don't know my background. I don't generally care about credentialism in internet forums, as I think facts and arguments speak for themselves. But I can assure you, mine are solid in this arena, and I'm not just astroturfing down on your turf. 

Speaking of professional backgrounds and "turf"...

What the teaching fellow (not professor) is discussing is whether DNA evidence all by itself is sufficient to convict. 

I suggest you do your research - Andrea Roth is very much a (full) professor (at UC Berkeley), and quite a qualified one to speak about these topics:

Andrea Roth joined the Berkeley Law faculty in 2011, after 3 years as a Grey Fellow at Stanford and 9 years as a trial and appellate public defender in Washington, D.C. Her research focuses on the ways in which pedigreed concepts of criminal procedure and evidentiary law must be retheorized in an era of science-based prosecutions. Her articles include “‘Spit and Acquit’: Prosecutors as Surveillance Entrepreneurs,” 107 [i]Cal. L. Rev.[/i] — (forthcoming 2019); “Machine Testimony,” 126 [i]Yale L.J.[/i] 1972 (2017); “Trial by Machine,” 104 [i]Georgetown L.J.[/i] 1245 (2016); “The Uneasy Case for Marijuana as Chemical Impairment Under a Science-Based Jurisprudence of Dangerousness,” 103 [i]Cal. L. Rev.[/i] 841 (2015); “Defying DNA: Rethinking the Role of the Jury in an Age of Scientific Proof of Innocence,” 93 [i]Boston U. L. Rev.[/i] 1643 (2013); “[i]Maryland v. King[/i] and the Wonderful, Horrible DNA Revolution in Law Enforcement,” 11 [i]Ohio St. L. J.[/i] 295 (2013); and “Safety in Numbers?: Deciding When DNA Alone Is Enough To Convict,” 85 [i]N.Y.U. L. Rev.[/i] 1130 (2010). She is a member of the Legal Resource Committee of the National Institute of Standards and Technology’s Organization of Scientific Area Committees.

In 2019 Roth was one of four recipients of the campus-wide Distinguished Teaching Award. In 2017, she received the campus-wide Prytanean Faculty Award given to one pretenure woman faculty member. In 2016, she received the law school’s Rutter Award for Teaching Excellence. She has also received teaching awards from Women of Berkeley Law and the Berkeley Criminal Law Journal.

I know our culture is descending towards mob rule and anti-expertise sentiment, but I tend to listen to credentialed experts on these matters, and there are many well researched, careful critiques of DNA evidence, including from the above cited professor.

And yes, of course there are good uses of DNA forensic technology - has anyone denied that? But DNA evidence is just a tool, and like any tool, it can be misused and abused. Same as facial recognition technology. And in an era where basically everyone - left, right, and center - seems to distrust authorities, I think moving cautiously on that front is more than prudent and far from ludditish.

[2006alum]

I think you may have wandered onto my turf, though I don't know what your professional background is.  In any case, the article you cite is apples and oranges.

You're right, you don't know my background. I don't generally care about credentialism in internet forums, as I think facts and arguments speak for themselves. But I can assure you, mine are solid in this arena, and I'm not just astroturfing down on your turf. 

Speaking of professional backgrounds and "turf"...

What the teaching fellow (not professor) is discussing is whether DNA evidence all by itself is sufficient to convict. 

I suggest you do your research - Andrea Roth is very much a (full) professor (at UC Berkeley), and quite a qualified one to speak about these topics:

Andrea Roth joined the Berkeley Law faculty in 2011, after 3 years as a Grey Fellow at Stanford and 9 years as a trial and appellate public defender in Washington, D.C. Her research focuses on the ways in which pedigreed concepts of criminal procedure and evidentiary law must be retheorized in an era of science-based prosecutions. Her articles include “‘Spit and Acquit’: Prosecutors as Surveillance Entrepreneurs,” 107 [i]Cal. L. Rev.[/i] — (forthcoming 2019); “Machine Testimony,” 126 [i]Yale L.J.[/i] 1972 (2017); “Trial by Machine,” 104 [i]Georgetown L.J.[/i] 1245 (2016); “The Uneasy Case for Marijuana as Chemical Impairment Under a Science-Based Jurisprudence of Dangerousness,” 103 [i]Cal. L. Rev.[/i] 841 (2015); “Defying DNA: Rethinking the Role of the Jury in an Age of Scientific Proof of Innocence,” 93 [i]Boston U. L. Rev.[/i] 1643 (2013); “[i]Maryland v. King[/i] and the Wonderful, Horrible DNA Revolution in Law Enforcement,” 11 [i]Ohio St. L. J.[/i] 295 (2013); and “Safety in Numbers?: Deciding When DNA Alone Is Enough To Convict,” 85 [i]N.Y.U. L. Rev.[/i] 1130 (2010). She is a member of the Legal Resource Committee of the National Institute of Standards and Technology’s Organization of Scientific Area Committees.

In 2019 Roth was one of four recipients of the campus-wide Distinguished Teaching Award. In 2017, she received the campus-wide Prytanean Faculty Award given to one pretenure woman faculty member. In 2016, she received the law school’s Rutter Award for Teaching Excellence. She has also received teaching awards from Women of Berkeley Law and the Berkeley Criminal Law Journal.

I know our culture is descending towards mob rule and anti-expertise sentiment, but I tend to listen to credentialed experts on these matters, and there are many well researched, careful critiques of DNA evidence, including from the above cited professor.

And yes, of course there are good uses of DNA forensic technology - has anyone denied that? But DNA evidence is just a tool, and like any tool, it can be misused and abused. Same as facial recognition technology. And in an era where basically everyone - left, right, and center - seems to distrust authorities, I think moving cautiously on that front is more than prudent and far from ludditish.

[Genuine Realist]

GR: your argument boils down to "errors are acceptable."

They are not.

https://www.npr.org/2020/06/24/882683463...-in-michig
https://www.wired.com/story/amazon-facia...forcement/
https://www.eff.org/pages/face-recognition
https://www.asisonline.org/security-mana...mographic/
https://www.bangkokpost.com/tech/1820554...ition-tech

Errors are unacceptable? Isn't that straight out of the anti-vaxxer playbook?

 in any new software protocol errors are inevitable.  You rely on common sense (sorely lacking in the Michigan anecdote) to mitigate and minimize. 
Greater sophistication leads to more and reliability. You don't get there without use and refinement.

Voice recognition with various automated assistants (Siri, Google Assistant, Alexa) is now outstandingly reliable and getting better. It did not begin at that level of refinement. You can't reach that level of refinement without continued development.

You use the tool as a tool - not a solution. The problem in Michigan was undue reliance by law enforcement on the tool. That happens, no matter what tools you use. Continued education is the answer. 

To assume that the 2020 version of any software technology is definitive, that the problems and crudities with use can't be resolved and eliminated, is plain silly, and belied by every software category you can name - including the platform I"m writing on.

The articles you cite, by the way, are all polemical.

[Genuine Realist]

GR: your argument boils down to "errors are acceptable."

They are not.

https://www.npr.org/2020/06/24/882683463...-in-michig
https://www.wired.com/story/amazon-facia...forcement/
https://www.eff.org/pages/face-recognition
https://www.asisonline.org/security-mana...mographic/
https://www.bangkokpost.com/tech/1820554...ition-tech

Errors are unacceptable? Isn't that straight out of the anti-vaxxer playbook?

 in any new software protocol errors are inevitable.  You rely on common sense (sorely lacking in the Michigan anecdote) to mitigate and minimize. 
Greater sophistication leads to more and reliability. You don't get there without use and refinement.

Voice recognition with various automated assistants (Siri, Google Assistant, Alexa) is now outstandingly reliable and getting better. It did not begin at that level of refinement. You can't reach that level of refinement without continued development.

You use the tool as a tool - not a solution. The problem in Michigan was undue reliance by law enforcement on the tool. That happens, no matter what tools you use. Continued education is the answer. 

To assume that the 2020 version of any software technology is definitive, that the problems and crudities with use can't be resolved and eliminated, is plain silly, and belied by every software category you can name - including the platform I"m writing on.

The articles you cite, by the way, are all polemical.

[BostonCard]

I think only the first post in this thread refers to COVID-19.  Perhaps it’s time to invoke the Terry rule?

BC

[BostonCard]

I think only the first post in this thread refers to COVID-19.  Perhaps it’s time to invoke the Terry rule?

BC

[2006alum]

I think only the first post in this thread refers to COVID-19.  Perhaps it’s time to invoke the Terry rule?

BC

Good point! So closed.

[2006alum]

I think only the first post in this thread refers to COVID-19.  Perhaps it’s time to invoke the Terry rule?

BC

Good point! So closed.