TSA's bomb-sniffing dogs
#196
Join Date: Nov 2007
Location: WAS
Programs: enjoyed being warm spit for a few years on CO/UA but now nothing :(
Posts: 2,507
I have to disagree.
We have to compare results in order to understand which method produces more reliable results - and at what cost.
If, for example, costs are similar but swabbing generates a 20% false positive rate and dogs lead to a 5% false positive rate, that's important to know. If costs are comparable, but one process takes 5 times more resources (staffing private back room 'resolution' gropes, for example, to resolve false alarms) and time, that's also a factor to consider (at least for the pax).
In this particular instance, it also might be affected by TSA's notoriously inconsistent and sloppy practices. A close examination might reveal, for example, that most of the 'false positives' with swabs are not because the process itself is unreliable, but because TSA's sloppy and inconsistent practices are: unchanged gloves, poorly calibrated machines, cross-contamination, checkpoints where it's clear that there's not a single person on duty who understands cross-contamination, etc.
We have to compare results in order to understand which method produces more reliable results - and at what cost.
If, for example, costs are similar but swabbing generates a 20% false positive rate and dogs lead to a 5% false positive rate, that's important to know. If costs are comparable, but one process takes 5 times more resources (staffing private back room 'resolution' gropes, for example, to resolve false alarms) and time, that's also a factor to consider (at least for the pax).
In this particular instance, it also might be affected by TSA's notoriously inconsistent and sloppy practices. A close examination might reveal, for example, that most of the 'false positives' with swabs are not because the process itself is unreliable, but because TSA's sloppy and inconsistent practices are: unchanged gloves, poorly calibrated machines, cross-contamination, checkpoints where it's clear that there's not a single person on duty who understands cross-contamination, etc.
I know you do.
Is it better to fly a plane or drive a car from LA to DC? Both are transportation and both might get you there early or late. But they are so drastically different in approach and have so many other variables affecting the equation of "which is better" that only in the most basic of areas can they be fairly compared.
Same with these two types of explosives detection.
Dogs are more sensitive than the machines. But dogs might be trained on a smaller range of chemicals. Dogs get tired and sick. Machines don't need to go to the bathroom, don't need to sleep and don't get sick. Dogs can detect in more than place, dogs can screen exponentially more people / things in the same amount of time as a machine. But machines can be built and put into service much sooner than a dog. Dogs can detect a scent that is moving. When a machine part breaks, replace the part; cant replace a part of the dogs brain....
Just because the basic rate of both giving false positives/negatives can be measured and compared does not mean the comparison of the two is a fair measure of cost effectiveness.
#197
A FlyerTalk Posting Legend
Join Date: Sep 2006
Location: where the chile is hot
Programs: AA,RR,NW,Delta ,UA,CO
Posts: 41,700
I know you do.
Is it better to fly a plane or drive a car from LA to DC? Both are transportation and both might get you there early or late. But they are so drastically different in approach and have so many other variables affecting the equation of "which is better" that only in the most basic of areas can they be fairly compared.
Same with these two types of explosives detection.
Dogs are more sensitive than the machines. But dogs might be trained on a smaller range of chemicals. Dogs get tired and sick. Machines don't need to go to the bathroom, don't need to sleep and don't get sick. Dogs can detect in more than place, dogs can screen exponentially more people / things in the same amount of time as a machine. But machines can be built and put into service much sooner than a dog. Dogs can detect a scent that is moving. When a machine part breaks, replace the part; cant replace a part of the dogs brain....
Just because the basic rate of both giving false positives/negatives can be measured and compared does not mean the comparison of the two is a fair measure of cost effectiveness.
Is it better to fly a plane or drive a car from LA to DC? Both are transportation and both might get you there early or late. But they are so drastically different in approach and have so many other variables affecting the equation of "which is better" that only in the most basic of areas can they be fairly compared.
Same with these two types of explosives detection.
Dogs are more sensitive than the machines. But dogs might be trained on a smaller range of chemicals. Dogs get tired and sick. Machines don't need to go to the bathroom, don't need to sleep and don't get sick. Dogs can detect in more than place, dogs can screen exponentially more people / things in the same amount of time as a machine. But machines can be built and put into service much sooner than a dog. Dogs can detect a scent that is moving. When a machine part breaks, replace the part; cant replace a part of the dogs brain....
Just because the basic rate of both giving false positives/negatives can be measured and compared does not mean the comparison of the two is a fair measure of cost effectiveness.
I thought I made it clear (as if it needed to be made clear) in my post that I am not talking about something as basic as comparing which process generated the most 'false alerts' - or the most real 'catches'. Money, time, practicality (we're not putting full-time dog teams in every single airport) and predominately useful and accurate results have to be factored in - as does TSA's uncanny ability to muck up almost everything they touch.
Remember how many times we were assured that properly used, the NoS's would be almost as fast but much more reliable than the WTMD? No mention of the cumulative delays and staffing impacts from the constant ridiculous false positives that TSA decided require a dedicated staffer to resolve using clever techniques like stroking a bald man's head or arm because it alarmed.
It's entirely possible that neither approach is the single best solution; it may be that both are so flawed that it's time to start looking for a different solution. Without actual data, we have no way of knowing.
#198
Join Date: Nov 2007
Location: WAS
Programs: enjoyed being warm spit for a few years on CO/UA but now nothing :(
Posts: 2,507
I still fail to see how keeping accurate measurements is a bad thing.
I thought I made it clear (as if it needed to be made clear) in my post that I am not talking about something as basic as comparing which process generated the most 'false alerts' - or the most real 'catches'. Money, time, practicality (we're not putting full-time dog teams in every single airport) and predominately useful and accurate results have to be factored in - as does TSA's uncanny ability to muck up almost everything they touch.
Remember how many times we were assured that properly used, the NoS's would be almost as fast but much more reliable than the WTMD? No mention of the cumulative delays and staffing impacts from the constant ridiculous false positives that TSA decided require a dedicated staffer to resolve using clever techniques like stroking a bald man's head or arm because it alarmed.
It's entirely possible that neither approach is the single best solution; it may be that both are so flawed that it's time to start looking for a different solution. Without actual data, we have no way of knowing.
I thought I made it clear (as if it needed to be made clear) in my post that I am not talking about something as basic as comparing which process generated the most 'false alerts' - or the most real 'catches'. Money, time, practicality (we're not putting full-time dog teams in every single airport) and predominately useful and accurate results have to be factored in - as does TSA's uncanny ability to muck up almost everything they touch.
Remember how many times we were assured that properly used, the NoS's would be almost as fast but much more reliable than the WTMD? No mention of the cumulative delays and staffing impacts from the constant ridiculous false positives that TSA decided require a dedicated staffer to resolve using clever techniques like stroking a bald man's head or arm because it alarmed.
It's entirely possible that neither approach is the single best solution; it may be that both are so flawed that it's time to start looking for a different solution. Without actual data, we have no way of knowing.
But determining that one is simply "better" than the other based on a single measure is not really fair because they are so drastically different.
#199
Original Poster
Join Date: Aug 2012
Posts: 3,526
I didn't say measuring their effectiveness is a bad thing. I quite agree with you that tracking and evaluating performance of both are necessary.
But determining that one is simply "better" than the other based on a single measure is not really fair because they are so drastically different.
But determining that one is simply "better" than the other based on a single measure is not really fair because they are so drastically different.
How many times does the ETD alarm and yet no explosives are ever found even when bags are torn apart and passengers are hauled off to a private room to be groped? 100% of the time?
That alone should be sufficient measurement.
#200
Join Date: Nov 2007
Location: WAS
Programs: enjoyed being warm spit for a few years on CO/UA but now nothing :(
Posts: 2,507
The dog or machine will provide a response (not an "alarm") indicating the presence of the chemicals it is trained or designed to detect. Based on the response it is up to humans to determine if the presence of the chemical is a threat (determining if explosives or a device are present through additional inspection).
Sometimes the presence of these chemicals means explosive material and even an explosive device is present. But the chemicals can be found on items for a variety of reasons (maybe the dangerous material/device was in the bag but is not now; maybe someone handled black powder and then the bag; maybe nitrates were present in a lubricant [completely benign] that would be dangerous if present in black powder pipe bomb). Not finding actual explosives or a device does not mean the dog or machine failed.
The dog or machine can be considered to have failed if (a) they do not detect the presence of the chemicals when the chemicals are present, or (b) they detect the presence of the chemicals when the chemicals are not present.
But the fact that the dog or machine alerted but no explosive or device was found does not mean the dog or machine was in error.
#201
Join Date: Dec 2007
Programs: DL, WN, US, Avis, AA
Posts: 662
To repeat a quote from a long time ago, "Everything you said is true, it is also irrelevant." I've added comments below to illustrate why.
The above question is quite instructive as it is a perfect example of a layperson's lack of understanding of what the dogs and machines do. Both detect the presence of certain chemicals that are typically or commonly used in explosives. (and in hundreds, if not thousands of other, perfectly benign products)
The dog or machine will provide a response (not an "alarm") indicating the presence of the chemicals it is trained or designed to detect. Based on the response it is up to humans to determine if the presence of the chemical is a threat (determining if explosives or a device are present through additional inspection).
Sometimes the presence of these chemicals means explosive material and even an explosive device is present. But the chemicals can be found on items for a variety of reasons (maybe the dangerous material/device was in the bag but is not now; maybe someone handled black powder and then the bag; maybe nitrates were present in a lubricant [completely benign] that would be dangerous if present in black powder pipe bomb). Not finding actual explosives or a device does not mean the dog or machine failed.
The dog or machine can be considered to have failed if (a) they do not detect the presence of the chemicals when the chemicals are present, or (b) they detect the presence of the chemicals when the chemicals are not present.
But the fact that the dog or machine alerted but no explosive or device was found does not mean the dog or machine was in error. But, using dogs or machines for this purpose when 99+% of the time the determination is that no explosive or device is present is an indication that the thinking behind using the machines or dogs in this manner was flawed.
The dog or machine will provide a response (not an "alarm") indicating the presence of the chemicals it is trained or designed to detect. Based on the response it is up to humans to determine if the presence of the chemical is a threat (determining if explosives or a device are present through additional inspection).
Sometimes the presence of these chemicals means explosive material and even an explosive device is present. But the chemicals can be found on items for a variety of reasons (maybe the dangerous material/device was in the bag but is not now; maybe someone handled black powder and then the bag; maybe nitrates were present in a lubricant [completely benign] that would be dangerous if present in black powder pipe bomb). Not finding actual explosives or a device does not mean the dog or machine failed.
The dog or machine can be considered to have failed if (a) they do not detect the presence of the chemicals when the chemicals are present, or (b) they detect the presence of the chemicals when the chemicals are not present.
But the fact that the dog or machine alerted but no explosive or device was found does not mean the dog or machine was in error. But, using dogs or machines for this purpose when 99+% of the time the determination is that no explosive or device is present is an indication that the thinking behind using the machines or dogs in this manner was flawed.
#202
Original Poster
Join Date: Aug 2012
Posts: 3,526
The above question is quite instructive as it is a perfect example of a layperson's lack of understanding of what the dogs and machines do. Both detect the presence of certain chemicals that are typically or commonly used in explosives.
The dog or machine will provide a response (not an "alarm") indicating the presence of the chemicals it is trained or designed to detect. Based on the response it is up to humans to determine if the presence of the chemical is a threat (determining if explosives or a device are present through additional inspection).
Sometimes the presence of these chemicals means explosive material and even an explosive device is present. But the chemicals can be found on items for a variety of reasons (maybe the dangerous material/device was in the bag but is not now; maybe someone handled black powder and then the bag; maybe nitrates were present in a lubricant [completely benign] that would be dangerous if present in black powder pipe bomb). Not finding actual explosives or a device does not mean the dog or machine failed.
The dog or machine can be considered to have failed if (a) they do not detect the presence of the chemicals when the chemicals are present, or (b) they detect the presence of the chemicals when the chemicals are not present.
But the fact that the dog or machine alerted but no explosive or device was found does not mean the dog or machine was in error.
The dog or machine will provide a response (not an "alarm") indicating the presence of the chemicals it is trained or designed to detect. Based on the response it is up to humans to determine if the presence of the chemical is a threat (determining if explosives or a device are present through additional inspection).
Sometimes the presence of these chemicals means explosive material and even an explosive device is present. But the chemicals can be found on items for a variety of reasons (maybe the dangerous material/device was in the bag but is not now; maybe someone handled black powder and then the bag; maybe nitrates were present in a lubricant [completely benign] that would be dangerous if present in black powder pipe bomb). Not finding actual explosives or a device does not mean the dog or machine failed.
The dog or machine can be considered to have failed if (a) they do not detect the presence of the chemicals when the chemicals are present, or (b) they detect the presence of the chemicals when the chemicals are not present.
But the fact that the dog or machine alerted but no explosive or device was found does not mean the dog or machine was in error.
This layperson also thinks it's perfectly valid to examine how many times the dog alerts vs. the machine.
And this layperson also agrees with:
But, using dogs or machines for this purpose when 99+% of the time the determination is that no explosive or device is present is an indication that the thinking behind using the machines or dogs in this manner was flawed.
#203
FlyerTalk Evangelist
Join Date: Mar 2008
Location: DFW
Posts: 28,114
Seems to me that if a detection device alerts and nothing is found then a lot of time, money, and effort went into generating the alert for no gain. Part of the problem is the extreme rarity of WEI that could be detected by either ETD or canine. I'm not sure that either one is a suitable solution
#204
Join Date: Nov 2007
Location: WAS
Programs: enjoyed being warm spit for a few years on CO/UA but now nothing :(
Posts: 2,507
The clear implication (to me anyway) in your question, "How many times does the ETD alarm and yet no explosives are ever found even when bags are torn apart and passengers are hauled off to a private room to be groped?" is that not finding explosive material or a device means the dogs or machine have failed or at least are ineffective if not useless.
The premise of that implication shows a lack of understanding. The dogs and machines do not find bombs; they detect the presence of chemicals that indicate the potential presence of a bomb. That potential requires additional investigation.
That the investigation does not turn up explosive materials or a device does not mean the investigation or what kicked it off was faulty, ineffective, or useless.
#205
A FlyerTalk Posting Legend
Join Date: Sep 2006
Location: where the chile is hot
Programs: AA,RR,NW,Delta ,UA,CO
Posts: 41,700
Instead of responding to every potential threat, however unlikely, by hastily adding another 'layer', the entire protocol should be reviewed. If there are overlapping layers, they should be evaluated to see if the redundancy is desirable or simply inefficient.
Think of the nonsense with LGAs. It never made sense to say that a liter of a particular substance might pose a significant risk - but that same liter, broken down into 3.4 ounce containers, is perfectly legal - because the inconvenience of combining these hypothetical dangerous liquids will be enough to deter the 'bad guys'. That is BS, it was BS from the start, it is still BS. Kip Hawley has said he didn't have the cojones to get the ban lifted while he was still in charge, but that it should have been lifted - yet we continue to waste enormous time and money. Most recently, a TSO (again) was barking at people that they 'should know by now' that they have to throw their bottles of water away.
?? Hey, Mr. highly-trained-but-still-clueless TSO - do you know how many of the people you are yelling at and insulting have never flown before?
Meanwhile, while he's obsessing about water and TSA is ramping up a renewed focus on male genitals, people are put at unnecessary risk standing in long lines, wasting time and productivity. Taxpayers will pay to haul away any water bottles not personally confiscated by TSOs.
And none of it improves security one bit.
This is another reason for privatizing security - and reducing TSA's role to high-level oversight. A well-run, for-profit company, unlike TSA, would be motivated to eliminate unnecessary, redundant, or ridiculous ROI practices.
If used appropriately, dogs can be a useful tool. Unfortunately, it's more likely that TSA dog handlers will be like BDOs - just another excuse for TSA to provide higher-paying, low-responsibility jobs for favored employees with zero benefit to the traveling public.
Think of the nonsense with LGAs. It never made sense to say that a liter of a particular substance might pose a significant risk - but that same liter, broken down into 3.4 ounce containers, is perfectly legal - because the inconvenience of combining these hypothetical dangerous liquids will be enough to deter the 'bad guys'. That is BS, it was BS from the start, it is still BS. Kip Hawley has said he didn't have the cojones to get the ban lifted while he was still in charge, but that it should have been lifted - yet we continue to waste enormous time and money. Most recently, a TSO (again) was barking at people that they 'should know by now' that they have to throw their bottles of water away.
?? Hey, Mr. highly-trained-but-still-clueless TSO - do you know how many of the people you are yelling at and insulting have never flown before?
Meanwhile, while he's obsessing about water and TSA is ramping up a renewed focus on male genitals, people are put at unnecessary risk standing in long lines, wasting time and productivity. Taxpayers will pay to haul away any water bottles not personally confiscated by TSOs.
And none of it improves security one bit.
This is another reason for privatizing security - and reducing TSA's role to high-level oversight. A well-run, for-profit company, unlike TSA, would be motivated to eliminate unnecessary, redundant, or ridiculous ROI practices.
If used appropriately, dogs can be a useful tool. Unfortunately, it's more likely that TSA dog handlers will be like BDOs - just another excuse for TSA to provide higher-paying, low-responsibility jobs for favored employees with zero benefit to the traveling public.
Last edited by chollie; Jun 22, 2016 at 10:37 am
#206
Join Date: Nov 2007
Location: WAS
Programs: enjoyed being warm spit for a few years on CO/UA but now nothing :(
Posts: 2,507
(and in hundreds, if not thousands of other, perfectly benign products)
But, using dogs or machines for this purpose when 99+% of the time the determination is that no explosive or device is present is an indication that the thinking behind using the machines or dogs in this manner was flawed.
.
But, using dogs or machines for this purpose when 99+% of the time the determination is that no explosive or device is present is an indication that the thinking behind using the machines or dogs in this manner was flawed.
.
sure, but the dogs do not alert to most of those benign products. Still, the inconvenience of alerting to a benign product is much less inconvenient than not screening and missing a malignant product.
That same logic would dictate that the thinking behind using the xray scanners and metal detectors was flawed, too, since 99+% of the time only water and nail clippers are detected.
Having said that, you raise a good point - are the dogs being used the most effectively? There is debate on how best to use the dogs; cargo screening is one example. Quite a few folks advocate that, in general, cargo screening is ill-suited for the dogs and that machines would be better. I believe they are correct but unfortunately the state of technology is still not quite there yet for large scale deployment of machines in cargo operations and airports to be reasonable; in most situation dogs are still more cost efficient.
Explosives detection machines for large scale cargo operations are possible for organizations with nearly bottomless security budgets such as the White House, CIA DOD, and NSA. But not so much for airports.
#207
Join Date: Dec 2007
Programs: DL, WN, US, Avis, AA
Posts: 662
That same logic would dictate that the thinking behind using the xray scanners and metal detectors was flawed, too, since 99+% of the time only water and nail clippers are detected.
Having said that, you raise a good point - are the dogs being used the most effectively? There is debate on how best to use the dogs; cargo screening is one example. Quite a few folks advocate that, in general, cargo screening is ill-suited for the dogs and that machines would be better. I believe they are correct but unfortunately the state of technology is still not quite there yet for large scale deployment of machines in cargo operations and airports to be reasonable; in most situation dogs are still more cost efficient.
Explosives detection machines for large scale cargo operations are possible for organizations with nearly bottomless security budgets such as the White House, CIA DOD, and NSA. But not so much for airports.
Having said that, you raise a good point - are the dogs being used the most effectively? There is debate on how best to use the dogs; cargo screening is one example. Quite a few folks advocate that, in general, cargo screening is ill-suited for the dogs and that machines would be better. I believe they are correct but unfortunately the state of technology is still not quite there yet for large scale deployment of machines in cargo operations and airports to be reasonable; in most situation dogs are still more cost efficient.
Explosives detection machines for large scale cargo operations are possible for organizations with nearly bottomless security budgets such as the White House, CIA DOD, and NSA. But not so much for airports.
#208
Join Date: Nov 2007
Location: WAS
Programs: enjoyed being warm spit for a few years on CO/UA but now nothing :(
Posts: 2,507
OK. Now I'm confused. If it is true that "(t)he dog or machine will provide a response (not an "alarm") indicating the presence of the chemicals it is trained or designed to detect" how is it that the dog can know that those chemicals are contained within a threat item or one that is benign? Are the dogs providing a response to the item itself or to the chemicals it contains?
The dog does not know or determine if the item is a threat or not - that is for humans to determine.
The dogs and machines are trained to respond to certain chemicals and compounds of chemicals.
However, in certain compounds those same chemicals are benign and therefore the dogs are not trained to respond to those compounds.
I hope that helps.
#209
Original Poster
Join Date: Aug 2012
Posts: 3,526
Those are good questions - sorry if I made it confusing. The dogs have fantastically sensitive olfactory systems - some breeds can detect in the range of parts per trillions.
The dog does not know or determine if the item is a threat or not - that is for humans to determine.
The dogs and machines are trained to respond to certain chemicals and compounds of chemicals.
However, in certain compounds those same chemicals are benign and therefore the dogs are not trained to respond to those compounds.
I hope that helps.
The dog does not know or determine if the item is a threat or not - that is for humans to determine.
The dogs and machines are trained to respond to certain chemicals and compounds of chemicals.
However, in certain compounds those same chemicals are benign and therefore the dogs are not trained to respond to those compounds.
I hope that helps.
That alone says to me that the machines are useless.
#210
Join Date: Nov 2007
Location: WAS
Programs: enjoyed being warm spit for a few years on CO/UA but now nothing :(
Posts: 2,507
The dogs and machines do their jobs very well; that they are not capable of applying judgment and discretion in these matters does not mean they are useless.