Does this technology have a slippery slope? Assuming that the technology isn't useful for recording conversations, what's the worst that could happen by implementing this in residential neighborhoods?
As far as I see it you waste money that could be better used in staffing police, implementing social programs etc - this thing is a solution to a very tiny problem: gunshots that are not reported via emergency calls. As far as I can see this does not help with robberies, burglaries, attacks with fists/knives/bats etc.
Even if its not a slippery slope - I'd imagine there are several better alternatives where you can spend the money.
> this thing is a solution to a very tiny problem: gunshots that are not reported via emergency calls
I wondered how big a problem this is, but the only number I quickly could find is annoyingly from their advertising material and claims that in some areas only 10 % of shots were reported, which is not really the relevant number. Does anyone have a number on how many people die of gunshot injuries that either weren't reported or reported to late?
That's also not the relevant number. In places like Chicago, what you end up with is a tiny proportion of the population (say 2000) that is perhaps 100 times more likely to be murdered and is a major drain on your city's image. This is fuelled by revenge killings; someone gets shot or shot at, their family/friends round up the posse.
So what you want to do is get people off the street before they start shooting, or if that fails, you get them off the street as soon as they start shooting in order to break the cycle.
At the moment, with the dispatching of officers etc, you find that it takes about 2 minutes from receiving a call to an officer being dispatched (that's from the London Ambulance Service). A lot of this is just the time between the call arriving and the operator working out where the person is. Before the call, there are perhaps 15 seconds where the bystander who calls up is getting their phone out and dialing.
In this case, with the new system you can in principle have
two squad cars each half a mile away knowing where the gunshot is to a reasonable level of accuracy, perhaps within a couple of seconds of the shot itself.
Within a minute of the first shot, the police can be on the scene. In this case, the suspects have only just jumped into their cars, and so are picked up easily.
There's no opportunity for any kind of revenge, because the suspects are in a cell.
Of course it's not a fantastic idea for every city; but if you're a company that focuses on military hardware, filtering off into LA, Chicago and NY can easily be an advantage for all.
>>this thing is a solution to a very tiny problem: gunshots that are not reported via emergency calls.
That's not exactly true. When people call in gunshots they frequently have at best a vague idea where the shots came from. If you're sitting in your house with the window open, the shots will most likely sound as if they came from that direction - even if they were actually in the opposite direction. So you could be sending the police to a place 6 blocks away, which in a dense urban environment is quite a long ways indeed.
Being able to reliably figure out the actual location has value.
source: we have lots of gunshots in our neighborhood. Police response and ability to actually catch people has gone up since the spotter was installed.
I'm not an expert, or even a layman in audio encoding, but it seems to me that a solution could be found that couples an open source capture program with their backend. The open source (audited and only updated from the open source branch) capture program filters the audio stream down to something that only contains a gunshot signature, and nothing else. If the motivation is there a solution is certainly possible.
But as the article says, they need the original audio in order to improve the gunshot detection - for example, if a gunshot is reported elsewhere but the detector doesn't classify it, they can grab the audio from the record to train the detector and hopefully improve it.
If you never capture the raw audio, you won't be able to train the system to better detect the gunshots and to reduce false positives.
I happened to start working with a similar system today. For our customer there are several factors they are interested in:
# Mapping gunshots in both time and space. These types of sensors can attempt to classify the type of weapon used based on the audio signature. If there are are several shots from similar-sounding sources moving through the city at a certain, or in a certain area over the period of days (or month), this can tell you more than you could determine from just emergency calls.
# Tieing the gunshot sensors in with other input devices, like cameras. You can reduce the time spent looking through, say, traffic camera recordings that happened in the vicinity of the gunshot looking for related video.
# Automatically pointing moveable cameras to look in the direction of a gunshot. Means you don't need as many camera for the same effective coverage.
Depends where you live. We've had gunshots in our neighborhood every day (or very nearly so) for 2+ weeks. I lost count of fatalities sometime around 9. Bit of a gang war going on here.
The ShotSpotter they put in last fall is definitely seeing some use here.
I don't doubt that you can hear gunshots in the rough parts of Oakland or LA, but this idea that people in the US are constantly in fear of getting shot is just so ridiculous.
East bay here, while I don't necessarily fear getting shot, it would be very nice if I was able to feel safe walking home at night, feel safe having my phone out, etc... I really would prefer to not be constantly checking my surroundings every half second
I've heard gunshots in SF more than a few times. Several times in the mission, once in the tenderloin, and once was a half block away from my house in the Richmond. I walked over right as EMS got there and saw them trying to stabilize a neighbor in a pool of blood.
I also made the dumb mistake of tapping on a window sitting on my couch once when I saw a bunch of funny dudes with helmets just outside it. When they looked in the window I realized it was actually SWAT, right before they raided the house two doors down.
I know that the amount of gun violence in the US is sometimes a shock to Europeans/foreigners in general, but I think a more pertinent question for most Americans is "Is the cost/benefit ratio worth it? Is this a proper allocation of public funds, or is it just another toy for local police departments to use?" My surprise in seeing that this is a sustainable business is this latter point: single-vendor sourced items and IDIQ contracts are a recipe for government spending disasters.
EDIT: My impression is that gun crime is heavily concentrated down to individual blocks or neighborhoods in urban areas. I don't like the idea of these devices in general, but I imagine that if they were deployed, they would be concentrated in so-called "high risk areas".
I was more under the impression that the gun violence is not really so high as to justify a constant monitoring system to watch out for (especially if you need a person to look over the reports and then alert law enforcement - I get the feeling a emergency hotline can cover this as well).
This looks like it targets a very small (but probably high profile) subsection of criminal activities. For example, it wouldn't detect someone robbing/beating/stabbing someone.
Ah, I might have mistook your surprise for an assumption in the other direction. I agree, I do not think that crime levels justify this sort of expenditure or monitoring. Maybe part of the appeal for law enforcement here is something that emergency hotlines can't replicate: this sort of system allows the police to be virtually omnipresent. An emergency hotline requires an engaged citizenry, one that would call the cops when a shooting happens, something that many pundits would argue crime-ridden communities are lacking.
Yes, that might appeal to law enforcement, even if their intentions are good. It would be interesting to compare the benefits to other programs, like raising the amount of patroling police etc.
Witnesses calling the police can't reliably locate the source of the gunfire. Wind, hills and buildings carry, echo and deflect the source beyond any hope. That's why a triangulating system like this is essential.
I think the business was actually built for the military (quickly locating where enemy fire is coming from is fairly high value there) and has since been marketed to civilian law enforcement.
At least one civilian agency which has adopted it seems to have found that its not worth the cost. [0]
That makes sense. I've seen something similar on the TV show "Flashpoint", mobile microphones/sensors were used to triangulate the position of a sniper iirc. This obviously makes sense, if you have the manpower to react to something like this.
But yeah, it might look cool at first glance, but I don't think its worth its money.
Note that that's extra tricky, because there are two generally sounds: the first received will be from the shock wave of supersonic bullet, then the report from the rifle.
Only if its a supersonic projectile AND its still supersonic when it passes the mic. Assuming a 115gr 9mm +P (.167 BC) that leaves the muzzle at 1275fps, the bullet slows to subsonic velocities between 50 and 75 yards. Assuming standard day at sea level.
Likely they’re looking for the muzzle blast. Also, rifles generally aren’t used to commit crimes, so looking for the shockwave exclusively will be next to pointless.
With 92 shooting victims in that same time period, I'm guessing that's a massive undercount. Probably only counts times they catch and charge a shooter.
Precisely my point. 92 shooting victims and 187 shots fired incidents means a 50% successful hit rate assuming every shots fired incident intended to hit someone.
Your last statistic ... isn't so useful, I haven't heard of Bloods and Crips laying down suppressive fire with General Purpose Machine Guns ^_^!
Even going to, say, the statistic I read some time ago of police hitting about 1 in 4 times they fired isn't so useful, because this I take this to be a metric of shooting incidents, in which multiple gun shots were probably fired.
To clarify, I doubt gang members in LA are firing hundreds of thousands of rounds... but I also doubt they're only firing at live targets and hitting as high as half of them. There's a reason they have reputations for "pray and spray".
The US military kills people with high powered machine guns, uses tactics like suppressive fire, trains constantly with live ammunition, etc. I'm not the comparison is very relevant.
> Given that the US military uses up 250k rounds per killed insurgent
I'm willing to bet that average engagement ranges in street violence in the US are closer, and the involved parties less effective at using cover and concealment, than is usually the case in battles between professional modern military forces and insurgent groups that still exist after several years of ongoing combat with professional modern military forces. Also, the weapons used are different.
So, generally, I wouldn't use statistics for the latter as a yardstick for evaluating shots/victim ratios for the former.
I don't know how much they're charging, but in violent cities it sounds useful. You can have the exact location of a gunshot and dispatch an officer in a matter of seconds, and microphone arrays should be pretty cheap. Together with cell-id positioning it's a powerful tool for investigation: you can have the exact location/time of the shot and exactly who was nearby that that moment.
The alternative for violent areas is relying on witnesses to both dispatch help and confirm date and place of gunshots.
And insanely, they're interested in "detecting gunshots" rather than stopping them in the first place, by, you know, taking guns away.
I'm not from the US, so it was quite shocking to be a friends BBQ outside LA and hear multiple gunshots over the course of the night. They are quiet teachers, and have kids. They don't need gunshots in their lives.
Yeah, well, trying to "[take] guns away" would trigger a civil war that would kill orders of magnitudes more people. That, in fact, is what sparked the American Revolution, when the British attempted to seize or destroy gunpowder, cannon, etc. (they did manage to damage 3 cannon).
If those people don't need gunshots in their lives, they should move to some place where they're not so common. Like the majority of the US where concealed carry is legal and, it is my impression, such bad areas are much more compressed.
I think you're reading more into what I've said so far than what I've actually said. In no particular order:
Adjust for demographics and that difference tends to go poof. Some time ago, and I should really check up on this, it was said that if you subtract the South, the rest of the US is pretty much normal. Being from the cultural South, I can believe this.
Decreasing homicides is not the end and and be all of what I consider to be a good society. The U.K. approach of outlawing effective self-defense per se, in the courts in the '50s and legislatively in the '60s, has resulted in a lot more overall violence, with innocents getting a lot more victimized. No thanks.
The history of genocidal slaughters of disarmed peoples in the 20th Century, which started with the Armenians in Turkey 100 years ago, ought to give pause to those desiring to disarm peoples.
A lot of Americans have a taste for freedom that doesn't seem apparent in much of the rest of the world; well, aside from all those who try to move here. I wonder why...?
> A lot of Americans have a taste for freedom that doesn't seem apparent in much of the rest of the world; well, aside from all those who try to move here. I wonder why...?
You should also note that much of the discussion in this thread has surrounded CA (specifically SF, Oakland, LA, etc.)
CA is largely hostile towards legal firearm ownership (getting put back into place by the courts, slowly but surely, though!)
These gunshots are not coming from law abiding citizens: most of these incidents are gang-related.
And you know what? Murder is illegal. More severe than any firearm-related law in the United States. If these individuals are attempting or successfully carrying-out murder, why and how do firearm laws prevent anyone but "good guys" from using them for legal purposes?
> That, in fact, is what sparked the American Revolution, when the British attempted to seize or destroy gunpowder, cannon, etc. (they did manage to damage 3 cannon).
This is a pretty absurd over-simplification of the causes of the American Revolution. Lexington and Concord certainly involved a British march on militia supplies including weapons/ammunition, but those weapons/ammunition were being gathered not for their own sake, but because of decades of tension over taxation, quartering, limitations on public gatherings, things like the Boston Massacre, etc. leading up to war. The Massachusetts militia weren't just a bunch of gun collectors who got pissed about gun control.
> it is my impression, such bad areas are much more compressed
I'm not sure ghettoization should be pointed to as a success story.
Not all US cities, and even then not all parts of those cities. In places with high poverty rates and gangs protecting their turf: yes. The gun violence rate is orders of magnitude higher there.
The privacy concerns with systems like these is not what they currently do, but what they are capable of.
It doesn't matter what your privacy policy says, and it doesn't matter if the default configuration is to only download the audio near the time of a possible gunshot.
The important part is that these devices can stream audio to remote servers. The NSA or anybody who finds or can coerce accesses to the system could use them to eavesdrop on everything within range.
No amount of "privacy policy" is going to prevent it.
That's assuming that audio from these sensors is at all useful for eavesdropping on people. Try a speakerphone lately? Even with higher quality codecs than we got in the POTS days and things like beamforming, the result is pretty terrible in decent sized conference rooms. Now try this outdoors, further away from people, with the noise floor of a city instead of a quiet room.
The cases people point to where ShotSpotter has picked up understandable words have been people very close to sensors who were shouting immediately before, during, or after gunfire the system picked up. But at that point, it's unreasonable to claim some sort of violation of privacy when people are screaming for all to hear.
Redwood City, CA had the first system. It was much simpler than this. They had analog microphones on phone poles, connected to a logarithmic amplifier which traded dynamic range for bandwidth. Phone audio is 8 bits at 8KHz; think of this as remapping to 16 bits at 4Khz, although it was done with analog hardware. They needed the dynamic range to see the full risetime of a gunshot, and gave up a big chunk of the voice spectrum to get it.
Each microphone had a wired phone line connection to the central system at the police station, which was a PC with a data collection card running LabView. A LabView program turned gunshot information into map plots. It was a simple, local system.
The new systems are much fancier. They're "cloud-based", which means the data goes first to ShotSpotter's HQ. They have synchronized clocks (probably from GPS), so they can send in timed audio snippets to be matched at the analysis center. It's a service now, rather than just hardware. Not clear what kind of data connection they have; probably cellular data.
Depends on the sophistication of the system, and how many microphones it has. The vast majority, if not pretty much all of this class of gunshots, are going to start with the bullet being supersonic, the shock wave of which contributes to the signature (in the case of sniper fire I mentioned elsewhere in this discussion, it will tend to fool people into thinking the source is as much as 90 degrees from where the rifle actually is).
I'm having difficultly coming up with anything that isn't essentially a blank firing gun, and I'll note those are deadly really close up.
To my thoroughly trained ears they sound very different.
Thinking about it, to avoid false positives, companies doing this need to filter out fireworks, the proverbial car backfire (do they do that any more???), etc.
I was looking for powerful magic spells for love. I was upset and depressed. Dr PayBack helped me finding my true love. I will never go to any other spell caster except Dr PayBack. quickrevengespell@yahoo.com, John (San Antonio TX)
58 comments
[ 3.0 ms ] story [ 84.8 ms ] threadEven if its not a slippery slope - I'd imagine there are several better alternatives where you can spend the money.
I wondered how big a problem this is, but the only number I quickly could find is annoyingly from their advertising material and claims that in some areas only 10 % of shots were reported, which is not really the relevant number. Does anyone have a number on how many people die of gunshot injuries that either weren't reported or reported to late?
So what you want to do is get people off the street before they start shooting, or if that fails, you get them off the street as soon as they start shooting in order to break the cycle.
At the moment, with the dispatching of officers etc, you find that it takes about 2 minutes from receiving a call to an officer being dispatched (that's from the London Ambulance Service). A lot of this is just the time between the call arriving and the operator working out where the person is. Before the call, there are perhaps 15 seconds where the bystander who calls up is getting their phone out and dialing.
In this case, with the new system you can in principle have two squad cars each half a mile away knowing where the gunshot is to a reasonable level of accuracy, perhaps within a couple of seconds of the shot itself.
Within a minute of the first shot, the police can be on the scene. In this case, the suspects have only just jumped into their cars, and so are picked up easily.
There's no opportunity for any kind of revenge, because the suspects are in a cell.
Of course it's not a fantastic idea for every city; but if you're a company that focuses on military hardware, filtering off into LA, Chicago and NY can easily be an advantage for all.
That's not exactly true. When people call in gunshots they frequently have at best a vague idea where the shots came from. If you're sitting in your house with the window open, the shots will most likely sound as if they came from that direction - even if they were actually in the opposite direction. So you could be sending the police to a place 6 blocks away, which in a dense urban environment is quite a long ways indeed.
Being able to reliably figure out the actual location has value.
source: we have lots of gunshots in our neighborhood. Police response and ability to actually catch people has gone up since the spotter was installed.
http://www.lightningmaps.org/realtime
http://www.blitzortung.org/Webpages/index.php?lang=en&page=3
TL;DR They do triangulated time of flight for EM emissions from lightning strikes.
In a city, you get more issues: double datapoints from the gun and the bullet, multipath effects, etc etc etc.
If you never capture the raw audio, you won't be able to train the system to better detect the gunshots and to reduce false positives.
Why would a city invest in something like this? Especially if you get the information via emergency calls?
# Mapping gunshots in both time and space. These types of sensors can attempt to classify the type of weapon used based on the audio signature. If there are are several shots from similar-sounding sources moving through the city at a certain, or in a certain area over the period of days (or month), this can tell you more than you could determine from just emergency calls.
# Tieing the gunshot sensors in with other input devices, like cameras. You can reduce the time spent looking through, say, traffic camera recordings that happened in the vicinity of the gunshot looking for related video.
# Automatically pointing moveable cameras to look in the direction of a gunshot. Means you don't need as many camera for the same effective coverage.
The ShotSpotter they put in last fall is definitely seeing some use here.
Though to be fair it was New Years in San Diego, and Santee is a bit out of the way
I also made the dumb mistake of tapping on a window sitting on my couch once when I saw a bunch of funny dudes with helmets just outside it. When they looked in the window I realized it was actually SWAT, right before they raided the house two doors down.
EDIT: My impression is that gun crime is heavily concentrated down to individual blocks or neighborhoods in urban areas. I don't like the idea of these devices in general, but I imagine that if they were deployed, they would be concentrated in so-called "high risk areas".
This looks like it targets a very small (but probably high profile) subsection of criminal activities. For example, it wouldn't detect someone robbing/beating/stabbing someone.
Yes, that might appeal to law enforcement, even if their intentions are good. It would be interesting to compare the benefits to other programs, like raising the amount of patroling police etc.
At least one civilian agency which has adopted it seems to have found that its not worth the cost. [0]
[0] http://www.sfgate.com/crime/article/Oakland-cops-aim-to-scra...
But yeah, it might look cool at first glance, but I don't think its worth its money.
Likely they’re looking for the muzzle blast. Also, rifles generally aren’t used to commit crimes, so looking for the shockwave exclusively will be next to pointless.
Between April 5 and May 2 of this year, there were 187 recorded cases of shots fired in all of LA.
[Edit: I misunderstood]
Given that the US military uses up 250k rounds per killed insurgent (http://jonathanturley.org/2011/01/10/gao-u-s-has-fired-25000...) I rather suspect LA has more than ~200 shots fired a month.
Even going to, say, the statistic I read some time ago of police hitting about 1 in 4 times they fired isn't so useful, because this I take this to be a metric of shooting incidents, in which multiple gun shots were probably fired.
I'm willing to bet that average engagement ranges in street violence in the US are closer, and the involved parties less effective at using cover and concealment, than is usually the case in battles between professional modern military forces and insurgent groups that still exist after several years of ongoing combat with professional modern military forces. Also, the weapons used are different.
So, generally, I wouldn't use statistics for the latter as a yardstick for evaluating shots/victim ratios for the former.
The alternative for violent areas is relying on witnesses to both dispatch help and confirm date and place of gunshots.
I'm not from the US, so it was quite shocking to be a friends BBQ outside LA and hear multiple gunshots over the course of the night. They are quiet teachers, and have kids. They don't need gunshots in their lives.
If those people don't need gunshots in their lives, they should move to some place where they're not so common. Like the majority of the US where concealed carry is legal and, it is my impression, such bad areas are much more compressed.
Adjust for demographics and that difference tends to go poof. Some time ago, and I should really check up on this, it was said that if you subtract the South, the rest of the US is pretty much normal. Being from the cultural South, I can believe this.
Decreasing homicides is not the end and and be all of what I consider to be a good society. The U.K. approach of outlawing effective self-defense per se, in the courts in the '50s and legislatively in the '60s, has resulted in a lot more overall violence, with innocents getting a lot more victimized. No thanks.
The history of genocidal slaughters of disarmed peoples in the 20th Century, which started with the Armenians in Turkey 100 years ago, ought to give pause to those desiring to disarm peoples.
A lot of Americans have a taste for freedom that doesn't seem apparent in much of the rest of the world; well, aside from all those who try to move here. I wonder why...?
If you believe that, you need to get out more.
CA is largely hostile towards legal firearm ownership (getting put back into place by the courts, slowly but surely, though!)
These gunshots are not coming from law abiding citizens: most of these incidents are gang-related.
And you know what? Murder is illegal. More severe than any firearm-related law in the United States. If these individuals are attempting or successfully carrying-out murder, why and how do firearm laws prevent anyone but "good guys" from using them for legal purposes?
This is a pretty absurd over-simplification of the causes of the American Revolution. Lexington and Concord certainly involved a British march on militia supplies including weapons/ammunition, but those weapons/ammunition were being gathered not for their own sake, but because of decades of tension over taxation, quartering, limitations on public gatherings, things like the Boston Massacre, etc. leading up to war. The Massachusetts militia weren't just a bunch of gun collectors who got pissed about gun control.
> it is my impression, such bad areas are much more compressed
I'm not sure ghettoization should be pointed to as a success story.
Sorry, going to need an amendment to the Constitution of the United States of America for that.
And even after said amendment: good luck preventing people from acquiring and illegally owning firearms.
Inner-city gang wars aren't even remotely reason to infringe upon one of the most important American rights.
It doesn't matter what your privacy policy says, and it doesn't matter if the default configuration is to only download the audio near the time of a possible gunshot.
The important part is that these devices can stream audio to remote servers. The NSA or anybody who finds or can coerce accesses to the system could use them to eavesdrop on everything within range.
No amount of "privacy policy" is going to prevent it.
The cases people point to where ShotSpotter has picked up understandable words have been people very close to sensors who were shouting immediately before, during, or after gunfire the system picked up. But at that point, it's unreasonable to claim some sort of violation of privacy when people are screaming for all to hear.
Each microphone had a wired phone line connection to the central system at the police station, which was a PC with a data collection card running LabView. A LabView program turned gunshot information into map plots. It was a simple, local system.
The new systems are much fancier. They're "cloud-based", which means the data goes first to ShotSpotter's HQ. They have synchronized clocks (probably from GPS), so they can send in timed audio snippets to be matched at the analysis center. It's a service now, rather than just hardware. Not clear what kind of data connection they have; probably cellular data.
Would be fun to put dozens up all around, on slow timers.
I'm having difficultly coming up with anything that isn't essentially a blank firing gun, and I'll note those are deadly really close up.
Thinking about it, to avoid false positives, companies doing this need to filter out fireworks, the proverbial car backfire (do they do that any more???), etc.
“You were so focused on whether you could do it, you never stopped to ask whether you should.” - Jurassic Park