"Arbitron: PPMs Track HD Radio"

[Mike Walker]

Why exactly anyone would want to listen to our conversations is beyond me (mine ain't that fascinating!), but awj does make an interesting point. I just neither buy into, nor care about the conspiracy theories. Life is short! But these things DO, apparently, listen and record with microphones. Something to at least think about. Ok, I've thought about it. Time for a snack

 

[awj223]

There are lots of motivations I can think of to use these things for evil purposes. The problem I see is that it might not be too difficult to justify adding this capability to the thing.

"It's for quality control purposes only. We want to know what types of background noise the PPM has trouble with so that we can make improvements to future versions. Audio recording will only be enabled when our field testers use it, and (should be) disabled in units sent to listeners."

"Not all stations will be running PPM equipment. We need audio recording to create more accurate diary returns."

etc., etc.

 

[davideduardo]

This thing is creepy. Doesn't anyone have any concerns about how easy it would be to turn this thing into an eavesdropping device? Imagine...a business exec has a discussion about confidential info from the car. The PPM, instead of recording only the tones from the radio, is also trojaned to record the entire audio spectrum, which gets compressed, stored in its memory, and sent off to a third party.

The memory capacity of the PPM is totally inadequate to store anything but the codes that indicate station and time stamp. You would have to rebuild the entire device to do anything else, and the nightly docking verifies the integrity of the device... plus it can only communicate with Arbitron.

The bigger issue is that nobody knows, outside of Arbitron, who has the PPMs and very few people will have them... in LA, with 10.3 million persons 12+, there will be less than 4 thousand meters.

Also I don't see how it would work when you have splatter and/or IBOC hiss partially obliterating the signal you're listening to. The human ear can pick information out of some VERY noisy signals. Can the PPM do this just as well? If not, the TSL values are going to be inaccurate.

The PPM only mesures at a specific audio frequency set. If a listener can hear the audio on the station, the PPM can extract the encoding. There is one code burst every 4.5 seconds. To "count" there has to be 5 minutes of cumulative listening in any quarter hour to get credit for the quarter hour. Shorter spans are not counted as "real" listening.

How many listeners do you think will be listening to unlistenable signals, anyway? If the PPM can not detect, the listener can not hear anything anyway. Remember, too, that the objective of all audience ratings is to determine the cost and efficiency of radio advertising.

 

[davideduardo]

.
This is just wonderful. Portable People Meters can only register and track stations using the encoded data.

Each station is encoded, using a piece of equpment provided, free, by Arbitron to every station in the market area, including fringe stations. Eventually, every station will be encoded, as well as satellite, HD-2 channels, storecasts, web streams...

How do they presume to measure accurately time spent listening to stations which do not wish to buy into this system?

In the test markets, every station is encoded, subscriber or not.

This seems designed to further marginalize smaller stations.

Why? If they are in or at the fringe of a PPM market, they will be given an encoder and technical assistance.

What about stations not wishing to inject data into their audio?

There are no cases of this.

What about low-level listening in arears with ambient noise?

If the listener can hear the audio, the PPM can detect the identification code.

If Arbitron ratings were considered questionable before, this only seems like a way for stations to buy their way into a higher cume.

Huh? The measurement is much more accurate, as it is vastly more precise than a diary which depends on not just listening but remembering what was listened to at the moment the diary is filled in.

I find more reasons to distrust the results and conclusions drawn from the PPMs, as opposed to diaries.

The results are amazingly similar overall... shares are pretty much the same.

Someone, please explain how stations not participating will be rated.

Every station participates... in tests in Philly and Houston over the last 5 years, Arbitron got 100% installation (the only resistence was due to negotiations with Arbitron on price, etc.). No non-subscriber failed to participate by installing the hardware.

Will there be 2 systems, or will the non participants cease to show in the books?

The diary method is being REPLACED by the PPM. And all stations are participating in the roll out markets.

How will non-participants negotiate advertising rates with advertisers in the face of apparent sudden loss of listenership?

First, even in the diary, non-participants can not use the ratings data to sell, so that point is moot. And the fact is that all stations in a market area will participate, as the tests have shown.

 

[awj223]

The memory capacity of the PPM is totally inadequate to store anything but the codes that indicate station and time stamp. You would have to rebuild the entire device to do anything else, and the nightly docking verifies the integrity of the device... plus it can only communicate with Arbitron.

The bigger issue is that nobody knows, outside of Arbitron, who has the PPMs and very few people will have them... in LA, with 10.3 million persons 12+, there will be less than 4 thousand meters.

But unless I rip open the unit, there's no way for me to verify that, is there? And I'm sure that that's against the terms of agreement if you sign up to use one.

I understand what you're saying, but in today's world 1GB of data fits on a device the size of a MicroSD card. It would NOT be difficult at all to slip enough memory in there to record several hours worth of audio. Even 12-18 hours of speech quality audio isn't out of the question. This thing is about the size of an iPod. How many hours of music do iPods store?

The PPM only mesures at a specific audio frequency set. If a listener can hear the audio on the station, the PPM can extract the encoding. There is one code burst every 4.5 seconds. To "count" there has to be 5 minutes of cumulative listening in any quarter hour to get credit for the quarter hour. Shorter spans are not counted as "real" listening.

How many listeners do you think will be listening to unlistenable signals, anyway? If the PPM can not detect, the listener can not hear anything anyway. Remember, too, that the objective of all audience ratings is to determine the cost and efficiency of radio advertising.

Well from what I know the human ear does an excellent job picking speech out of a crappy signal, and technology doesn't do nearly as well. You can easily understand someone talking to you in a crowded room but speech recognition software has a long way to go before it can match that feat. For the same reason, I can hear and understand advertisements on a desired station over sideband splatter, IBOC hash, and power line noise, and other types of interference. Maybe the PPM can do well under these circumstances because it's not doing speech recognition, but only has to pick a few bits of information out of a signal over a long period of time. That has to be an advantage.

 

[davideduardo]

But unless I rip open the unit, there's no way for me to verify that, is there? And I'm sure that that's against the terms of agreement if you sign up to use one.

First, it can not be opened by the end user. It is sealed. Second, every night the unit does system verification on the unit to make sure it is not damaged or otherwise not the same unit.

Again, the panel is a tiny percentage of the market population, and participants (not "users") are kept secret. If any panelist goes public, the person is kicked from the panel.

I understand what you're saying, but in today's world 1GB of data fits on a device the size of a MicroSD card. It would NOT be difficult at all to slip enough memory in there to record several hours worth of audio.

The device would know it has been modified when it docks and talks to Arbitron. The dock and the device can only communicate with Arbitron. The meter is sealed, and the size of a pager, and is (I've seen the insides of one at the Arbitron lab) so tightly packed nothing could be added or moved... and doing so would cause the device to be rejected.

Even 12-18 hours of speech quality audio isn't out of the question. This thing is about the size of an iPod. How many hours of music do iPods store?

It is half the size of an i-pod, and does not have a hard drive. Or any place inside to add memory. This is getting very "black helicopter." If we have no reports of modified cell phones, this is not going to happen with PPMs since the chances of anyone you are interested in spying on are very low since the panels are not large.

Well from what I know the human ear does an excellent job picking speech out of a crappy signal, and technology doesn't do nearly as well. You can easily understand someone talking to you in a crowded room but speech recognition software has a long way to go before it can match that feat. For the same reason, I can hear and understand advertisements on a desired station over sideband splatter, IBOC hash, and power line noise, and other types of interference. Maybe the PPM can do well under these circumstances because it's not doing speech recognition, but only has to pick a few bits of information out of a signal over a long period of time. That has to be an advantage.

The PPM detects a code burst that gives the unique station ID code and the time. If a listener can hear the audio, the PPM can detect the code.

As you say, this is not speech but a data stream. The PPM has essentially a filter that only hears a specific very narrow notch of the audio specturm. If there is a code stream there, it it will detect. I saw a demonstation in my office a few weeks ago where a radio was put in a computer bag... and the meter was outside on a table. When the meter was docked, it showed the station as being identified.

Since the code and time is sent about 12 times a minute, missing moments are not critical.

 

[awj223]

First, it can not be opened by the end user. It is sealed. Second, every night the unit does system verification on the unit to make sure it is not damaged or otherwise not the same unit.

This assumes that
1) the programmers who made the firmware didn't insert trojan horses into their code that would allow the system to be modified without sounding an alarm, and
2) the firmware contains no exploits that can be used to modify it without sounding the "alarm".

Then the system has to report back to Arbitron. As soon as it connects to a public network to send its data, it's exposed to potential hacks from 3rd parties. It has been shown again and again that people make mistakes when writing code. Exploits are constantly being found and fixed.

Again, the panel is a tiny percentage of the market population, and participants (not "users") are kept secret. If any panelist goes public, the person is kicked from the panel.

How can the panelists be kept secret when they go around in public wearing the PPMs?

The device would know it has been modified when it docks and talks to Arbitron. The dock and the device can only communicate with Arbitron. The meter is sealed, and the size of a pager, and is (I've seen the insides of one at the Arbitron lab) so tightly packed nothing could be added or moved... and doing so would cause the device to be rejected.

Again, if the right type of exploits exist, the verification checks can be disabled. The firmware was written by humans. Humans make mistakes. Maybe there isn't enough memory to store hours of audio, but it could still be used as a real-time eavesdropping device while it sits in its dock.

It is half the size of an i-pod, and does not have a hard drive. Or any place inside to add memory. This is getting very "black helicopter." If we have no reports of modified cell phones, this is not going to happen with PPMs since the chances of anyone you are interested in spying on are very low since the panels are not large.

Reports of modified cell phones? Here you go. http://news.zdnet.com/2100-1035_22-6140191.html Farther down in this article is mentioned another eavesdropping method that can and has been used in the real world: OnStar. Another article on this is here http://www.theregister.co.uk/2003/11/20/court_limits_incar_fbi_spying/

All I'm saying is that if it can be abused, someone's going to find a way to abuse it. Do I carry a cell phone? Yes. The difference is that I'm allowed to inspect and modify my own cell phone which isn't the case with the PPM. There's a reason I don't view OnStar as an asset (and in fact I would never consider buying a car with it). Think about it. The fact that you can call OnStar and have them shut down your engine if the car is stolen or unlock your doors if you lock the keys inside means that anyone who calls OnStar and can convince them that they're you can shut down your engine or unlock your car doors. Or, anyone who can gain access to OnStar's system for that matter. I'd rather have to pick the lock or call a locksmith if I ever lock the keys inside than let someone in a call center 1000 miles away have the ability to shut down the engine of my car. The scary thing is that OnStar systems could be used in this way even if the owner hasn't subscribed to the service. The only way to avoid this is to physically disconnect the power and remove the components from the vehicle.

 

[davideduardo]

This assumes that
1) the programmers who made the firmware didn't insert trojan horses into their code that would allow the system to be modified without sounding an alarm, and
2) the firmware contains no exploits that can be used to modify it without sounding the "alarm".

SInce the whole process is audited and accredited by the MRC, the chances are minimal.

Then the system has to report back to Arbitron. As soon as it connects to a public network to send its data, it's exposed to potential hacks from 3rd parties. It has been shown again and again that people make mistakes when writing code. Exploits are constantly being found and fixed.

The docking station calls a telephone number. The data is not sent on a public network.

How can the panelists be kept secret when they go around in public wearing the PPMs?

Only people in direct contact with a panelist would know that. As for anyone trying to blanket hack the system, they could not identify multiple panelists with any ease.

Again, if the right type of exploits exist, the verification checks can be disabled. The firmware was written by humans. Humans make mistakes. Maybe there isn't enough memory to store hours of audio, but it could still be used as a real-time eavesdropping device while it sits in its dock.

The "ET Phone Home" function is similar to the way TiVo updates. The docks know where they are and they only send a very compact set of data to arbitron.

Reports of modified cell phones? Here you go. http://news.zdnet.com/2100-1035_22-6140191.html Farther down in this article is mentioned another eavesdropping method that can and has been used in the real world: OnStar. Another article on this is here http://www.theregister.co.uk/2003/11/20/court_limits_incar_fbi_spying/

In this case, they exploit public networks. The dock talks by wired in home phone lines that are identifiable.

All I'm saying is that if it can be abused, someone's going to find a way to abuse it. Do I carry a cell phone? Yes. The difference is that I'm allowed to inspect and modify my own cell phone which isn't the case with the PPM. There's a reason I don't view OnStar as an asset (and in fact I would never consider buying a car with it). Think about it. The fact that you can call OnStar and have them shut down your engine if the car is stolen or unlock your doors if you lock the keys inside means that anyone who calls OnStar and can convince them that they're you can shut down your engine or unlock your car doors. Or, anyone who can gain access to OnStar's system for that matter. I'd rather have to pick the lock or call a locksmith if I ever lock the keys inside than let someone in a call center 1000 miles away have the ability to shut down the engine of my car. The scary thing is that OnStar systems could be used in this way even if the owner hasn't subscribed to the service. The only way to avoid this is to physically disconnect the power and remove the components from the vehicle.

I can't see why someone would go to the expense and trouble to spy on one person or one panel family by trying to modify PPMs when just bugging the home would achieve the same purpose. The chances of knowing that someone you want to spy on was a PPM panelist are minimal, and panelists are not permanent participants... they roll over.

The only real security issues involve radio stationst rying to influence ratings... and the MRC is satisfied that these issues are handled by the software that, just as with the diary, detects unusual listening patterns and triggers contact with the respondant.