Schneier on Security

Clive Robinson • February 26, 2010 7:40 AM

@ Bruce,

You note that the Dubi suspects did not where hats and glassess or other items that hide the base facial features that “facial recognition” software is supposed to use.

It is one of those things from the “Al-Mabhouh” killing that has not actually been talked about.

It is important because if facial recognition is not reliable (and the suspects obviously thought so) then the next round of Tax Payer Billking involving back ends for CCTV systems is about to start.

Along with unmaned passport control points.

Clive Robinson • February 26, 2010 9:10 AM

@ Tim,

‘Isn’t their value in simply letting us know what happened, and providing a record of it? I.e. preventing the “Oh shit some crime happened here – I wish I had some kind of video recording of what went down.”‘

If that is the only use for CCTV then they are not even usefull for that.

They usually are of such poor quality that you cannot usualy tell the sex of the person when full face or what ethnicity they may be.

They do not record sound and the poor quality is usless for lip reading.

Thus a very large part of what went into the start of an incedent is missing.

Also they are very rarely used in court in the UK as evidence even supporting evidence which suggests that the CPS (UK Criminal Prosecution.

Of more interest is the quality of “police cams” those that officers carry on them or in their vehicals which are in some cases better than home video and very close to broadcast quality.

Invariably these appear to be used for “World’s Most loony driver” or for evidence when a police officer is attacked.

As a lay person you would get the very clear message that it’s one effective system for the Police benifit and one usless system for the people who pay for it.

In practice I suspect part of this is due to the usuall when, where and how the systems where installed and how they are maintained.

If you look around to find the information you will find that in reality the cameras and the recording equipment is at best 10% of the cost of these systems.

It is worthwhile looking at this because you can easily identify a vast amount of poor planning and thus waste.

You can easily see from the bulk figures Bruce gives for London (5e+5 cameras 7e+8 USD) of 1400USD/camera is very high.

Even so I think Bruce’s figure may be on the very low side of reality (ie based on Gov fiddled figures).

The last time I looked at it with independent figures from within the industry (not Police or Gov) it worked out at 1500GBP (which at the time was nearer 2500USD).

To give your eye brows some excercise 😉 for that sort of money and in those sort of quantities I could buy 8-12 inch telescopes with ultra low light CCD 1024×1024 sensors with wide area network interfaces.

Alternativly I could go looking on the web and find cameras of similar quality to the “Street CCTV” for less than 20USD and much higher quality TCP/IP cams with 100baseT connectors for less than 50USD.

So Using Bruce’s figs (1400) and being generous on the camera cost (100) you have 1300USD of other infrastructure and instalation costs.

To identify the waste we also need to remember that most of those systems have very proprietry analogue or digital comms to get the pictures back to the command/control center.

Thus if you were to look at a 100baseT camera it’s invariably “full duplex” thus 100meter Ethernet “time based” distance limit does not apply. Just the cable crosstalk and loss limit which is nearer 500meter with reasonable quality cable (I’ve actually had an IP camera powerd localy to it, send reliable data through two 300meter drums of cable).

So you could setup a 16 camera system (1600USD) at upto 300meter (100USD/drum) from a switch (240USD) using POE (500USD for frame and plates) connected to a computer (1000USD) used for storage / backup to provide pictures almost anywhere in the world.

All for a basic starting price of less than 5000USD or around 300USD/camera these days. Which is a lot lot less than 1400USD.

But this does include other infrestructure it would be nearer 2000USD for just the cameras and recording equipment giving a generous 125USD/camera (which is nearer 8% than the 10% I indicated originaly).

However unlike the “custom” systems currently installed around London the cost of upgrading in place of such a system over time would be minimal as the cables etc would not need to be changed just the cameras and computer.

Which probably indicates why electronic security is such big business in the UK.

Clive Robinson • February 26, 2010 9:20 AM

@ Anon.,

Would you care to back up any of your statments with even UK Gov figures?

Last time I looked the UK Met was basicaly saying with the best interpretation the video footage was about equivalent to “not worth it” when compared to conventional investagative effort.

Also various police officers have been talking about CCTV as “anti-PR”.

That is if there has been a crime Police knocking on doors “walking and talking” is not just pro-active at bringing in other information it actually lets people see crime in their area is being addressed. Where as time spent viewing video footage not only produces less usable evidence than the “foot time” it is unseen and locals think that crime is not being investigated.

Clive Robinson • February 26, 2010 12:34 PM

@ Jason, John D,

“If the vandalism stopped after installing cameras, who’s to say installing fake cameras that record nothing but have a lens and a blinking LED wouldn’t have accomplished the same thing with much less cost?”

I suspect it would.

The reason being is the type of crime being committed.

John you call it “our university computer labs were being consistently vandalized: mostly petty, e.g. coke poured onto the keyboards, balls stolen out of the mice, etc.”

Or “petty vandalism”, it probably was not “vandalism” at all. I worked at a Uni where there where several labs and yes there was some of this going on.

When we looked into it we found that what the students considered the “better labs” where virtually un-touched. We did a bit of “covert” survalence on the lab getting most abuse to see what was going on.

It quickly became clear it was down to accidents and mindless bordom.

The computers in the lab where often (glacialy) slow and you could see the students sit back and pick up a drink or start fiddeling with a mouse etc.

On reporting the findings the managment decided to put in cameras (apparently cheaper than upgrading the computer network…).

We started to get different behaviour not spilt drinks or mouse balls missing but broken keyboards and damaged cables, and much more rubbish and litter than before.

The keyboards and cables where getting broken by “frustrated behaviour” you could actually see students literaly “pounding the keys” in frustration and yanking the mouse etc around. And importantly they complained much more about the “crap facilities”.

All the cameras had done was change the behaviour and it’s severity not stopped it.

Oh and the better labs without cameras carried on not needing cameras only more computers as the students went there rather than the “big brother lab” which was rapidly becoming scruffy and very unreliable.

Sometimes you need to ask “why” a problem occurs rather than put in obvious survalence to frighten people, it just adds to the problem.

Some councils in the UK are investigating this by improving the suroundings fixing the pavements putting in seating and good lighting, allowing pavement tables and chairs outside cafes and pubs etc. They still have CCTV but it is much more discreat and blended in with the suroundings.

Some reports indicate they are seeing drops in crime but unlike “just CCTV” it does not appear (yet) as short term in results. Also they are reporting more people being on the streets actualy sitting, shopping etc not hurrying through. Also the amount of litter is down in some areas and the bins are not being set on fire etc.

Clive Robinson • February 26, 2010 5:11 PM

@ Anon.,

‘”Cameras can help solve crimes” – “surveillance footage is used as evidence in convictions” – what’s to back up? Both are trivially obvious.’

Sadly what appears “trivially obvious” in practice is not.

For instance you say,

‘Surveillance cameras are useful after crimes have been committed to supply evidence.’

It depends on your definition of “evidence” most CCTV footage is not admisable in UK courts as primary evidence for many reasons. Even as secondary or supporting evidence it is usually only of use to confirm other addmisable evidence.

Thus it is actually of little direct use.

If however you mean “evidence” in a less strict sense that is to provide the police with leads then yes occasionaly it is of use.

Which is what you probably ment with,

‘Footage can provide clues that more quickly lead investigators in the right direction before other critical evidence is compromised.’

Usually (in the UK atleast) very little evidence is gathered from your average crime scene in fact you will be lucky if they even dust for prints these days, let alone photograph anything.

Usually the only thing they collect is forign objects that can be clearly identified as having been brought to the scene of the crime (such as dropped tools cloathing or other similar items).

For instance in the case of robbery in a home where the burgler has clearly been consuming food or drinks and have left very clear bite marks or saliva they won’t bother with them only clear fingerprints on an otherwise clean (of fingerprints) objects.

It is only in serious crime (by value or violence) or one that is potentialy going to have high press mileage do the generaly put in any real forensic work (and this is becoming less common).

So called “CSI myth” which is well documented is actually hurting trials as juries falsly believe that each and every crime will have unarguable forensic evidence.

The simple fact is there is not the money to gather let alone process forensic evidence from 99% of crime scenes. And way to many “first responders” have the training to preserve crime scenes from contamination. This has actually got worse with the introduction of Community Support Officers (CSO’s).

Thus you will find in the UK the police have found CCTV footage to be usefull in around only 1 in 13 cases where it has actualy caught the crime in progress in sufficiently clear detail. In the other 12 they got the same or better evidence faster from more traditional methods.

With regards to,

‘In a TV-world, people “understand” CCTV and that kind of evidence resonates with juries – leading to convictions.’

As I noted above (CSI Myth) this actually raises juries expectations way way above what is actually achivable, and when they see grainy footage with light bloom turning people into ghosts because there is no auto iris on the camera they are less than impressed and defense barristers usually have little difficulty convincing a jury that the person in the footage could be one of a thousand others.

Which is why CCTV is generaly only used in court when the identity of the criminal is not in question. Usually to show how calous or ruthless or sensless an attack was or to stop the defence barrister claiming the deffendant was provoked or defending themselves (ie not to show guilt but to get longer sentances on conviction).

As a general rule the low definition of CCTV cameras (~300×400) requires a full on human face to fill about 1/8th of the screen area for reliable identification.

Cleaning the image up is tampering with evidence and it is generaly not allowed for court presentation except within very very strict guidelines.

Also apart from J.Random nutter who takes no precaution to protect her identity most street criminals and burglers wear baggy cloaths with hoods and baseball caps and look down making CCTV effectivly usless even their footwear is often of the wrong size…

It is why I was very very surprised by the Dubi footage by normal CCTV standards it is not just good it’s out of this world and the hotel in question must be using the newer generation of digital cameras that effectivly take 20 or so “high res” still photographs each second as opposed to the 1990’s CCTV analog low bandwidth therfore low res with no processing.