If you want to fuck up your reference sample try wanking when you phone. If you can time your vinegar strokes to coincide with you answering your name you’ll be fine.

If it makes things more easier I’ll just point out that all call centre staff are young, horny and up for it.

Don’t agree.

Woo is the misapplication of science as much as it is the non application of science. Or, if you prefer, coming to a conclusion based on dodgy, sciency, evidence. Ben Goldacre is very good on this. As are his nemesis, the drug and cosmetic companies’ PR wings. Ben is a good guy, IMVHO, whereas, well, you fill in the details….

Here is an example of a Local Authority trying to sound, err, authoriative, and getting taken to bits:

http://tinyurl.com/5htsn9

Which is exactly where we are with VSA. It is clearly wrong to link stress and lying. They are not the same things. And it is clearly wrong to deny the gross failures of the technique or the apparatus to actually measure anything beyond random chance.

Anyway, if what you say is correct, then we are being comprehensively lied to. It seems to me that we need a double blind test of this product in exactly the same way as we would test a clinical drug. So, why aren’t we?

1. What it is, is a piece of moderately interesting science, albeit one of very limited application – and actually an offshoot of some of the very early efforts to try an understand conditions like Parkinson’s Disease – one which has been grossly misapplied to a purpose founded on a false premise.

Woo is stuff like homoeopathy and astrology which has no scientific foundations whatsoever and science is science and can either be correctly or incorrectly applied, the latter being the situation here.

2.This is NOT ultimately a scientific debate because the decision being challenged here is not a scientific decision but a political one.

If it was simply a matter of science then there’s no way anyone would consider using these systems, but as I noted in my previous comment, what the government is interested in here is not the scientific validity of the system, other than in the limited sense it provides a modest level of cover for the decision to use the system, but a very simple bottom line – how much money does the system save by propelling people off benefits.

That alone dictates that a straightforward challenge to the scientific validity of the system in extremely unlikely to work, unless you can convince someone like the Public Accounts Committee that the lack of scientific evidence makes the system a waste of money, which is a tough ask as long as it appears that its saving the public purse more than it costs to use it.

One of the political dimensions to this issue which hasn’t been raised yet is that, on the last published figures, the government spent £154 million on anti-fraud measures and investigations related to benefit claims but only saved £106 million as a result of successful investigations. Currently, the government is spending 50% on combating benefit than its saving by detecting fraud while using this system in Harrow, on paper, saved between three and five times as much in withdrawn claims than it cost to put in the system.

As far as the government is concerned, the system works, because it appears to save them considerably more money that it costs to use the system – and beyond they couldn’t give a toss about the technical arguments…

…unless you can pin them on a discrimination charge.

You have to have valid legal grounds to mount a judicial review and, sadly, the mere fact that a piece of technology fails to do what its supposed to is not enough to get a case into court, unless you can persuade the government to sue the developer for selling it a pup.

In theory, an individual facing a prosecution for benefit fraud after being ‘fingered’ by this system could try to challenge the legal validity of the prosecution based on the fact that the trigger for the investigation derives from ‘evidence’ that is inadmissible in court. But you’re hardly going to be batting on friendly wicket by putting up an alleged benefit fraudster at the spearhead of efforts to invalidate the use of the system, which still leaves you to find a valid legal basis for challenging its use – which is where that list of confounding factors and the fact that several of them make the use of systems highly questionable under things like DDA and, potentially, the Race Relations Act, comes into play.

If that system cannot process voice data provided by a disabled subject, or by a subject for whom English a second language, on exactly the same basis that it would any other subject, then you have a discrimination claim that, if successful, would force the government to either pull out of using the system or find some means of screening claimants and discarding those whose disability or ethnic background is a confounding factor.

The first outcome is a straight win, the second would massively increase the administrative overheads and costs that using the system entails and almost certainly kill is a viable proposition – either way you skin the cat.

That’s what’s emerging from this debate, a valid legal basis from which to challenge the government’s decision to roll out this system.

Whether it does that in fair manner is, to a significant extent, immaterial in their eyes and, joking aside, if some of these bastards really thought they could get away with rolling a D8 and investigating people at random then they would – its only the likelihood that such an arbitrary approach would land them with a shedload of bad press and an awkward judicial review under HRA that would hold them back from such a move.

OK.

I am quite willing to argue beside you on this anti-science, anti-authoritarian platform.

It is just a bit disheartening that we have to do it, don’t you think?

For goodness sake!

From ‘Assessing the Validity of Voice Stress Analysis in a Jail Setting’ , page 89 – conclusions:

These findings add to the growing literature on tests of voice stress analysis theory and devices. Even though early tests of the “theory” suggested that stress is related to measurable changes in voice patterns (Cestaro 1996, Smith 1997, Hansen 1996, Hansen and Zhou 1996, Haddad, Ratley, Walter and Smith 2002) it is not clear that VSA devices are able to distinguish stress from efforts to deceive (Haddad et al 2002) We are unable to find any peer reviewed and published studies that showed significant support for the effectiveness of VSA software to detect deception. All previously published research conducted in a lab setting has failed to find for VSA theory or technology (Brenner, Branscombe, and Schwartz 1979; Cestaro and Dollins 1996; Hollien, Geison and Hicks 1987, Horvath 1978, 1979; Janniro and Cestaro 1996; Lynch & Henry 1979; O’Hair, Cody, Wang and Chao 1990; Suzuki, Watanabe, Takeno, Kosugi, and Kosuya 1973; Timm 1983; Waln and Downey 1987) Some researchers have tried to test VSA products in “the field” but with limited success (Palmatier, n.d.: 1999, 2000). Our research therefore complements previous research by failing to find support for the VSA products in a real world setting.

I apologise to Messrs Damphousse, Pointon, Upchurch and Moore if there are any typing errors in the above quote. However it is only available in pdf, which is another governmental nonsense.

So, Unity, what have you and your chums added to that debate?

All that Lippold proved was that this subsonic component of voice was a fight or flight response.

Which applies equally to the innocent as to the guilty.

I stand by what I said. This is woo woo science.

What might be useful is getting this ridiculous idea confronted, right now, in the judicial system. As both you and Mike have pointed out, this is a lottery. The fact that politicians take it seriously suggests that it is not only Washington that is crawling with vested interests.

If you’re serious about challenging this then the ‘it’s just woo’ argument doesn’t fly – we have put the evidence base behind any challenge and that evidence base has to acknowledge that there are at least some valid scientific foundations.

The Lippold tremor is real, and I’ve read several of Lippold’s original papers, the earliest of which I’ve been able to track down dates to 1957. Likewise, the underlying technical premise of the system is real in the sense that it ‘works’ using the same base technical methods as any other voice recognition system – and its that which ‘elevates’ this system from being just mere woo to its being the misapplication of science.

Unless you acknowledge that element then you lose the argument straight away. You can’t just dismiss that component of the debate, you have to explore and expose its limitations, giving the context that the company behind this system is trying to obscure in order to sell it to the suckers in government.

You need to realise that, so far as the DWP is concerned, this system works simply because it delivers an outcome they want – a reduction in the number of benefit claimants – anything else is gravy.

Whether it does that in fair manner is, to a significant extent, immaterial in their eyes and, joking aside, if some of these bastards really thought they could get away with rolling a D8 and investigating people at random then they would – its only the likelihood that such an arbitrary approach would land them with a shedload of bad press and an awkward judicial review under HRA that would hold them back from such a move.

So, we need a solid counter argument and of the ones open to us, by far the strongest isn’t the ‘it doesn’t work – full stop’ argument, its actually the extensive list of confounding factors that would compromise the system even if it did work because many, if not most, of these cut across statutory duties to promote equality and are, arguably, discriminatory –

- and in the context of this debate (and knowing how the government think) there’s the focal point of the attack.

If the system can be shown to be more likely to generate false result when fed a subject who has a speech or cognitive impairment or has English as a second language due to its technical limitations – and I think it can – then you have legal basis to challenge its use.

We have to use the system we have to get after this, and unless the Commons Public Administration committee takes an interest in the question of whether the government is spending money on a bit of technology that doesn’t work, then the best route through this is likely to be to push for the Commission for Equality and Human Rights to go after this as a discrimination issue or find a suitable plaintiff and a human rights lawyer who’s prepared to mount a HRA challenge pro bono.

Unity. This post at least made sense.

It acknowledges the simple fact, that lie detectors are shite.

Could I suggest to you that the challenges to lie detectors have a good and honourable history? Could I suggest to you that most effects, are in fact interperable in a number of ways, and that there is little or no peer reviewed support for this woo woo?

You’d seem to accept that, now.

As you are much more connected that I am, could you please tell your political masters that this is, indeed, the piece of shite you and I say it is?

It is, by the way, a form of psychological torture. As it is not evidence based it is an attempt to play God on the part of the inquisitor. With confessional results that have nothing to do with reality.

Such is stupid.

In cost-benefit terms, if the fraud rate is established (let’s say it’s 1%) would it cost a local authority £63,000 to promote a private Bill empowering it to reject 1% of claims at random? After all, we all hate claimants, so there couldn’t possibly be an ethical objection.