Breath-test ruling jeopardizes thousands of state DWI cases

I don't understand why the source code is needed. What would be needed in my understanding would be the a statistical analysis that shows the results of the breath test match those of a direct measurement blood test to a certain confidence limit.

 

They have to be able to test the algorithms to make sure they are accurate...

DD

 

Exactly, not sure what having the source code is going to prove.

Its not like they have code

if(suspectname == "JOE")
{
BAC = .2; //suck it joe
}

The part that could possibly be flawed is the hardware sensor. But the source code wouldn't prove anything about that.

 

What acceptable error rate? Hasn't everyone just assumed that these things are correct? And isn't this what the defendant is arguing against?

 

Algorithms don't matter if another piece is faulty. It is the total package, the results that count. The machine needs to be calibrated the same way any other lab machine is with QA samples. I don't care if the breath test is a monkey in a box provided he provides quantifiable accurate and reproducible data.

 

Every piece of technology has an error rate. Nothing is flawless. Leave it to the scientists to determine what the error rate should be for something to be scientifically accepted. That can be done in clinical trials without the need for the source code.

All of that being said, I never stated definitively that the source code should necessarily be out of bounds. I have stated over and over again that it is the federal courts that should decide being that intellectual property law is a creature of Congress.

 

The software calculates the data that is input by the breath....I think they have a right to see how those calculations work...what if they are all screwed up?

What are the calculations based upon? How do they measure a person's breath? What about body size, what if someone burps, does the size of a person's throat and exhaling ability make a difference?

There are TONS of factors, how do they use the same machine to test each person, and how can you be sure the data is right?

The accused have a right to defend themselves.

Software is only as good as the people who wrote it...

DD

 

I was thinking the same thing. Who cares about the method as long as the result is correct? I mean, unless there's suspicion that the company is doing something illegal inside those breathalyzers, why should they give out the source code?

I don't know the error rate nor the standard deviation for breathalyzers. If they're not reliable enough, then the cour should simply stop using them ala polygraphs. But if they are reliable enough, then there's no reason to ask for source codes.

 

How do you determine if they're correct? People have always just assumed they are.

 

Well, I assumed the government ran tons of tests to make sure it's reliable enough before using them. If not, then I don't know WTF these things are doing out on the street in the first place. Because they should just be thrown away.

Surely you can have an outside company run tests on all the equipment and personnel to match its quality vs. blood tests. None of this require actually digging out the source code for.

 

"DRUNK!!! DRUNK, Will Robinson!!!"

 

Assuming the government gets anything right, is a MAJOR assumption.

And I would not want my guilt or innocence decided by an assumption would you?

DD

 

My friend beat a speeding ticket by asking the cop when the last time he calibrated the radar gun and how far from the vector he was when he got him. The cop stammered something like 10 feet off the road and was then asked the correction factor. He had no answer, ticket dismissed.

 

I always see stories of this and books/websites on how to beat them. Is this true? Do the cops actually have to answer you?

 

Not sure about on the street at the time of the ticket, but if it's in court, yes.

I know a girl who got out of a ticket by using HS geometry to prove that the cop couldn't have clocked just her from his position (distance and angle) and the direction she was heading. It's incredibly simple, but you have to have the right parameters.

 

Ah, for some reason I read it like he did it while he was getting a ticket. That makes sense.

 

Statistical analysis would seem fine to me. However, at this point, can the company realistically refuse to share the source code if keeping it secret means that no US police force can buy their product? There are plenty of opportunities for compromise here. The courts could allow an alternative method for verifying accuracy; legislatures could make some new rules around it; police forces could use other tests; the company could release the code.

 

Yes, there are tons of factors that go into geting the results, but all that matters is if the output of the machine matches direct measurement of blood alcohol level. Comparing results of the machine to direct measurement is how you know the data is right.

The only caveat I would have is that the initial calibration tests would need to be done for different height/weights and/or other inputs that is needed to give an accurate conversion to BAL. I would think it would be very difficult for a breath test to be acceptable for Yao Ming or Mini Me.

 

Trying to figure BAC through a breath test seems like indirect way of measuring it so there is bound to be some amount of error in it.

 

How can a single machine be calibrated to give an accurate reading for every single person on the planet?

You stand on the side of the road and blow, the officer looks at the read-out and the read-out gives the supposed blood-alcohol content. They don't adjust it for height, weight, breath/blood ratio.

So if there are differences between individuals, those differences have to be accounted for in the software..... and yet, we don't know how those differences are accounted for.

Honestly, that software would have to be amazingly sophisticated (bordering on magical) to be able to pick out differences in breath/blood ratio simply by having someone blow into a tube. I presume that the calibration doesn't take into account such differences, and it would be interesting to discover how far off the machines really are, on average.