Thursday, December 18, 2008

Thoughts on Melendez-Diaz: The Product of Machines

One of the issues addressed by some of the cases dealing with lab reports, and in the Melendez-Diaz argument, is the status for Confrontation Clause purposes of the output of the machines that do the testing.

As some courts have noted, the output of a machine is not in itself an assertion that can come within the scope of the confrontation right, or of the hearsay rule. Nor are the raw data. But raw data – the numbers themselves – are meaningless without being linked to the case.

Let’s say Officer A brings to the lab a sample labeled “Melendez.” Lab technician B takes the sample, puts it into a machine, and runs a test to determine whether it contains cocaine. When the test is completed, the machine spits out a piece of paper bearing only test results that an expert can testify indicate the presence of cocaine, without any identification of the sample tested.

Now suppose that the prosecution wants to prove its case based only the testimony of Officer A, the printout, and the testimony of an expert that the numbers on the printout reflect the presence of cocaine. This is plainly inadequate: We have sufficient proof that a sample linked to the defendant was brought to a lab, and a piece of paper bearing numbers that, the expert’s testimony indicates, would, assuming they are the results of a test, demonstrate the presence of cocaine in the tested sample. But we don’t have proof that any test was actually performed on the sample linked to the defendant or that the printout reflects the results of that test.

So let’s say technician B wrote an affidavit stating, “I took the sample Officer A gave me, put it into our machine, and performed a prescribed test for determining the presence of cocaine.” I will call this the input proposition. B’s statement of the input proposition seems plainly testimonial; it is made with the understanding that it will very likely be used in prosecution of a crime. The affidavit should not be admitted unless B testifies to the input proposition subject to confrontation, at trial if she is then available and otherwise in some other setting such as a deposition.

Even with B’s live testimony that she performed the test, the prosecution’s case is incomplete absent proof that the particular piece of paper presented is in fact the printout of the machine after performing the test on the sample in question. I will call this the output proposition. Proof of the output proposition could be supplied by B’s testimony asserting it – but an affidavit stating the proposition is testimonial and should not be admitted absent confrontation.

Now suppose that instead of writing an affidavit asserting the input and output propositions, B types into the testing machine an identification, such as “Melendez,” and when the machine spits out the test results it includes that identifier at the head of the page. Doing so does not render the printout admissible. In context, typing the name amounts to an assertion of the input proposition. It also amounts to an assertion of an altered form of the output proposition – altered because the printout does not exist at the time the assertion is made, so the assertion is that the printout that the machine will produce will be a product of the test on the sample in question. So the fact that the printout generated by the machine bears the name of Melendez does not obviate the need for B to testify. I do believe that incorporating the identifying information on the printout does relieve the prosecution of one potential burden. Suppose B does not stand continuously by the machine during the entire interval between the time she puts the sample in the machine and the time the machine generates the printout. Then absent an identifying mark on the printout she could not say that the printout is the one generated by the test on the sample in question. But if she types in the name, and the name is on the printout, then even if she did not see the printout being generated, I believe she has can testify that this printout is the one for the sample in question: Given the way the machine works, she has effectively tagged the printout by typing the name in.

In the actual Melendez-Diaz case, it was not a machine printout at all that the prosecution offered, but a certificate based on the machine readings. I think it is very clear that this certificate, obviously prepared for prosecutorial use, is testimonial, and the fact that it reports machine readings does not diminish this fact. Even more clearly than the printout bearing an identifying marker, the certificate is a human assertion prepared for prosecutorial use. Obviously, it asserts the input proposition – that the sample in question was submitted to the given test. And it asserts a broader form of the output proposition – not that a machine-generated printout bears the results of the test, but that numbers as described by the writer are the results of the test. The certificate cannot be admitted consistently with the Confrontation Clause unless a witness with personal knowledge – presumably but not necessarily the author of the certificate – testifies to both propositions.

A recent opinion usefully summarizing the decisions on this matter, and coming to the proper result, is United States v. Crockett, 2008 WL 4937029 (E.D. Mi. Nov. 14, 2008). The opinion does not go into as much detail as I have here, but I think its analysis is consistent with that presented here.


Anonymous said...

What if the technichian is only given a number along with the substance he is to test, for example, "sample #1a." He does not know from whom the item was seized or whether it is to be used in a future prosecution. Would the output of the machine, and an affidavit of the technician stating how he performed the test on sample #1a, still be testimonial?

Mitch Ignatoff said...


This assumes that the machine is accurate. When the machine contains a computer we are asuming that the design of the machine and it's software are accurate. This is especially so if the machine does it's own accuracy testing.

What if we want to challenge the accuracy of the machine? Here I think we have to distinguish between machines produced generally for all kinds of industry and machines produced for litigation, such as breath testing machines for alcohol and radar guns for speed measurement. These latter machines are made almost entirely for the enforcement of the criminal laws. Under Palmer v. Hoffman and what I expect to be the decision in Melendez-Diaz I think the state has to produce a witness with first hand knowledge of the design and the software of the machine to prove the machine accurate. And this witness will not have first hand knowledge of the particular machine. We are assuming that the machines do not differ in performance.

Justice Kennedy's concern with the impact of holding that the lab tech makes a testimonial statement when he reports the results of his test-that the test was made in anticipation of litigation- is reason for concern. I know that in California the state must produce a lab tech in drunken driving prosecutions to testify to the accuracy of the machine. Melendez-Diaz could make this a requirement for every state which forces the states to hire more techs and train them so that they have personal knowledge of the design and software of the machine that was made in anticipation of litigation.

Another answer to Justice Kennedy's concern-which seemd to be shared by Justice Breyer and Ginsburg-is that the confrontation clause is absolute. There is no balancing test in the clause.

I'd like to know your thoughts.

Mitch Ignatoff

Richard D. Friedman said...

In response to the comment by an anonymous reader: The situation doesn't change whether the identification is by name or by some tag such as "Sample 1a". There has to be a chain of custody -- if the officer who brought it in labeled it as "Sample 1a" and can testify that that sample is the one he took from the defendant, and the lab tech can testify that she identified the sample as "Sample 1a" -- or whatever other label she used -- to the machine then that's good enough. An affidavit to that effect would clearly be testimonial.

On Mitch's point: I think the Court can, and should, resolve Melendez-Diaz in favor of the petitioner without resolving or squarely addressing the issue he raises, which is an interesting one.

When one testifies to the time of the event based on what a wristwatch said, we don't demand that a witness testify as to the inner workings of the watch. Is the result different if one testifies to the output of a machine designed principally or exclusively for litigation? I don't know that it is.

I understand that the argument is that whoever designed the machine or the software was effectively making a statement to the effect, "If the results of a run on this machine are X, then this indicates Y," and that this statement was intended for prosecution. Let's assume that the design should in fact be considered a statement, and let's also assume that a statement geared for multiple prosecutions, of crimes not yet committed, is testimonial. (That's an issue that arises when prosecutors offer a certificate that a machine was in good working order.) Even given these two pro-accused assumptions, the fact remains that the prosecution is not presenting, or at least need not present, that statement. That is, the witness who testifies as to the result presumably does not testify, "The person who designed this says it works in this way . . .," or, "The designer set up the machine like this because it works in this way . . ."

So on what basis could the jury conclude that the machine is accurate? Here are three possibilities:

(1) As Mitch suggests, the tech might testify to how the machine works. If there are lots and lots of these cases -- and only if there are many is there a significant problem -- then techs can be trained efficiently in how the machine works and in how to explain that. This does happen in some settings, and I don't think it's cost-prohibitive.

(2) Perhaps the workings of the machine or of the process are a mystery either to the tech or to everybody -- but if it works, it works. That is, if the tech were able to say, "I don't know [perhaps no one knows] how this works, but I can tell you that I've tested this machine [or process] and it works," that's good.

(3) Arguably the prosecution does not have a substantial burden here. Suppose a prosecution witness testifies, "It was 23 degrees," based on a thermometer on a bank wall. I don't think that as a constitutional matter the prosecution bears a burden of presenting evidence that the thermometer was in proper order. It seems to me that, as a constitutional matter, the state can provide that it's enough for the prosecution to demonstrate that this was a thermometer that had been in operation in the public for some time -- that would support an inference that it is probably working reasonably well -- and that the defense may, if it wishes, challenge the proposition that it was working properly. Perhaps the same should hold for a radar gun or a drug-testing machine that a police department or lab says they use regularly. Interesting issue. Not presented by Melendez-Diaz, not one that the Court has to resolve to decide the case.

No, the Confrontation Clause does not balance -- just as we don't tote up the pluses and minuses of providing counsel or having a jury trial in each case. But holding for Melendez-Diaz does not compel the Court to impose intolerable expenses on the states.

Mitch Ignatoff said...

Thank you professor.

The reason why I keep pushing this argument is that there is substantial scientific evidence that breath testing for drunken driving is not that accurate. See the website of Dr. Michael Hlastala. The theory on which breath testing is based appears to be wrong. And when guilty and innocent is often seperated by .01,
it matters.

Mitch Ignatoff

Anonymous said...

machines are many types,electronic machines,diesel machines etc.that's very impotent for our

EOT Cranesevery work do with machines.