Hearing - Tues: Davis Direct II
Filing back in. Brenna was looking unhappy in the hall; he doesn't have the poker face of the attorneys.
SUH: given you copies of the batch sequences; also in folders, but easier to follow this way.
q: USADA 155 seen?
a: yes.
[more]
q: and subsequent pages, the individual processing results for each part of the run.
a: the operator has put 14 bottles in the machine, and the machine has controlled the robot to run all 14 in a row. Results all should be from that run. If it stops, it was a manual intervention. No summary sheet if run is halted.
q: and the rest are supposed to be the pages corresponding to those steps. Conclusiong about those pages?
a: not all were printed as part of this batch run.
q: which are not?
a: look at USADA 157, top block "batch name", the batch under which the sample was run. The front pags was 23016, on this it is empty. So this paper is not from conte mporaneously with the batch run; also 160, 163, 166 and 169 172 and 177, where the batch name has re-appeared.
q: so 177 is the first time the batch name appears in a run with the summary on 155.
a: correct.
q: keep going
a: [ lost page counts ] some in batch, some not.
q: on 155 there are values in the summary, page looks like it got cutoff.
a: os2 problem
q: did you try to match summary results with subsequent pages.
a: the numbers are different.
q: so there's stuff in the pages that is not in the batch?
a: we don't know where it's from.
q: find the pages that are supposed to match...
a: mixcal 003... there are different numbers...
q: USADA 179 and USADA 155 on screen
[ Mclaren is looking perplexed ]
q: I can't see the numbers at all. They're different. This is the result, and this is supposed to be the summary. But they have different values.
a: correct.
q: do you know where the data from USADA 179 comes from?
a: ???
q: other examples of data not matching summary sheet?
a: yes, it goes all the way through.
q: Let's look at the B, USADA 331, which appears to be the batch summary.
a: yes.
q: looking at the following pages, supposedly supporint the summary, what pages don't match.
a: some do not appear to be part of the batch. 2nd mix cal irms 358 miss.
q: so the data doesn't match.
a: nope.
q: describe what reinjection is?
a: injecting again and again.
q: are the logs consistent with re-injecting?
a: yes; we don't generally know, but it appears they are reinjecting samples and standards.
q: what would that mean, isn't that ok.
a: It's ok if documented in a proper chain of custody -- even running someone eles's samples in the middle. Here we have no idea what is going on. If an engineer came back with a log with gaps like this, I'd be suspicious. Maybe they are in a rush to get home, and pick the best 5 of 10 samples as the contemporaneous results.
q: can you describe manual reprocessing?
a: there are as many strategies as there are operators. It's subjective interpretation.
q: please sit at the computer, and please describe computer.
a: it's very old with very old software, a 486.
q: because it can't run on newer computer
a: right -- because of disk partitioning and floppy install; if processor to fast, software breaks. good for it's time, but it's time was 1987.
q: this is what LNDD has on its isoprime?
a: yes.
q: used for S17 samples.
a: yes.
BARNET: exact same software?
a: same version.
q: walk through how manual reprocessing works -- which peak identification and background subtraction.
Ooo- we're getting a color demo, and the resolution is low enough it's big and visible on the plasmas
McLaren: Botre speaks to software versions.
Botre: 167-4?
a: 167-2 change was change for head amplifier, doesn't affect integration.
YOUNG: objecting -- head amplifier was an issues.
BARNETT: not the exact version. This is why we can't properly do cross.
[ They don't want to see this demo ]
a: we can load 167-2 if you like?
BARNETT: that they didn't do it yet, we shouldn't do it.
BRUNET: we decided we'd do this. If it's 5 minutes, OK.
BREAK
29 comments:
Is brenna still being paid to just sit there every day?
I don't blame him. If Davis is anywhere near as credible in person as his answers appear to be in text form, he has to be killing the USADA/WADA guys listening who have already testified that they believe the LDNN analyses were high quality. He is making EVERY WADA lab rat who stood up for LDNN look extremely bad.
does anybody else wonder what lab they'll be testing at for this year's TDF?
You know, I'm wondering if we can re-open some other cases now that this stuff is all coming out.....
This is incredible.
Am I hearing this correctly??? Is he saying that they cooked the books???
Sort of sounds like they may be going for the tampering argument after all? Is that what this line of questioning - unaccounted for pages - is about? That would surprise me.
And I thought yesterday was big. Sheesh!
-Skip
Chaz, sounds like it to me, but I'm not sure yet. Maybe just more unpredictability, or just more variance from SOP?
It sounds to me like they are saying that they reinjected repeatedly until they got the results they were looking for. This implies that they were manipulating the results somehow.
No kidding USADA doesn't want to see this demonstration! What this really does is give the arbs the information and understanding they need to question the interpretation they get from Botre.
Great preparation - having the other version of the software ready to go if needed.
Looks like BAD prep to me. They should have had the exact version on the machine. They might lose their demo because of this.
I am worried. I hope this isn't crucial.
TBV
There are two ways to look at that. The system is designed to have an automated run. In other words, you stick all the bottles in, and let it go. If they are reinjecting, it means they are interrupting the manual process. That could be harmless and innocent (reinjecting to correct an operator error on the machine like the one described by the analyst in her testimony) or it can be to inject another sample or some other compound. Without a coc of the samples or notes in the log of what was reinjected, there is no way to know if it was sinister or innocent.
The second issue is the batch paperwork. The idea is that everything that is done during a given batch of samples gets printed out as a part of the summary report so that anybody looking at the data can validate the process of what was done and evaluate for themselves why the reinjections might have needed to be done and what the results of the reinjections were. However, the out of sequence batch pages and missing pages suggest that the result of the reinjections and some of the other summary pages were either removed or pages printed manually were inserted in their place. Either way, an outsider looking at the batch summary now has no way of knowing what the analyst did, why they reinjected, or what the results of the reinjection were. Tampering with QC packages like that is a huge red flag for any data validator looking at sample data from a lab. In the environmental sampling world, a data package like that would be thrown out and the samples would have to be reanalyzed if the lab could not provide a full and complete summary package that includes all original pages.
TBV -- i agree! everything was going so smoothly, and then this sloppy mistake.... what were they thinking???
Anon3:57 wrote: If Davis is anywhere near as credible in person as his answers appear to be in text form....
Davis appears even more credible if you can watch the video. Absolutely prepared, absolutely knowledgeable, can take the entire system apart and put it back together again. Excellent witness.
When contrasted with 60 year old lab heads who haven't touched an actual piece of lab equipment in a decade, he is stunning.
Anon 3:57,
He looks very impressive. Another expert confident in his stuff, and can explain things to the point I think I could understand.
Regards,
I think the Dr. already testified that the part of the software which deals with what he is about to deminstrate, reprossessing, is indentical.
Is Floyd's team going to get the 3-4 extra hours offered to them by USADA yesterday? USADA would look very bad if they tried to back out of that now.
Swim,
Right on! As to the printed out pages I suspect that it is more a matter of the techs picking and choosing what they print, similar to their manual adjustments, without being aware of the ramifications. Guessing that it may reflect the instances when the stopped the process, made an adjustment and then continued. The software may have treated as different tests and given different reference numbers. Those numbers might appear on pages they discarded because they were for identification purposes and did not contain data. Pure guess, but I don't think that these government employees had much vested interest in the test results, other than getting the work done and going home.
And, remember, these test result packets were "verified"! has to toss that in.
pcrosby
Alright now you have me worried! I bet the 167-4 would have provided exactly the same information, but presented it better for the panel. Now they have to slog through the older version. I guess they should have asked ahead of time which the arbs would like?
It was good to see that Botre knew which software to ask about. He may be tracking this closer than I was thinking.
Are they back at it yet???
If the 167-2 was run on a 486, it probably will run sloppy on a machine running windows XP to link to a projector. I bet the old version uses more text commands and keyboard controls and less mouse/windows GUI. Probably why FL's lawyers wanted to use the newer version.
Anon 4:30,
When the "government" employees work for the host country of an event and just watched another stinking american win "their race", I doubt any of the samples they run are "just another day at the office". When it came to the B sample, knowing that the sample she was running was the "stinking american's" B sample and that the credibility of their lab and the woman who trained her was on the line, I assume she had plenty of reason to make sure she got the "right" result from her test, regardless of what she had to do to get it. Making the paperwork in the summary package match the right answer is just part of the housekeeping to sweep away the evidence of all the reinjections and manual corrections used to get to the "right result". I have seen EPA descend on and shut down an environmental lab for exactly the kinds of QC problems being demonstrated by Davis.
Manual correction of any analysis is not something that is taken lightly in any lab, and it should be meticulously documented when an analytical run deviates from the standard automated run to which the machine is designed and calibrated. The apparently blase' attitude and lack of documentation of manual deviations from the standard method are stunning, and I am finding it harder and harder to keep reading without expecting to see the WADA people step in and stop the hearing and announce an audit of the lab's practices by outside experts.
Wow. It is obvious to me that LNDD labs made errors. The question is how serious were those errors. From what I have read, the errors were so serious that the results cannot be verified.
The testimnony also shows me that USADA is not interested in finding out the truth....they want to convict Landis regardless of the evidence.
Everyone of the USADA people knew that there was no difference in the software at all in the part that he was using. The part that was different affected the actual machine it ran on, not the reprocessing part.
Their objection isn't that they want to make sure the demo is correct, they don't want the demo period because it will show how easy it is to manipulate the results.
-WDS
Anon 4:23,
I disagree. I worked in a lab for many years doing chromatography. If a run is not completed because of a problem the correct way to address it is to re-run the entire sequence of samples.
Something of note is that the really, really bad chromatograms all seem to be the ones from the most recent testing where they had lots of witnesses. If they did rerun anything to clean up their results on the original A and B, they couldn't do the same on the newer ones.
I trust Davis when he says he doesn't think they were doing it to deliberately get false positives, if for no other reason than the fact that it appears they wouldn't have known how to intentionally jigger the results. But that still doesn't mean the results they did get are correct, or that they didn't rerun things a few times to get "clean" results.
Nomad
Post a Comment