How do you cross-examine a whistleblower?

Erika Cheung, who took the stand yesterday for her direct testimony, which continued today, is known for being one of John Carreyrou’s sources for his blockbuster article about Theranos. You know, the article that’s part of the reason we are now witnessing US v. Elizabeth Holmes. (Holmes is on trial for several counts of wire fraud and conspiracy to commit wire fraud.) 

Cheung is, famously, one of the people who blew the whistle on Holmes. She’s even given a TED talk about it. The other major source for Carreyrou was her friend, Tyler Shultz, who also worked at Theranos and was the grandson of one of the board members. He is expected to testify later in the trial, since he tried to bring problems in the lab to Holmes’ attention.

At Theranos there were two labs, Cheung testified: the research and development lab and the clinical lab. She worked in both, starting with the R&D lab. Her position in both labs was largely the same — an entry-level job that involved running quality control tests to make sure the machines were working properly. About a quarter of the Theranos machines failed the test at any given time, Cheung said. By comparison, the commercially available machines failed much less frequently.

Theranos was Cheung’s first job out of college, and she worked there only about six months. During that time, she testified, she saw problems with the Theranos blood tests that gave her serious pause about their use on patients. In one email we saw today, Cheung wrote that she was happy to do quality control on a hepatitis C sample to confirm the test worked, but “I don’t feel comfortable running the patient sample,” she wrote. She didn’t want to give results to a patient from something that failed so often.

Cheung was “stressed and uncomfortable with what was going on,” she said on the stand. “Behind closed doors, we have all these problems and they [patients] think they have accurate results” they can use for medical decisions.

Cheung’s testimony established what the policy was for running VIP samples at the Theranos labs, that some results were thrown out so that machines could pass quality control, and that Cheung raised her concerns quite high up the food chain — though never directly to Holmes. Instead, Cheung talked to both Sunny Balwani, Holmes’ co-defendant who is being tried separately, and George Shultz, a member of Theranos’ board of directors.

Balwani told her she was unqualified to critique Theranos, and that she should unquestioningly run the tests or quit. So she quit.

She eventually went to Carreyrou and the Centers for Medicare and Medicaid Services, which are the regulators that oversee those labs. As a result, she was served with threatening legal papers by Theranos’ law firm.

The failures in the lab aren’t really in question — the defense acknowledged them in opening statements. Those failures, argued Lance Wade, Holmes’ lawyer, just weren’t Holmes’ fault. And so, in his cross, he made a point of establishing in tedious detail all the experts that Cheung worked with, and the existence of their PhDs and medical degrees.

And instead of trying to call Cheung’s credibility into question, Wade used Cheung to introduce the people in her chain of command — emphasizing her distance from Holmes. He also used her to introduce documents about lab regulations from CMS. There was a 120-page long document that set out standards from the lab assistants like Cheung to the very top of the lab. Per these regulations, a lab assistant — the lowest-level employee — must have a bachelor’s degree in science. That means Holmes wasn’t qualified even to do that work.

You can look at that two ways. One is that Holmes, unqualified founder, couldn’t have known how bad things were in the lab because she didn’t have a degree. The other is that it is a big red flag that the person who supposedly invented this revolutionary technology couldn’t work in her own lab.

Wade also went through documentation that showed that throwing out “outlier” samples was mentioned — bolstering his opening argument that nothing Theranos was doing was secret.

We saw pages of documentation with signatures affirming the procedures. Cheung was asked to read the signatures aloud. None of them belonged to Holmes.

It’s still possible Wade will rip Cheung apart on Friday — I’ve seen some gnarly cross-examinations — but this approach so far is like water. He’s agreeing with some of the damning things Cheung has to say, but laying the blame elsewhere. For the defense, it’s maybe the smartest move on the board.