support. Noticing that the miniLab running the potassium test was
stuck at 70 percent completion, he took the cartridge out and rebooted
the machine. He had a pretty good idea what had happened.
Balwani had tasked a Theranos software engineer named Michael
Craig to write an application for the miniLab’s software that masked
test malfunctions. When something went wrong inside the machine,
the app kicked in and prevented an error message from appearing on
the digital display. Instead, the screen showed the test’s progress
slowing to a crawl.
This is exactly what had happened with Parloff’s potassium test.
Luckily, enough of the test had occurred before the malfunction that
Kyle was able to retrieve a result from the machine. The breakdown
had happened while the device was running the test again on the
control part of the sample. Normally, it would have been preferable to
have the initial result confirmed by the control, but Daniel Young told
Kyle over the phone that it was OK to do without it in this case.
In the absence of real validation data, Holmes used these demos to
convince board members, prospective investors, and journalists that
the miniLab was a finished, working product. Michael Craig’s app
wasn’t the only subterfuge used to maintain the illusion. During
demos at headquarters, employees would make a show of placing the
finger-stick sample of a visiting VIP in the miniLab, wait until the
visitor had left the room, and then take the sample out and bring it to a
lab associate, who would run it on one of the modified commercial
analyzers.
As for Parloff, he had no idea he’d been duped. That evening, he got
an email from Theranos with a password-protected attachment
containing his results. When he opened the attachment, he was happy
to see that he’d tested negative for Ebola and that his potassium value
was within the normal range.
—
BACK IN CALIFORNIA, Holmes and Balwani were laying the groundwork
for a bigger show-and-tell. Holmes had invited Vice President Joe