My background is biomedical engineering, and I just never, ever understood how the sampling rate could ever work. If you're looking for a low-energy signal and your sampling rate is really low, you're just going to miss it and this should've been screamingly obvious to everyone since the beginning.
It's like going to the beach, drawing a 5 gallon pail of seawater, looking into it, and then declaring that the beach is shark-free today and it's safe to swim.
This has been my favorite tech story of the decade. Watching a rich, well-connected phony get knocked down and called out on all the bullshit fills me with glee. It excites me in ways we don’t talk about in polite company. So many hard working smart people out there toiling away in obscurity while these “elite” fakers get all the recognition and money. The schadenfreude in watching one of them finally and spectacularly burn up is too good to bear.
However, I have had a related experience that stops me, even in these cases, from immediately shouting "the king has no clothes":
Some 25 years ago I was giving a talk at some point, that mentioned spread spectrum communication to a small crowd of physicists and engineers, none of whom was familiar with the concept. When I drew an energy/frequency graph in which the signal level was below the ambient noise level, I got laughing/mocking/disbelieving response, and I had to actually show (a simplified) version of the math to proceed.
None of those people were stupid. In fact, I believe most if not all of them are way smarter than me; but they've all adopted the "s/n ratio must be large enough at every distinct frequency " as an axiom, which is simply wrong.
Since then, when I see extraordinary claims, I just wait for the extraordinary proof.
> It's like going to the beach, drawing a 5 gallon pail of seawater, looking into it, and then declaring that the beach is shark-free today and it's safe to swim.
If shark shed sharkticles that can easily be detected, and the water mixing speed is sufficient, then -- yes, you might be able to make that claim from a 5 gallon pail of seawater.
But before I believe you can do that, I would want to see the theory and a working demonstration. It's astonishing that wasn't demanded of Theranos by any of the investors.
Theranos' claims fundentally misrepresented the state of medical diagnostics, and showed at best a failure to grasp the nature of the industry, and at worst a deliberate misrepresentation. Apparently it was the latter.
If any of the investors now saying "gosh who could have guessed" had taken 2 seconds to ask anyone who worked on or with real medlab tech about Theranos' claims, they wouldn't have touched that company with a 10 mile pole.
See Siemens Advia 1800 specs under "Microvolume Technology" https://www.healthcare.siemens.com/clinical-chemistry/system...
It only uses 30ul of sample to run up to 15 assays.
How is that any different from what Theranos was doing?
Only difference is that you collect a bigger vial when using the Siemens machine. And then don't use most of it. So why not collect a smaller vial?
This is easier to do consistently in larger vials, reducing the error in this part of the process to acceptable levels. It’s the same reason why a recipe will normally be more consistent when making a larger batch.
So you could collect less, but the way the modern lab industry operates (collecting samples, preserving them, and shipping them to an offsite lab for analysis) there is little business benefit to using smaller collection vials.
1. Multiple tests. Most people aren't going to the doctor for a single test. A majority of the testing is Comprehensive Metabolic Panels, which covers 20-30 different things at least. Many patients get a CMP plus other stuff, so enough is drawn for all the tests at once.
2. Possible retesting. There are dozens of things that can go wrong between drawing the blood and sending a result to the doctor. Maybe a tech drops the tube and spills some blood out. Maybe the machine has a problem mid-test and you have to rerun the samples after it's fixed. Maybe the patient has a critically high/low range of something and you need to retest for confirmation. The worst possible scenario for a lab is to have to call the patient back in for more blood, because that causes patient pain and could make the doctor start using a different lab if they're perceived as unreliable. There's actually a legal requirement for labs to store blood for a week after testing, just in case.
3. Different test requirements. The CMPs usually test the serum only (i.e. your blood gets put through a centrifuge), but there are a variety of tests that run on whole blood. Because of the possible re-testing above, and because whole blood test samples need to have anticoagulant added, you'll usually have two separate tubes drawn for whole blood vs. serum tests.
There are definitely areas where a wunderkind can advance the frontier by ignoring dogma about what is and isn’t possible; but biomedical engineering doesn’t seem like one of them.
I always had my suspicions too; her idea seemed so simple that it had to have occurred to larger med device makers too. If they couldn’t solve the problems related to the small sample size with many years of engineering experience, what made her think her chances of success were any better?
The Hacker News commenters have always been quite kind and optimistic about Theranos and Ms. Holmes. Were your doubts downvoted?
* Holmes started with the best of intentions
* Theranos raised capital from starry-eyed investors that didn't carry out proper due diligence
* Realization set in early on that it wasn't going to work out
* Theranos got scared as so much of their time and other people's money had been invested
* The questions mounted
* Stalling became outright deception
* The realization of the consequences after the WSJ article became starkly obvious
* Theranos got spooked and doubled-down on covering up out of fear
.. but this is all speculation, and we'll never really know.
It does seem though that this didn't turn into fraud (which by the end it's quite clear it was) until late on.
It's a bit tragic, really. It seems Holmes could have come clean early on and avoided the worst of the fallout (but for whatever reason that didn't happen).
That said, from a personal perspective, I honestly hope Ms. Holmes is either a sociopath or was actually in the dark about how bad things were (e.g. her direct reports were lying to her).
It seems like it would be a pretty excruciating mental burden to go into work every day for an extended period of time, knowing your company was inevitably failing, and just trying to keep the plates spinning. Ugh.
The exception is that this science had already been explored, for decades, by the major companies that specialized in this technology. There simply was nothing to show that worked, yet, so there was nothing to commercialize. And the big players aren't going to publish their interim research, or Holmes wasn't motivated to actually do the homework.
She wasn't the first to think of it, but had enough traction (combined with a vaccuum of commercial products) to convince people she had, or had some mystery secret sauce that was going to make it work.
The fact that Theranos hired Cass Grandone, an Abbott Diagnostics exec that pioneered the previous generation of microparticle immunoassay/ELISA technology, and Grandone walked away after six months shows that he knew their science never worked, and was never going to work.
There were two companies that I simply couldn't understand, at all, from media reports in the last couple decades. One was Enron, 'tother Theranos. Turned out there was a reason why.
Among startup founders, puffery is endemic. You have a vision and you try to talk it into being. Whatever you're doing probably won't work, but you can't say, "Well, there's a 90% chance we'll fail," because nobody gives you money or labor then. You have to develop this (unjustified) confidence, and a deep understanding of what people want to hear. Silicon Valley culture definitely encourages this.
To succeed, one also of course has to develop an equally deep understanding of the reality of the business domain. But as we see with Theranos or Hampton Creek or uBeam, you can get surprisingly far on vigorous self-promotion and creating the appearance of success.
The thing that really keeps me awake at night is that there are surely a lot of companies that faked it nearly as much but managed to find a real business before things caught on fire.
This Vanity Fair piece provides a fair amount of character background: https://www.vanityfair.com/news/2016/09/elizabeth-holmes-the...
I would like to comment jokingly here with the saying "If you wish to [defraud] go into politics." But let's not support people in that cause.
Seriously, in this case it appears to me it was more and more a series of cascading lies that had to be added to cover the original lie. Which either stemmed from hope, misconception or lack of due dilligence.
If she thought actually her idea was going to work, and nobody raised a flag from her peers, faculty, ..., I think after some point the weight of capital invested made the executives follow the "fake it till you make it path".
I think in that case there was a lot of emotion, pride and "our name is on the line" that came into play.
However, in that case I wonder, if it is believable that major investors did not do their homework? Or are they also part of the game. Cass Grandone bailing on such a promising project should have been a red flag for any serious investor. I bet in that case, a lot of major investors knew and chose to follow the path. In that case they are accountable.
If the story reported about Tyler Shultz is right that means that George Shultz, and his peers, knew but turned a blind eye. In that case, I wonder if Holmes ended up as the scapegoat just to keep the boat floating a bit longer (or even recover). In that case her mistake was writing down a lie in an official document and signing it. Alas, I see clues everywhere that the whole board and investing body are complicit in this deception.
I decided in undergrad that Honors programs exhibited a lot of the same characteristics. Kids in those tracks are treated with velvet gloves by professors, come to believe their own superiority, are inevitably unprepared for the real world / a colleague proving them wrong.
Hype trains create their own RDFs that effect even the subjects.
http://micheleincalifornia.blogspot.com/2016/05/theranos-cul...
> I used to refer to it pretty regularly as the Blood Unicorn, Elasmotherium haimatos, because it was a Silicon Valley unicorn with a peak valuation of $9 billion that managed to raise $700 million from investors.
I suppose if "unicorn" is meant to describe fundraising activities, it's accurate. They really raised that money.
EDIT: This article[2] (also by Matt Levine) predates [1]. “And here is a unicorn being barbecued in what looks like a medieval manuscript that I will just assume prophesies the coming of Theranos. ‘And lo, a Unicorne shall come among ye, and ye shall call it by the name Theranos, or in the Old Tongues, Elasmotherium Haimatos. And it shall take your Bloode, but only a lyttle bit of your Bloode, and it shall do strange Magick upon said Bloode, and tell ye many Things. But then it shall come to pass that its Magick was [makes 'so-so' hand gesture], and that those Things were mostly not true. And ye shall barbecue that Unicorne.’”
[1] https://www.bloomberg.com/view/articles/2016-08-02/mergers-u...
[2] https://www.bloomberg.com/view/articles/2016-05-26/libor-cha...
Bloomberg and most other media companies pumped Theranos up for all it was worth, then Murdoch's WSJ deflated the balloon, having previously blown it up to dekacorn dimensions.
It feels like the pump and dump plays so popular in the dot com era by the money.
Also, the way she compatmentalized her business and threatened former employees proves she was complicit.
Theranos history is littered with staged demos (not being described as such), deception (running tests with other manufacturer machines), and outright lies.
For all the upbeat press release regurgitation (of which there was a lot, absolutely), there's limits to investigative journalism of those who _were_ skeptical.
I can't decide if the investors were foolish for trusting in news articles, or if the news articles should have been trustworthy.
He’s the living embodiment of the law’s “reasonable observer”.
> Holmes was paid a salary of approximately $200,000 to $390,000 per year between 2013 and 2015
> Holmes has never sold any of her Theranos stock.
> Due to the company’s liquidation preference, if Theranos is acquired or is otherwise liquidated, Holmes would not profit from her ownership until – assuming redemption of certain warrants – over $750 million is returned to defrauded investors and other preferred shareholders.
So she probably didn't profit that much financially from the fraud (especially not compared to the sums in play), unless Theranos somehow magically comes back from the grave.
