Wednesday, March 20, 2019

On aspects of Theranos

...and what it teaches us about bad, hyped science...

I admit that I have almost completely missed the story of Theranos and its founder, up to yesterday or so (I missed it partly because they have "only" made big claims about a limited issue, blood tests, not about a world revolution). The media were full of reviews of "The Inventor", an HBO documentary about the most famous recent Silicon Valley fraud. Interestingly enough, most of the footage in the documentary was shot for Theranos ads – and by Errol Morris whom I met in person at the 2005 SidneyFest (and a dinner in the Society of Fellows).

If you want to learn more, you need to Google search for "Theranos" or "Elizabeth Holmes". But let me start with a basic story.

Elizabeth Holmes, who is 35 now and awaiting up to 20 years in prison for massive wire fraud (she's officially broke but still lives in a hyper-luxurious apartment with various paid servants now, it turned out), has been born into an important dynasty. Her ancestors built important hospitals etc. and her father was a vice-president of a hot company – whose name happened to be Enron. I would think that even this fact should have raised some red flags – but it seems largely unknown to the public, even today.

She studied some biochemistry at Stanford but became a dropout, starting with Theranos (from Therapy+Diagnosis, originally named Real-Time Cures), a company with gadgets that make equally reliable blood tests fast and only need a droplet of your blood or so. It's not terribly important how many droplets the gadgets needed, it was nonsense for all small values of that number. For a young entrepreneur, she looked like a remarkably average teenager as a high school student.

Her Stanford instructors generally told her many reasons why it couldn't work back in 2003 (e.g. Phyllis Gardner said you can't nicely deliver antibiotics by patches) but apparently no one cares about Stanford experts' opinions about blood tests, anyway. In modern California, the self-confidence of a female dropout is clearly more important than the expertise of Stanford's professors who have studied the very same thing, sometimes for many decades! One might wonder why the Californian people contribute to these professors' salaries at all. Maybe to keep them away from the public life so that they don't talk about things such as their field!

Gardner said that Holmes took the strategy "try it until you succeed" which is "ludicrous in the context of healthcare". I would add it's ludicrous whenever you're facing some real limitations, starting with the laws of physics LOL.

As some of you know, I did buy Felisa Wolfe-Simon's claims about the arsenic life for a week. But that was different. The discoverer with some required training was just said to semi-accidentally come somewhere and see a new architecture of the DNA. It was really a question not about her but about the molecule – can there be arsenic instead of phosphorus? It looked plausible, she had personally passed the first consistency check by knowing that arsenic is in the same column as phosphorus, and that – along with some NASA supporters – was almost enough for me to believe LOL. I've seen many more details and my thinking is more complicated but I think it's healthy to honestly admit that our decision making is sometimes simple-minded, indeed.

Here, Holmes was said to have actively invented something that no one had before. I think that if I had paid attention back then, I wouldn't have bought it for a second. Incidentally, except for the unnaturally deep and high-contrasted blue eyes, Holmes looks like an ultimate average, boring, uninteresting blonde woman to me and I am baffled by most of the myths about her charm. However, when I listened to Alison Foreman's comments about the documentary, Foreman made me smile happily. The facial gestures etc., the apparent inability to stop the smile or expressions of doubts, so cute, almost like my Tom & Jerry video. OK, I would be much more likely to pay $1 billion to this Miss Foreman. ;-)

OK, so Ms Holmes started to dress in black like Steve Jobs, a role model of hers, having dozens of copies of the same clothes in her wardrobe, and she got a support from huge senior characters such as Henry Kissinger, George Schultz, and a dozen of comparable men (old powerful men have largely "created" her). She also apparently changed her voice, a point that the documentary made very popular now. Her corporate voice was apparently several octaves deeper than her natural voice. Not sure whether she just used throat muscles or some technology to achieve that.

Her corporate voice's pitch was in between Lisa Randall's, which is already deep enough to be viewed as authoritative by many, and Penny with some sulfur hexafluoride in the throat or what was that gas (I suppose that helium turns you into a chipmunk instead and nitrogen is too close in density to the normal air). Well, if women's authority is really diminished by the female voice, maybe it should be normal and tolerated for them to have voice synthetizers that rephrase everything they say in a deep male voice. ;-)

Also, as a Halloween monster, Holmes never instinctively blinks, only rarely and intentionally when she wants to stress a point.

Those Stanford professors and most experts' voices must have been totally ignored for some 12 years. Meanwhile, the company was growing – and growing into a structure that spied on all employees, had gorilla glass around the headquarters, and was doing fraudulent things everywhere. The main problem was that the products with the promised features have simply never existed. Finally, in October 2015, John Carreyrou published his story in the Wall Street Journal (the same man wrote "The Bad Blood" book that you may buy everywhere). I think it was a relatively friendly one but it really implied that nothing works. The gadgets of Theranos are very imprecise and they probably secretly use old-fashioned, state-of-the-art blood testing machines internally, anyway. So the added value of Theranos-anything is zero, it's a complete fraud. The story was published despite lawsuit threats from Holmes' lawyer (I almost forgot what it reminds me of) and other strong-arm tactics. Well, let me mention one. Holmes also wrote to Rupert Murdoch – who was invested both in Theranos ($125 million, oops) and the Wall Street Journal, you know – to stop the story. As a saint or a Lumonator who places integrity above his money, Murdoch refused to interfere against the journalist.

The events from late 2015 may be described as a rather slow journey from the top echelons of the business world to the jail.

Now, there are probably many reasons why tiny droplets of blood aren't being used for comprehensive blood tests. You know, even I can understand some possible problems – and I am sure that the best experts' explanations would be more accurate. But the point is that the blood isn't totally uniform so that every small droplet could be assumed to contain the same information as every other.

All matter is composed of atoms – but those are still small enough relatively to a droplet. The size of an atom is some 0.1 nanometers. Maybe the atomic composition may hurt you when you detect some really rare chemicals. OK, but the blood isn't composed just of atoms. The atoms are clumped into larger structures, in some cases to cells. And the size of a cell is several micrometers or so. So if you have a cubic millimeter of material, it contains at most (one thousand cubed i.e.) one billion cells. If it's mostly water, the content is vastly lower, especially if you look for rare cells.

Now, a cell that is still rather omnipresent in blood is the white blood cell, a part of the immunity system. If you search for "blood count", you may see that there are some 5,000 white blood cells in one milliliter (cubic centimeter) of blood. So in a tiny droplet, a cubic millimeter, there are just 5 white blood cells. In such a droplet, there's some probability that you get zero. How big is the probability?

Well, quite a few times, I was looking at the mining of the Bitcoin blocks and none were coming for an hour, so I watched it, hoping that the whole network was finally hacked or otherwise collapsed. ;-) It's a Poisson process of a sort – because the miners are basically "randomly shooting large numbers" and checking whether they have some rare property (then they hit the target and the number is used in a new block which defines the target problems for the next block) – it's designed so that there's no known or currently plausible way to speed up or reverse the process and avoid the brute force.

The average block is mined in 10 minutes. So the probability that you get 0 white blood cells instead of the expected 5 in a random tiny droplet is the same as the probability that a Bitcoin block hasn't appeared in a randomly chosen 50-minute interval! And I've seen it often. To be honest, I've seen a couple of 2-hour gaps, too. Hint: exp(-5) is some 0.007, almost one percent probability. (I would bet that Ms Holmes couldn't correctly solve even these homework exercises on basic probability distributions.)

And now appreciate that the blood tests may want to look for cells that are vastly less frequent than the white blood cells. Like the actual cause of syphilis, a bacterium which is just another cell (which looks like a thin helix, a slinky like a matrix string, a magnified helix from the DNA in some sense). Clearly, a small amount of blood may overlook things. The statistical errors are just way too high. You may easily overlook lots of things that you shouldn't. And there are probably other reasons why the hospitals haven't switched to tiny droplets decades ago.

The basic problem of small droplet blood tests is the same as the problem to find supersymmetry in a limited amount of \(13\TeV\) data. The last bin shows 12 events instead of the expected 5.4 and you can't know whether the excess proves supersymmetry or not.

But what is amazing is how long time it takes for the advanced society to collectively figure out that this business was just complete fraud – and how many people (and how many famous, celebrated people) can keep on worshiping a totally fraudulent CEO. The company peaked in 2014 or so, more than 10 years after it was founded, and its capitalization was some $9 billion. (Tesla is 5 times larger now but Tesla actually produces products that work on the street – they "just" don't add up financially. So Tesla is clearly a less "pure" example of fraud than Theranos although $40 billion out of Tesla's capitalization could be said to be "fraudulent".)

For more than a decade, all those experts' explanations had to be totally ignored – and I feel that even in the new HBO documentary, the scientists' understanding what is possible plays a very minor role. In fact, I only see two seconds, 0:52-0:54, where an older female scientist says "it's impossible, physically". Just a detail! ;-) Don't you think that the scientific plausibility of people's multi-billion claims should receive more than 1% of the attention? The company happily grew and she was celebrated as the heroine of business and women in engineering etc. etc. The set of supporters was amazing, included lots of famous names, and many of these names continue to do their shows with recommendation advises (greetings to Tim Draper, still on CNBC, and Larry Ellison who moved from Theranos to Tesla). And even in 2016, after the Wall Street Journal basically unmasked that the company was complete fraud, Holmes still organized fundraisers for Hillary Clinton. I guess that with the "very clear" status of Holmes' business that everyone sees now, Trump would emphasize Hillary's links to Holmes in his campaign if the campaign were ongoing now and Hillary were running.

On the other hand, it may sometimes happen that a Stanford dropout makes a revolution related to the applied science like this one. But the likelihood is generally small and I think that even when it would happen, most of the Stanford experts would simply agree that their former student has done something important, regardless of the coursework. America is great for performing well-funded experiments and this 12-year-long Theranos experiment was sort of an example, although one with a bad conclusion. To optimize the outcomes, there must be some healthy balance between the expertise that insists on "what has been demonstrated" on one side and the "courage" of the entrepreneurs who go in a new and different direction (accompanied by the society's support for such "brave people" – I used the quotes because the support clearly makes them less brave de facto) on the other side.

In this case, it seems spectacularly clear that this balance has been tilted to the direction of hype. The probability that a female Stanford dropout would know better how to perform effective blood tests than her professors is simply small (needless to say, my very taking her sex into account would be considered non-PC in California – and the fact that these concerns and arguments are being suppressed is a reason why we're going to see women overrepresented among similar crooks, and this trend already started to be visible: it's just easier for women to become crooks now; they get a greater support to become anything than the men but it's easier for them internally to become crooks than actual "disruptors"; for this reason, you may say that the political correctness is statistically responsible for sending many women into jail).

Hype became way too important and the "standard" experts' opinions have become too unimportant. I think that there are many cases like that in California, America, and the West. I wouldn't claim that it's a majority... But there is surely a greatly enhanced room for Holmes-like crooks in our "Western" culture and especially in the contemporary Californian culture.

Her "biochemistry" is bad engineering and she's created a company that looked like a $9 billion company to its investors (yes, the number has already been adjusted to zero). At least statistics indicates that people have died because of its dysfunctional methodology (and broken syringes). Pure science and theoretical physics in particular has many examples of hype-driven pseudoscience that is really analogous to Holmes'. The difference is that in theoretical physics, people aren't dying when someone claims that some non-existent structure exists in the Standard Model. And people aren't really paying billions of dollars for such individual wrong ideas, either. Instead, one billion dollars is what the whole field (with thousands of real experts) gets annually in the U.S.

So pure science such as fundamental and particle physics is too disconnected from the everyday problems – and from the really big money (fundamental physics is a Cinderella that annually got just slightly more than a single medicine industry crook, Ms Holmes!) and people's survival (well, high energy physics became important and reasonably supported in 1945 when it reduced the survival chances of some Japanese people). But if you care about the truth as found by science, and I do care about it more than I care about the size of blood droplets (although, like Holmes, I totally hate needles!), then the message about the Theranos' "hype that was beating the merit" is directly applicable to fundamental and particle physics, too.

In fact, some examples are strikingly similar because the hype is sometimes also heavily energized by the presence of a physically attractive, and sometimes young, female. So maybe this is a good label that rational readers should keep in mind when they read about some wonderful stories about breakthroughs in science that were made by some seemingly unlikely folks (and it doesn't have to be arsenic life or octonions in the Standard Model): "Haven't I learned something from the story of Theranos?"

Well, some people are learning nothing, ever, but let me hope that others still are learning.

P.S.: Despite the possible 20 years in jail, Holmes keeps on living like a rich, relaxed, and successful woman, apparently with no fear. I guess she wouldn't lose sleep in a whole night because of a threat of a $10,000 lawsuit over a dozen of expletives, like your shy and humble correspondent did. A behavioral economist claims that she could lie in the court and elsewhere because she really believes she is saving the world. I have some trouble to believe this explanation (she must just know that nothing works) – an excuse for the criminals' self-confident behavior. I really think that these people, and it's true for Musk to a lesser extent, too, simply believe that they will get away with everything because they're surrounded by Yesmen who make their gurus look impossible to defeat in their own eyes. In other words, these self-confident people have an almost literal "thick skin" composed of their subordinates.

No comments:

Post a Comment