[This is part of a long series which started October 9, 2021. The beginning point is at this link. You have joined us for comments about Chapter 3 of Paige Harden’s book]
This week we were treated to headline
Included in the story that followed was this passage:
“In his final remarks, prosecutor John Bostic said Holmes was ‘especially fond of using’ half-truths, which he described as something that is ‘arguably technically correct, but still leaves the listener with an unmistakable [sic] incorrect understanding about what the truth is.’ "
The trial lasted nearly four months. Of course, a report of 41 words (especially those 41 words) out of that trial is less than a complete analysis. What struck me when I read those words is how close they are to words that may describe some of Paige Harden’s conclusions. Granted, it is premature. We are only in Chapter 3. But since I do not want to be arrested for inadequate disclosure, I need to say two things now. One is a repeat: I have read all the way to the end of her book. The second is that I need to preview now, in generalities, what initially sent me down this rabbit hole. That will take some time and Elizabeth Holmes’ story forms a curious backdrop for it.
On September 9, 2021 I finished reading the New Yorker article on Paige Harden, which I found fascinating, all of it - the article, Paige as the subject and her research topic. Her book had not yet been published. I sent Paige an email, commending her on her work and mentioning a particular technical issue, hinted at in the New Yorker article, that troubled me. She was gracious in her reply, said she was aware of the problem and had dealt with it in the book (as the correspondence between us was private, I see no reason to quote her precisely as it does not really change the story below).
Now we need to go back another year or so and remind us all that I also read and enjoyed Stuart Ritchie’s book Science Fictions: How Fraud, Bias, Negligence, and Hype Undermine the Search for Truth, a damning account of unsavory practices of some researchers. Focus especially on the word “Hype” in the title while I introduce the phrase “Fake It Until You Make It.” This was reported as an industry-wide sentiment, deep in Silicon Valley culture, to be used in Lizzy’s defense. Essentially, standard tech salesmanship involves “pitching” to venture capital firms using outrageous claims to motivate investors. Who knew?
Stay with me here…
If you have been on this journey with me for a while you have read that I am not a genetic biologist. Neither am I an attorney. I am just a guy with a computer. I will now prove that with an embarrassing confession. [spoiler alert: I am not perfect].
Now we go back even farther in time. In 2011 a couple of academic friends approached me with a project involving calculations they did not understand. Theirs was not my field but, thought I, “Numbers are numbers, how hard can it be?” My job in the project was to play the “dumb terminal,” a role for which, as it turned out, I was uniquely qualified. They handed me the data, I read up on the methodology, cranked up my computer, input their numbers, spit out the results and delivered them, thinking little more of it.
Time passes. In 2017 I happened upon a statistical curiosity I had never heard of. It was not new, having been mentioned in the literature more than 100 years earlier. I had not gotten the memo. It occurred to me that the work I had done in 2011 was flawed in that I had not considered this phenomenon when I should have. I spent the next three years (not full time!!) reworking the data and concluding that, indeed, the problem with my output was very real. I approached my co-authors only to find that our paper had by then been cited more than 100 times. Great, in 2011 my little Wuhan Lab had unleashed intellectual COVID. It took nearly another year and examination by two independent statisticians to convince my co-authors I was right about being wrong, to write up the correction and get it published. In the revision some, but not all, of the conclusions of the original article were still supported by the data. The strength of that support, naturally, varied. I did not write the narrative either time. I still do not understand their theory and that is part of the problem. For all I know it is arguably technically correct, but still leaves the reader with an unmistakable [sic] incorrect understanding about what the truth is.
My co-authors will never speak to me again and I don’t blame them.
Why are the Feds not pounding on my door? Perhaps I am not as photogenic as Elizabeth Holmes? Or, could it be because I raised $945 million dollars less than she did? Actually it went the other way. Not wanting to taint my co-authors and hoping to avoid a public display of my bonehead error, after 2017 I paid some real mathematicians some of my own real cash to take me through the process thoroughly before I pressed for a published correction.
This is only partially a story of how I wanted the record corrected about something with my fingerprints on it. It is also about my own desire to KNOW a more accurate answer when that answer is available to be known. I got my wish, albeit painfully. In today’s context, it is also about how I (now) know something about a problem in Paige’s book, for it is in the same methodological area of my 2011 mistake.
In the early 1960s I am sure there was a social engineer who whispered in the government’s ear “If you create a welfare system based on the absence of a father in the household you will get The Great Society.” Lots of money was paid for studies to support this idea. Possibly, those studies were arguably technically correct, but still left the reader with an unmistakable [sic] incorrect understanding about what the truth is. Essentially, standard social engineer salesmanship involves “pitching” to politicians eager to use outrageous claims to motivate voters. Who knew?
Politicians stumping in the 1960s seized upon the line “Studies show…” The machine they created worked about as well as the blood testing machines manufactured by Theranos. In 1964 the food stamp program had 370,000 participants and a budget of 75 million dollars. Sixty years later it has 44 million supplicants at a present cost of about 90 billion dollars. That’s annually by the way. I don’t know about you, but I would not want to be remembered as the social engineer who cooked up the idea that resulted in an Epic Fail of that magnitude.
Makes Liz look positively small time.
Most of the sad results of the last six decades of mistakes are in the data Paige uses to complain about today’s not-so-great society and to advance her alternate vision of utopia. She will have no trouble convincing today’s politicians, but at what cost?
Liz and Paige have something in common. They each have an entry in Wikipedia. Today, Paige’s is more complimentary. But will it remain so? Paige might, like I did in 2017, want to think about her legacy. Wikipedia pages are subject to continuous updating. Sixty years from now Liz and my mistakes, mercifully, will be forgotten. Paige has the potential of being remembered as the woman who tried to Save The World with genetics, got the math wrong and ended humanity. It’s not nice to fool Mother Nature.
I don’t have a Wikipedia page. I am lucky. I am just a guy with a computer.