Coming Soon:

Now Available: Volumes I, II, III, and IV of the Collected Published and Unpublished Papers.



Total Pageviews

Tuesday, November 19, 2013


As I set out at six this morning on my daily walk, under a beautiful full moon, my thoughts turned to the very helpful comment by Marinus Ferreira about the underlying engineering problems behind the botched rollout of the Affordable Care Act insurance exchanges.  I reflected, as I often have before, on the curious fact that most of us really have little or no idea how the things that we use actually work.  Several years ago my older son, Patrick [the chess grandmaster and hedge fund manager] gave me a very interesting book about the design engineering problems in the invention and development of the commonplace things we all use.  The first chapter was devoted to the invention of the paper clip.  Very few of us, certainly not I, have ever thought about the invention of the paper clip.  I recall that when I came upon the letter file from the 1917-18 term that my socialist grandfather served as New York City Alderman, I found that sets of papers were held together by straight pins, presumably because at that time the paper clip had not yet been invented.  The second chapter of the book dealt with the tricky engineering task of designing a pop top beer or soda can.  The central problem was that if you made the top too easy to take off, the pressure of the carbonated beverage in the can would blow the top off prematurely, whereas if you made it too difficult to take off, the tab would just break off in your hand.  Who knew?

When I got my IPhone, which seems to have caused as much trouble in cyberspace as that coke bottle did in the Kalahari in The Gods Must Be Crazy, my younger son Tobias, the law professor, who serves as my IPhone guru [he solved my problem of the rotating screen] pointed out to me something that had never occurred to me, namely that a mobile phone is essentially a shortwave radio.  A proper appreciation of that fact would put in perspective the widespread anxiety about NSA spying.  I mean, we are all now amateur ham radio operators.

Which led me back, finally, to Marinus Ferreira's point.  Most of us have very little idea at all of how computers actually work.  Oh, we -- by which I mean the readers of this blog -- all know how to word process and surf the web and play computer games and maybe use an Excel spreadsheet, and some of at least [not I] are adept at photo shopping a picture.  But how many of us know, all the way down, how computers work, in the way that red-blooded American boys are expected to know how the internal combustion engine works?

Let us think about this in layers, as it were, starting [as Aristotle would say] with things that are first in the order of knowing and proceeding to things that are first in the order of being. [Philosophy is really very useful for expressing commonplace ideas in fancy ways.]  All of us can use a computer, as I have said.  A handful of us [again, not I] can actually program -- that is to say, write sets of instructions for a computer, even perhaps create entire applications.  But that is probably a very small fraction of all the reasonably skilled computer users -- say, one percent or fewer?

Even if you can program, you are still operating very much on the surface, as it were.  You may have no very precise idea of what you are actually causing to happen when you type the commands on the keyboard.  Now, I understand the theoretical significance of John von  Neumann's brilliant idea of expressing all mathematics in a binary number system.  Briefly, for those of you who have never given it any thought, a number system based on ones and zeroes perfectly models the on/off or open/closed structure of an electrical network -- 0 for no current flowing through a connection and 1 for current flowing through.  Every on/off switch is representable by a "bit," and a string of eight zeroes and ones in binary notation, representing a particular specification of ons and offs in an electrical circuit, is a "byte."  If you use the eighth binary digit as a test of the success of the transmission across a junction, that leaves you with a seven place binary number, which is to say 128 different combinations of ones and zeroes [two to the seventh power].  When I look at my computer keyboard, I find 47 keys with symbols on them, which, what with upper and lower case, gives 94 different binary numbers to which one can assign letters and symbols.  That leaves another 34 for other uses.  Why not a sixteen bit byte?  Because in the early days that was too complicated to build into the wiring of the computer's central processor.  Why not fewer?  Because then the letters and symbols you wanted would not all be modeled by a single byte of binary code [the next step down from 128 is 64, of course.]

But all of this, which I understand at a perfectly useless abstract level, is still very much on the surface.  What is really going on is the flow of electrical currents along circuits -- at first, circuits of wires and tubes, subject to heating problems very difficult to manage, then later printed circuits, and then in solid state transistors, which I do not understand in any real sense at all.

Not one in ten thousand of us actually understands the physics and engineering of a transistor [which suggests that there are thirty thousand people in America --a bit of an undercount, maybe, but not by much, I would imagine.]  And yet all of us every day use computers [and cell phones, which are really little computers as well as shortwave radios].

If we consider the entire sweep of the history of human beings, which is to say one hundred thousand years or somewhat more, this is a very odd and unusual state of affairs.  For at least ninety percent of that time, and maybe more like ninety-nine percent, most people had a grasp of how their technology worked.  There were specialists, of course -- ironmongers, silver smiths, carpenters, shipwrights and wainwrights and wheelwrights [I omit philosophers, kings, and theologians] -- but even if you did not have the skill to fashion a piece of iron into a sword, you could watch a blacksmith do it and grasp what was going on.  Today, nothing remotely like that is true.  How many of us really know what is involved in making a machine that can turn out Barbie Dolls?

Which, by a circuitous route [Tristram Shandy has nothing on me when it comes to digressions], brings me back to Barack Obama and the botched rollout of the insurance exchanges website.  Obama is an intelligent man, but I very much doubt he understands as much as Marinus Ferreira does about the problems inherent in launching a website of that nature.  Lord knows, Kathleen Sibelius certainly does not seem to.  [For my foreign readers, she is the former Governor of Kansas who is currently Secretary of the Department of Health and Human Services, and hence the person in charge of the website.]

So perhaps I should not be as censorious of Obama as I was.  Then again, the mark of an effective manager is knowing what he or she does not know and finding someone who does.




Wholeflaffer said...

This is an excellent point and needs to be made. I work at a Fortune 500 company building architecture for 1000s of web application, none quite taking quite the load as the single "application" that is, but in the aggregate taking millions of requests an hour.

So knowing what I know, I can say that there were several mitigating factors which make the rollout ridiculosly complex:

1) The interactions with numerous backends of all of the insurance companies makes this an exceedingly complex application. These backends are most likely using different protocols and requirements for authorization to pull date from them. This is hard to do with several, stupidly hard if there are 100s of them.
2) The backends do not belong to the same companies, so working with each company and their differing cultures adds even more complexity.
3) The scope of the work changed due to political machinations. The Fed had to take over the exchanges caused by Red states abrogation of running their own. The Fed did not know this when the project started (it takes years to build this sort of system) and added scope creep to an already complex project.

I do not think, necessarily, that Sibelius running the project was a bad decision. She could certainly have gotten the right people into right places to get the job done. And they certainly needed someone with a large stick (cabinet member) to run it so that HHS had enough authority.

I would put the blame more squarely on the hiring of the consulting firm(s) and their lack of due diligence. There is some reports that the consulting requirements meant that the firm(s) were hired for the wrong reasons (i.e., cheaper is better).

So finally, I would say the issues with the rollout were not related to micro issues on the level of people not knowing the hardware and physics. Rather, the issues stemmed from the usual places failures happen in IT: in side the craniums of a lot of people making poor decision.

Robert Paul Wolff said...

That sounds very plausible indeed. I speak as someone with ABSOLUTELY NO EXPERIENCE of anything remotely like this!! Many thanks.

Unknown said...

"If we consider the entire sweep of the history of human beings...For at least ninety percent of that time, and maybe more like ninety-nine percent, most people had a grasp of how their technology worked."

I think this is false. Before the 20th Century, the overwhelming majority of people were illiterate. Those of them who were not metalworkers could see that to make a bronze, the metalworker would put some stuff together and heat it and then cool it, but I don't see this is any more of a "grasp" than Bob Wolff and I have of what programmers and circuit designers do to make it possible for us to word-process or blog.

And I'm quite sure that none of my grandparents (all immigrants from eastern Europe) knew what made an electric motor run.

Robert Paul Wolff said...

I am not so sure you are right. I have in mind any time up to the Middle Ages. Remember, illiterate does not mean ignorant or stupid. Until quite recently, historically speaking, almost everyone was illiterate.

chrismealy said...

Another professional web site developer here. My diagnosis of the problem is that no part of Obamacare should have relied on a web site at all. Something like 20% of Americans don't even have an email address!

They should have just sent out a letter to everybody telling them whether they're qualified for Medicare or what their subsidy is for ACA compliant plans. The IRS basically knows everything already, there's no point in asking people for that info. Then folks could just sign up on paper or in person or on an insurance company's website. It would be up to the insurance companies to connect to the government systems to collect the subsidies. There's no point in putting regular people in between.

The politics would have been a hell of a lot better. Millions would get a letter saying, "You get free insurance." Others would get a letter saying, "You get $200/month to help pay for insurance."

chrismealy said...

For example, why should the government handle payments to insurers? Payments programming is insanely difficult! Just stay out of it! Let the insurance companies do the work and take the blame.

Obamacare is basically $100 billion/year for Medicaid expansion (oops, said Medicare in the previous post) and another $100 billion/year for insurance subsidies, and somehow they've turned a huge benefit for working people into the world's worst web site project. Minus the web site it would be a pretty popular.

T Gent said...

I just keeps blowing my mind that the ony way to provide people with affordable healthcare in the US has been to give taxpayers' money to the same people who have been making a profit out of callously denying care at all costs. How sad...

Simon B said...

Hi Bob,

Just wondering what the book on design engineering problems is called? It sounds like a very interesting read!


Robert Paul Wolff said...

I couldn't remember the title so I looked on my shelves. But since my books are shelved in a perfectly rational way, alphabetically by author [except for major subjects like Hume or Plato], that did me no good at all! Maybe my son can remember. I will ask him. It is a really fun book, especially the early chapters.

CCH said...

Sounds like, "The Evolution of Useful Things: How Everyday Artifacts-From Forks and Pins to Paper Clips and Zippers-Came to be as They are" by Henry Petroski.