Thursday, May 29, 2014

Creative Destruction, thankfully, applies to human capital

Tyler Cowen linked to two recent stories:

National Nurses United has an ad warning about the dangers of replacing nurses with data.

And here is a story of the battle between farmers and "big data".

The nurse ad is pretty funny, but probably not in the way the union intended.  It reminds me of these wonderful old ads from musician's unions in the 1930's campaigning against recorded music (ht: KPC).

These conflicts are an example of how difficult progress can be when it causes obsolescence of something we consider dear or sacred.  Is there a more quintessential image of human character than a farmer's intimate relationship with the sight, smell, and feel of his dirt?  Monsanto has technology that makes this knowledge increasingly obsolete.  Likewise, AI diagnosticians are becoming more dependable than doctors and nurses.

Our reaction to this is much different than our reaction to the obsolescence of technology.  Or, even think of cars replacing horses.  Stable keepers' unions probably had ads against those infernal new machines, but with distance, we recognize these advances as boons to human opportunity.  (Auto repair diagnosis has also become highly automated.  I think it is an interesting exercise to compare this to medical automation.  There was relatively little despair over this development in auto repair.  How does this relate to our feelings toward this service, and toward auto mechanics versus doctors and nurses?)

It seems compassionate and pro-human - pro-community - to support the skilled farmers and professionals in these matters.  The sentiments being prodded in the nurse and musician ads are universally sensitive.  And, come on, are you really going to side with Monsanto over farmers?  Is it outrageous and mean spirited of me to feel disappointment that our educational institutions are shielding teachers from these same pressures?

But, the irony is that these easy sentimental biases are wholly anti-progress.  If we apply them in these contexts, we are allowing our subconscious knack for social signaling to overbear progress in the very society we are signaling to.  We have a bias for supporting obsolete human capital because it shows our loyalty and support for people doing good in our communities.  It shows our sophistication - that we can recognize talent and skill.

But, what are the consequences?  In the end, the most powerful force for improvement of the most poor, the most egalitarian tool in the history of humanity, is automation.  Canned music means that paupers can now enjoy the same entertainment previously only available to the elite.  AI medical tools will mean that the triage office in rural Appalachia or the Kenyan savannah will offer the same quality of diagnosis as the Mayo Clinic.  Farming automation means that basic foodstuffs are abundant for many poor people across the globe.  Imagine the possibilities for marginalized kids if teaching were largely automated.  Not only would the poorest kids have access to the same sources of knowledge as middle class kids, but resources would be freed to provide them with even more services.

A central dilemma of political advocacy:  Being truly pro-progress sometimes necessarily means taking postures that make you look like an ass.  ("Don't you see?  We're all better off because the skills you built over a lifetime in a widely honored craft are no longer needed!  Huzzah!")  Of course, also being human means that if you commit to this discipline, sometimes you will bravely be defending progress at your own social cost, and sometimes, you will just be being an ass.  Since these matters can never be universally determined with certainty, not looking like an ass is probably the smart choice most of the time.  Thus, I would humbly suggest that, if your economics or politics don't frequently put you in the unavoidable position of supporting positions only an ass would support, you might just be a cog in a giant collective action problem.  But, I suppose I could just be saying that because I'm an ass.

6 comments:

  1. Great post. The knowledge use we want, we can save (i,e. provide space and measure) for the social reasons that matter. That's the real product, and that's what we can coordinate for.

    ReplyDelete
  2. There is a big difference between cars and human beings, though. Cars are built to exact specifications and thus their making can be automated. Human beings are unique. And even with the most detailed statistical analysis, the sample size of a human will always be one.

    ReplyDelete
  3. Computerized Diagnostics in Healthcare Potentially Hazardous to Our Health; Nurses Right to Oppose to Them

    By Neco Copiosus

    National Nurses United has been running ads on TV and radio lately about the obvious dangers of a troubling trend in our increasingly corporatized, for-profit healthcare system (consisting of the large hospital chains and health insurance cartels), replacing nurses and doctors with computerized diagnostics and protocols, which must be rigidly followed.

    The nurses contend they are concerned about the potential harm these unfeeling, non-human computerized protocols might cause their patients. Such rigid adherence to computerized diagnostics seems to this observer self-evidently to be a bad idea and I am pleased to see that the nurses understand this as well. We are all potential patients someday, and I for one want humans to remain in the diagnostic equation with the ability to override or ignore computerized or automated patient assessment.

    But not so fast say some critics who have accused the nurses of being “anti-progress.” One critic of the ads recently unabashedly stated, “Auto repair diagnosis has also become highly automated. I think it is an interesting exercise to compare this to medical automation.” In other words, computerized diagnostics help Joe at Joe’s Auto Repair down the street to figure out what’s wrong with your car therefore, (insert quantum leap of logic here), why wouldn’t the same hold true for the hospital down the street with automated medical diagnostics? This critic apparently sees no difference between the computerized assessment of an automobile’s breakdown and the breakdown of a human being. Lost on this critic as well, is the reality that while there is very little variation among automobiles, there are no two human beings alike. The critic went on to equate the nurses, in so many words, with head-in-the-sand, anti-progress Luddites.

    In Utah in the early ‘80s many in the Provo area opposed the expansion of a local steel smelter, which was polluting the air and causing an inordinate amount of respiratory illness among area children. The opponents of the smelter expansion were accused of being “against progress!” The smelter was later closed for financial reasons and lo and behold, the air became cleaner and childhood respiratory illness dropped drastically.

    The promoters of automated healthcare diagnostics similarly argue their new gadgets represent “progress!” But just like in the example of the smelter expansion, the focus should not be solely on some vague notion of progress or whether it saves money, but should also include the question – Is this safe? What impact will this have on my health and the health of my children?

    We’ve become accustomed to believe technological innovation represents “progress” but have forgotten that such innovations do not inexorably result in improvement of the human condition. Blind obedience to notions like -- “progress is always good” -- is tantamount to the worship of a false idol. Modern capitalism and its adherents have been worshiping at this alter for so long it has become sacrilegious even to question whether the latest innovation will actually improve anything.

    Humans have “advanced” in the name of “progress” with such alacrity that the rest of us can hardly stand it; e.g., Chernobyl, Fukushima, climate change. Blindly contending the latest innovation or improvement equals “progress” is truly modern day idol worship. Or as the late Colorado governor William Gilpin once said, we have come to believe that “progress is God.”

    We are all potential patients someday, and I for one want humans to remain in the diagnostic equation with the ability to override or ignore computerized or automated patient assessment. I am therefore glad that the nurses union gets this and is resisting human-less medical diagnoses. Hang in there and keep up the good fight nurses! We need and trust you!

    ReplyDelete
    Replies
    1. Thank you for putting so much thought into your response.

      Fundamentally, I think we completely agree: "the focus should not be solely on some vague notion of progress or whether it saves money, but should also include the question – Is this safe? What impact will this have on my health and the health of my children?"

      That is precisely my point.

      I will suggest that the rest of your comment is simply begging the question.

      Delete
    2. We don't understand medicine nearly well enough to replace nurses' "knowhow" with automated systems.

      Automating *too early*, before we really know what we're trying to do, suffers badly from the garbage-in-garbage-out problem and is inevitably a disaster.

      So, although we might want to do that *eventually*, we shouldn't do it now. Case closed. Anonymous tried to make that clear, but you missed the point.

      What should we do? Well, we have a database of all prescription drugs with their known side effects (with occurence rates). We should make a similar searchable database for all *diseases* with their known symptom patterns (including unusual ones).

      This would free doctors from doing a particular sort of diagnostic work which they are bad at (coming up with diseases off the top of their head). It would free them to look at the 20 diseases which match the search pattern and use their expertise to figure out *which one* it is.

      There are certain parts of medicine which are amenable to automation now. Nursing really, really isn't one of them. Nurses are trained to spot symptoms which doctors overlook. Data can't spot symptoms.

      Delete