Sunday, October 4, 2015

Signs of the Times

This week the Leader Times published an editorial criticiing the PA Turnpike Commission for dragging its feet on implementing fully-automated toll collection and another editorial criticizing the defense attorney for Michael Cinefra, the Plum School District substitute teacher indicted for having sex with a 15 year-old student.  Both editorials make good points based on what we know so far about the situations.  The board deserves kudos for the second editorial.  Not so much for the first.  It is a bit misleading on the facts and leaves unanswered larger questions about our society that the toll collection program raises.

The first editorial supposes that at least one of the reasons the Turnpike Commission has decided to initiate a study of full automation rather than implementing it system-wide is the desire to preserve the jobs of as many toll-takers as possible as long as possible.  Under the assumption that full automation will save the state large sums of money in the form of toll-collector salaries and benefits and the costs of upgrading toll booths, the editorial board complains that the Turnpike Commission is just wasting taxpayer money, money that could be better spent on ... oh wait, the editorial board didn't say anything about that at all!    So, what is the point of the editorial besides taking shots at state government spending money?

Of course the editorial mentions that the toll-collectors are unionized, (i.e., lazy, overpaid beneficiaries of bloated government make-work).  Because, according to the free-market orthodoxy of conservative/libertarian religion, the costliest single resource that just about every organization must account for in pricing its products and services is its employees, and therefore, the proper target of our anger, resentment, and downright hatred about the costliness of the products and services we pay (too much) for is other people.  The editors of the paper would never say this, of course, because that would make them sound like sociopaths. 

The editorial also implies that Democratic (you know, the party that always wants to spend more taxpayer money, especially when it can be spent generating government-dependent voters, securing permanent electoral majorities for Democrats) Commission Chairman Sean Logan is dragging his feet in the implementation of automated toll collection.  But the actual timetable for the study was established during the Corbett administration, years before Logan became chairman or Tom Wolfe became governor.  Not only that, but the plan to automate the toll collection system was already on the drawing board in 2012.  In short, the foot-dragging accusation is bullshit.  Logan has expressed some scepticism about the long-term cost savings of a fully-automated system.  You could attribute that to a pro-union bias, but an article in the Tribune-Review -- which I suppose is the factual basis for the editorial -- points out that Colorado's experience with automated tolling has not realized the cost savings they anticipated.

None of this is to say that fully automating toll collection on the Turnpike is a bad idea.  There are plenty of good reasons to do so, as pointed out here and here, for example.  Converting to a fully-automated system will most likely reduce toll-collection costs eventually, if done correctly.   The commission anticipates upfront conversion costs of about $320 million.  One article estimates that about $70 million in annual costs will be saved by eliminating the jobs of toll collectors.  Taxpayers further benefit in several ways.  They avoid slowdowns at toll booths, auto emissions are reduced, engineering connections to other roadways are simplified, some physical maintenance costs are reduced.  These savings are offset by the higher ongoing cost of the toll collection technologies and higher enforcement costs.  I could not find a detailed estimate of how these costs balance out over the long term, but it will probably result in the savings of millions of dollars annually.

Now for the larger question that this brief editorial leaves unanswered.  Or, to be quite frank, trivializes.  I've seen different numbers, but it appears that an estimated 600-800 toll booth collector positions will be eliminated.  The Turnpike Commission has been talking about this project with their employees to prepare them for the eventual job reductions.  The commission hopes to transfer some of these employees to the enforcement positions that will need to be filled as part of the conversion.  Other jobs simply won't be filled as current employees retire.  For the rest the commission is likely to offer other incentives to find work elsewhere.

This relatively small workforce reduction is one example of a larger trend that has been going on for a long time and will likely accelerate in the coming decades.   Automation has been replacing human labor for a long time now.  In most cases, economies that invested in labor-saving devices eventually expanded enough to create new forms of employment that absorbed the excess labor displaced by automation.  Unfortunately, this did not always happen in the right magnitude, with the right speed, or with compensating jobs that matched the abilities and skills of displaced workers.  As a result, there were large numbers of losers, people whose livelihoods were ruined and as a result were plunged into poverty or at least a much lower income level.  This was especially the case before national governments adjusted better to the problems caused by worker displacement.  Modern, post-industrial societies tend to do a better job helping displaced workers transition to new opportunities.  But better is just better, not good in many cases.  

Many leaders view improved educational and training programs as the key to helping displaced workers quickly transition to new employment in a high-tech economy.  This approach assumes that automation mostly replaces rather low-skill workers while opening up new opportunities for more highly-trained workers.  In general this is true, although the number of highly-skilled workers needed to maintain a more highly-automated process is usually far less than the number of low-skill workers replaced by the automation.  Many economists argue that the difference is made up indirectly, by future investments of the money saved via automation in other economic activities from which displaced workers obtain new employment.  Whatever its merits, this arrangement only works when automation replaces low-skilled workers.  When automation is able to replace highly-trained, highly-skilled, or experienced workers, improved and training and education will not by themselves enable many displaced workers to find new employment.  First, some of these workers may simply lack the native talents needed to succeed at a higher level of intellectual effort.  Second, as the level of skills and knowledge needed to outcompete machines rises, so does the time and money needed for a displaced worker to obtain the needed retraining.

We are fast approaching a new phase in workplace automation in which the machines are becoming capable of replacing knowledge workers.  This is already happening in information technology, although only to a very limited degree.  Information security anomaly detection and incident response, automated legal discovery, algorithmic optimization are just a few of the areas that have become increasingly automated, reducing the number of human beings needed to fulfill such roles.  Examples like this could be repeated for any number of other fields usually classified as knowledge-focused or knowledge-intensive.   It is true that up until now, no large numbers of knowledge workers are losing their jobs to machines.  Instead, what is happening is that as AI and related technologies make incremental progress, human knowledge workers have been gaining in productivity.  The machines have served as very sophisticated tools.  What has changed is that many entry-level or low-level positions are now filled by machines.  The result is a net loss of job opportunities for new workers

Artificial Intelligence researchers may not be able to produce machines capable of human-level intelligence for decades, or more.  But incremental progress along these lines is likely to continue, and as it does it should eat up more and more of the upper-level employment opportunities that white-collar workers assumed would be safe from automation.  When machines do attain human-level intelligence, they are bound to quickly surpass it as well.  From a purely market-based perspective, at that point most highly-valuable work will be altogether beyond human capabilities, and the occupations humans are capable of will be accomplished more effectively and cheaply by machines.  In short, we become virtually unemployable, no matter how smart and well-educated we are. 

What to do about this?  You can always make the argument that it's not our problem.  Some future generation will have to deal with it when the time arrives.  Really, that is precisely the attitude taken by the editorial board in this case.  Toll collectors losing their jobs?  Not my problem.   Trouble is, if it isn't the editorial board's problem now, it will be sooner or later.

So, to the editorial board of the Leader Times: There are already automated systems that have become capable of writing basic financial news articles  that to an unsuspecting reader appear to be written by a human reporter.  Research and development along these lines will not stop in order to save the jobs of free-market zealots.  Those zealots may hope that the fawning they have done at the knees of the 1%'ers who pay their salaries will save them from the unemployment line, but I doubt it.  The machines are coming for your jobs and mine, editors. 

We can act now to prepare for this eventuality.  This is not a natural disaster; it can be averted or remedied by economic policy changes.  One change, hinted at by this open letter, involves changing the rules of capital ownership (Gasp!).   If current trends continue those who own the machines or the businesses using machines will see their wealth grow, while the rest of society slips into poverty.    (Well, it would continue that way until the machines surpass human intelligence.  At that point they may decide to thrust their owners into poverty or worse along with the rest of us.)   Clearly, only the most immoral, committed Randian would find this outcome appealing.   Would the editorial board be willing to engage in examining economic policy alternatives that would reverse our current slide?

 None of this has ever perturbed the editorial board of the Leader Times enough to broach the subject before ... why should it?  "Who gives a f*ck about union members, most of them probably Democrats, losing their jobs?  Why don't they all just move to Sweden and suck from the government teet over there?"  Like I said, sociopaths.  Editors, are you listening?  Do you realize that your excessive admiration for free-market principles not only makes you sound like sociopaths, it makes you act like them?