The End of Computing

The End of Computing.  We put forth the question as to the end of computing.  That is, we ask when will computing and computers come to their end of innovative applications, though this is not a discussion about bigger and faster machines.  Sure, bigger, faster computers can and will push to new limits ordinary and well explored topics.   They have this, and will so continue.  We are entered into a discussion about the use of computers to solve new, even revolutionary, problems  of this world. 

Examples of innovations now at the end of their road.  Of course, these examples may simply reveal this author’s lack of futuristic insights. 

·         Watch making – long the epitome of machines, the watch is now engineered with precision and at least mechanically do just about everything ever desired – extremely accurate time keeping.  Even still there has evolved a new technology for this task.

·        

 The horizontal milling machine- Just about everything a milling machine can do has been done.  It is a complex but simple machine that has its mission well defined and explored.

·        

 Data storage – we now have Exabyte storage capacities.  Just about everything you do and I do can be stored in real time – and maybe is.  The new problems may involve making sense of this data, bringing it to reign and then meaning.

·          

Time – With the advent of the Bulova Accutron, using a tuning fork, and then the quartz crystal, accurate timekeeping now at the atomic level is basically complete.  There is likelihood that no application will ever require any measurements more accurate than we have now.

·         Art.  The techniques of painting have been developed to such heights that modern artists must change the message just to gain any attention.  There are likely no current artists that can even closely match the technique of the masters of only a couple of centuries ago.  Thus, the birth of modern art – including many genres.  Artists have learned the importance of moving the target to hit the mark.

Where are we at?  Computing furnishes us with a generalized tool for doing new things, though only things for which quantification, numbers, and information are at play.  Practitioners have now spent several decades devising ways that qualitative data and information can be made quantitative.  We say some decades because that is the currently longevity of computing machines.  As usual, it was the real world that spawned the need for machine encoding and processing of large quantities of data.  Beginning with the complexities of processing census data, Herman Hollerith (1860 –1929) devised the punched-hole card that could be machine read and input data analyzed using mechanically complicated but pre-electronic simple machines.   This was years before the computer, but note that complex processing machines were developed years before the computer as we know it.   Indeed, the most striking aspect of computers was the concept of the stored program and now adaptive program.  

Our mission here is not to recount the history of computing, but rather to suggest where it is going, and importantly where and when it may reach an end.   Yet it is important to note that the idea of computers evolved from the works of Cantor and Gödel on the incompleteness of our mathematical systems which concluded there are problems within any mathematical system that cannot be solved.   This is to say, that within any logical system, there must be impossible problems solvable within it.  The implication is the system must be expanded, but then the expansion will lead to new impossible problems.  This is a spiral situation.    It may be best to get back to computing. 

When asking when computing is at an end we are discussing whether computing can resolve any question we need to answer.   Not simplistic questions as implied by “Do I love you?” Such questions are emotional and may never have any meaningful answer as the answer, if so posited, may change by the moment.  Such simple questions are virtually impossible by constructs of the human system.

Numerics – The world functions on mathematical models of reality and essentially what they predict, what they can show, and importantly what they can solve.   Astonishingly successful, numerics generate a wealth of information and lead to solutions of problems impossible to consider, if not pose, well beyond the scope of all even two centuries ago.

Data – The accumulation of population, climate, weather, and other statistical data has amassed at such a rate and have exceeded all estimable bounds that it cannot be any longer processed without computers to help.  Just imagine a large bank trying to keep track of its customers with index cards and phone numbers on a Rolodex.   Below, we take up the similar-sounding topic of “big data,” quite another issue.

  

Modeling – The most successful technique for the explanation of phenomena, following Sir  Issac Newton’s theory of planetary motion, has been the model.   Clearly, this was a mathematical model, but the idea has caught on – despite philosophical debate.  We now have all sorts of model s including the statistical models derived from data sets.   Most mathematical models have no tight, closed solution coming by way of a formula. Most problems involving these models required intense numerical computing.  Indeed, in the past four decades an entire field of numerical analysis has evolved simply to help provide solutions to such problems.   Statistical models are models generated from small and more recently massive data sets to correlate and thus to explain cause and effect.

Big data – This is the newest application for computing.  It involves massive amounts of data with the goal to determine patterns within.  There are specialized software (RapidMiner, SAS, etc) designed to do just that.  Remarkable results have been achieved.  Patterns in medicine, in finance, in education are just a few of the bigger topics that have been rendered to undiscovered and even unconsidered conclusions.  Perhaps this is the end of computing, or is it just the beginning.  The tools for analysis follow rather standard constructive models.  When one is found there is celebration and delight.  It is then used for predictive ends.  Often the models are used to predict backwards in time to validate their predictions toward the future.  Herein lies the issue.  Unless the reality is time-reversible,  validation in the manner is suspect.  Indeed, it can lead to conclusions that are simply false.   We could call this a modeling false positive.  

Here is one new example.  In a Cornell study of Facebook pages of more than one million participants, there has evolve though big data analysis a predictive of relationship breakups  The study suggests that their models can predict a breakup before even the participants are so aware.  They show, though big data modeling, that If you both all have the same set of friends, this is an indicator of a possible breakup.  But this is just a model – NOTHING MORE.  In fact, there may be other factors of the personality types of people having same sets of friends. What conclusions should be determined?  See, http://www.mirror.co.uk/news/technology-science/technology/facebook-can-predict-your-relationship-2659859

Conclusions.  Given in the form of questions, our conclusions are not even that; they are just suggestions of possible futures ahead. 

A.      Can it be that computing and computers is one of those elusive techniques and machines that adapt endlessly to ever new and challenging problems – without end?  This positions computers to replace humanity as perhaps some sci-fi writers have called our next big step of evolution.

B.      Can it be that we will soon (whatever that means) reach  the limits of computing, and a new technology or concept will emerge to offer solutions to ever more impossible problems?

C.      Can it be that modeling will create a world of lemmings, wherein we see a rush to the latest conclusion, which in turn rush to a new conclusion based on systemic changes.  

D.      Can it be that the tried and tested “cause and effect” paradigms are to be replaced by degrees of correlation?

We are decades, if not centuries, from the end of computing.  Yet the co-mixture of numerics with modeling of big data gives on the one hand assurances of defined reality and on the other hand the predictive modeling based on data sets.  This creates an internal conflict.  The dark side is what happens if society becomes overly intoxicated by its models, and begins to reinterpret them as facts, much less truths.  This is scary.