The Complexity Factor

	Sign In Sign-Up

By Ed Meagher

------------------------------------------------------------------------

Much has been written about the potential effects of the World's computers inability to calculate properly (or at all) in the face of the rollover from 1999 to 2000. The problem is usually explained in a single paragraph or two as being the result of decisions innocently made several decades ago to save precious computer memory and storage. These decisions resulted in ad hoc conventions and protocols that mandated a six character date representation long after the need to conserve memory disappeared. The logic that was applied in nearly all of the computer programs written since then assumes the two "missing" century characters to be "19" and when the year 2000 arrives (or calculations concerning the year 2000 arise) the logic of these programs will be wrong.

The debate that ensues is usually technical, boring, and unresolved. This is actually reassuring because it lends support to the notion that this problem is actually a business problem and not a technical one and that the issues that are being discussed are not the real issues and therefore unresolvable. The debate usually focuses on a single company or industry and the fact that if proper attention is paid to the problem certainly this company or industry will be able to solve their problem in time to avert any serious problem. I would liken it to the task of catching your dinner at a turkey ranch. Certainly catching a single bird in a huge flock is not too difficult. Even catching a specific bird, while more challenging is achievable if sufficient attention is paid and certain skills are applied.

The problem arises when everyone has to catch their dinner within a very restricted time frame, each has to capture a specific bird, and all have to share limited, specific skill sets. While this analogy is useful it doesn't completely express the difficulty of the task. This is where the complexity factor enters the calculation.

One of the biggest difficulties in getting people to understand and appreciate the threat posed by the year 2000 computer date problem is the apparent simplicity of the problem. Six characters versus eight characters. 06/12/97 versus 06/12/1997. Add the two characters and be done with it. It is counter intuitive that this could be difficult. Even given that there are hundreds of billions of lines of code in millions of different organizations that need to be searched to find the 3 or 4 percent that are date related, it would seem that this could be automated and accomplished rapidly. Even the complications of old, undocumented code written in over 2500 different computer languages and executed on thousands of different hardware platforms being controlled by hundreds of different operating systems should be manageable. Further complexity in the form of billions of six character date fields stored in millions of databases that are used in calculations while daunting is not insurmountable or so it would seem. At this level we are still simply faced with a business problem potentially resolvable by automated tools and effective management. The find and capture your turkey phase.

Now compound this by the fact that under 800 days remain before the Year 2000, that 150 of those days are weekends or holidays and that until an emergency is declared 40 to 50 additional days will be spent on vacation or sick leave and the number of available work days is under 600. Reflect on the well documented, 30 year track record of the average American programmer in terms of productivity and accuracy and then perform the requisite mathematics and the first inkling of the "you can't get there from here" realization begins to dawn. These concerns can be somewhat mitigated by claims of improved productivity and accuracy based on new, untested automated tools and methods but it must be remembered that these are only claims made by the developers and owners of these new tools and that they have not been tested in any meaningful way. Still at worst this represents the everybody trying to capture a specific turkey at roughly the same time phase.

In fact the problem is more complex than that. During the early part of the 1980's as a result of the adoption of "Reagan Economics" it was appreciated that the only part of the productivity equation of raw material, energy, and labor that could be readily controlled was labor. Rather than simply slash labor from the equation it became the goal to make labor more productive. Breaking the power of unions and adopting more efficient labor models were but two of the methods employed. By far the most successful approach was the coupling of the power of the rapidly evolving computer with limited amounts of more highly specialized labor. This approach succeeded more rapidly and completely than even its' most ardent proponents could have imagined. While major disruptions to the labor force occurred and continue, the Information Age Revolution has resulted and the health and well being of our economy and our society are totally dependent upon it.

While knowledge of this revolution is accessible to anyone who cares to read and consider such matters the implications of this dramatic change to our economy and society as a whole are less well understood or appreciated. The first phase of this revolution involved giving workers access to databases and preprogrammed computing power to make them more efficient and eventually requiring fewer workers. As computers got faster, cheaper, and more powerful the bottleneck became the human element. The next phase in this evolution involved the use of networks to allow computers to communicate with each other, further reducing the need for human intervention. Inexorably this has led to computer systems being linked to multiple other computer systems which require little or no human involvement.

This process of "systems integration" occurred over the course of the last 15 years spawning the emergence and growth of an entire industry. Hundreds of billions of dollars and millions and millions of man years have been spent integrating and reintegrating these multiple complex systems to form highly interdependent meta- systems that have resulted in productivity gains for the highly paid information worker that makes them competitive with lower skilled, lower paid workers anywhere in the world. This tightly integrated economy has been a 15 year success story. This success story is what is at the heart of my belief that these two "missing" characters are going to lead to major disruptions in our economy and potentially devastating consequences for our society. How can this be so ?

Back to my belabored analogy. This integration of systems can be likened to the numbering of the turkeys. Amidst the capturing of specific turkeys in a fixed and unchangeable time frame add the requirement that they be captured in a specific sequence. This roughly equates to the requirement to dis-integrate and then re-integrate these complex systems. Finally add the notion that there is no single entity keeping track of the capture sequence and no single rule or set of rules governing the how of the capture or even what constitutes a capture. Only now can the true complexity of the task at hand be appreciated. Even sadder is my belief that this convoluted analogy only represents the situation in the United States. A far darker analogy would have to be conjured up to represent the situation representative of the world economy.

And as bad as this situation might be, it is not what I worry about. With less then 800 days left and no evidence of any type of magic bullet solution, I am still enough of an optimist to believe that some how, some way, methods will be found to solve much of this problem. I really have no clue what these solutions might be and I am confident that they don't yet exist but I am humble enough to allow the possibility that something may be found to solve the majority of these problems. My concern focuses on the margin of error that we have allowed ourselves in any of these situations. It is not enough to solve simply "most of these problems". The integration of these systems requires that we solve virtually all of them. Our ability as an economy and as a society to deal with disruptions and breakdowns in our critical systems is minuscule. Our worst case scenarios have never envisioned multiple, parallel systemic failures. Just in time inventory has led to just in time provisioning. Costs have been squeezed out of all of our critical infrastructure systems repeatedly over time based on the ubiquity and reliability of these integrated systems. The human factor, found costly, slow, and less reliable has been purged over time from our systems. Single, simple failures can be dealt with, complex, multiple failures have been considered too remote a possibility and therefore too expensive to plan for.

The only place serious consideration has been given to this problem is in the military contingency planning in regard to Information Warfare. The unclassified reports of these planning exercises on the part of the military establishment are illuminating. These reports assume hostile forces launch an attack on our nations information infrastructure resources. These attacks take the form of overt military attacks as well as covert, computer based assaults on these systems. The bottom line assessment attached to these reports is that our systems are vulnerable to these types of attacks, that they are becoming more vulnerable as a factor of increased sophistication on the part of our systems and the availability of more sophisticated attack systems, and that the ability to disrupt our economy and our society as well as our defense capability is greatly increased as a result of the highly integrated nature of our systems infrastructure. This enhanced integration of systems increases the risk of "collateral damage" to systems, that is, damage caused to systems not directly attacked but damaged due to their reliance on information from damaged systems.

These Information Warfare assessments all assume an external threat either from conventional forces such as attacks on power plants and transportation facilities or electronic attacks from digital terrorists or felonious hackers. They describe in chilling detail the cascading negative impact of these attacks on transportation, communications, and public safety. Scenarios describing the effect on food distribution resulting from disruption to the transportation scheduling systems are played out. Within 3 days there is limited fresh food available anywhere in the United States. Six days later there is virtually no food at all at a retail level. Similar scenarios are played out for each of the major digital infrastructure systems. What is not mentioned is the obvious. That all of these scenarios apply to an internal threat on these same systems. Failures due to programs, databases, and systems being unable to process accurately or at all due to date miscalculations or date related logic failures will have the same devastating effect as a bomb or a computer virus.

The dilemma is not the two characters that represent the century in dates stored in computers. The problem is that we as an economy and a society have become totally dependent on millions of unseen computer systems. Our standard of living and in some cases our very well being have become dependent on complex computer systems operating, within limited margins for error, reliably and accurately, without human intervention and with very limited human oversight. No conscious decision was ever made by anyone to get to this point. It occurred over decades, a simple, straight forward decision at a time. Each decision being reasonable and justified. No one ever said. "Hey, we are becoming overly dependent on computers." or " "This decision has far reaching consequences and will eventually result in a situation where we as a society will be dependent on these computers operating flawlessly." Each decision that has led us to this point was limited, reasonable, logical, defensible. Only the most philosophical among us saw the implications of this digital revolution and few heard or listened to them. It has taken this absurdly simple problem to bring it to our attention.

Consider the last time you dialed a business to ask a question. Did you get a human being or a recorded message walking you through, "...if you want to order an item press 2, to check on an order press 3...". Tomorrow morning as you go to work be mindful of what is occurring around you. The computers that control the traffic lights, the electric power generating plants, the cars, the gas pumps, the elevators. When you get to work think about the phones, the heating or air conditioning, the email, the internet. Think about the reports and information you rely upon. On the way home think about the banks, the hospitals, the police and fire departments you depend on to do their job. What degree of inconvenience would result if one or two of these services were limited or absent? That's the worst that has ever happened to us. What would the consequences be if all or most or just many of these services were limited? What would you do ? Who would you call ? What would you use to call them with? We have become so trusting of these systems that we will stare in disbelief at a light switch, repeatedly cycling it on and off, if the lights don't come on instantly every time we beckon. Who hasn't repeatedly pushed the elevator button if the door doesn't open immediately?

These are the systems that it is possible "to see". What isn't known or appreciated are the far more complex systems that operate in the background of our digital lives. The banking system, the business systems, the healthcare systems, the communications systems, the transportation systems that make our economy run smoothly. These systems have the highest degree of integration. Traditional boundaries between businesses or governments or nations have been blurred or erased altogether. We are an on-line society, an on-line economy, an on-line government and in fact an on-line World. KMart and General Motors require their thousands of suppliers to have their systems on-line to them. Orders pass between companies, bills are submitted and paid, parts are scheduled for shipment, inventories are adjusted and finished products are bought and shipped all without human intervention. The manual processes that could accomplish any of these functions has long since been eliminated and the human beings required have been down-sized, right-sized, or re-engineered. The reliability of these individual systems has been so remarkable that backup or contingency planning has become an academic exercise at best. Even if manual processes can be identified and human beings found to implement them where is the infrastructure to support them? The banks of phones, the desks and chairs, the offices to house them? Who will you call to order these items? Will their systems be operating? Will they be overwhelmed by everyone else trying to reinvent manual, human based processes?

So the problem is not a technical one, a simple insertion of the requisite century information. The problem is a societal one. We have allowed our digital slaves to become essential to our well being. We have been lulled into believing that they will always be there doing our bidding. We are in for a very painful surprise in less than eight hundred days.