Home       Articles / Books    Projects     Barnett Consulting     Weblog

Year 2000 International Security Dimension Project Report

Explaining Our X-Y Axis

Our X-Y Axis (shown below as Slide 4) begins with two simple questions:

• Horizontal axis asks the "What?" question: What is the nature of the Y2K Event?
• Vertical axis asks the "So What?" question:  What is the impact of the Y2K Event?

There is a huge difference between these two questions, for the first question focuses on cause, while the latter focuses on effect.

One way we like to differentiate between the two questions is to employ a medical analogy.  Think of the horizontal axis (What? question) as the nature of the trauma or illness and the vertical axis ("So What? question) as the patient's overall health.  Two extreme examples show why this analogy is illuminating:

• Example 1 is an elderly man who is stricken with a very slow growing bladder cancer.  While this elderly man could have lived with this cancer for several years, the stress of his hospitalization, exploratory surgery, and the frightening diagnosis stresses his already fragile system to the point where he suffers a stroke and is dead within two weeks as a result of major organ failures cascading throughout his system.  To sum up, while the initiating event (bladder cancer) was more minor than major (placing it on the left side of the horizontal access below), the man's overall system robustness was weak (placing him on the lower side of the vertical axis).  The medical outcome was--irrespective of its modest origins--disastrous.
• Example 2 is a two-year-old child struck with a very aggressive kidney cancer that--by the time of diagnosis--has spread to both her lungs.  Other than that, though, the child is in excellent health, and as such, is more than able to survive the surgeries, radiation, and months of chemotherapy with no lasting negative effects of clinical value.  To sum up the child's case, while the initiating event (kidney cancer) was more major than minor (placing it on the right side of the horizontal axis), the child's overall system robustness was strong (placing her on the higher side of the vertical axis).  The medical outcome was--again, irrespective of its profound origins--quite positive.

These two very different medical case histories, drawn from the author's family history, highlight the importance of juxtaposing the "What?" and "So What?" questions to create the four quadrants of the X-Y axis, for it is not enough simply to ask how bad Y2K may be.  Given how bad it may be (i.e., how many computerized systems fail), Y2K's ultimate impact will depend greatly on the targeted system(s) in question.
 
 

Looking at Slide 4, we then explain our X-Y Axis as follows:

• The horizontal axis, asking the "What?" question of the Y2K Event, posits the minor extreme on the left as being "Y2K events are discrete and episodic" and the major extreme on the right as being "Y2K event is widespread and sustained."
• The vertical axis, asking the "So What?" question of Y2K's impact, posits the minor extreme on top as being "Systems are robust," and the major extreme on the bottom as being "Systems are vulnerable."

Two caveats are in order:

• By "Y2K Event(s)," we refer only to network failures directly attributed to Y2K or those caused via subsequent cascading system failures, to exclude any social, economic, or political responses that exacerbate or reduce failure rates.
• By "Systems," we refer not only to a country's network systems (broadly defined to mean any network that moves something--e.g., bytes, people, electricity),  but also its political, economic and social systems, with the key attributes of robustness being:
  • Distributiveness
  • Recovery capacity
  • "Workarounds" capacity
  • Trust "capital."

Having defined the extremes of our axes, we break down the four quadrants in the following manner:

• Best Case is when Y2K events are discrete and episodic and systems are robust
• Next Best Case is when the Y2K event is widespread and sustained, but systems are robust
• Next Worst Case is when Y2K events are discrete and episodic, but systems are vulnerable
• Worst Case is when the Y2K event is widespread and sustained and systems are vulnerable.

Y2K Onset Model #1: The Ice Storm

The Ice Storm onset model is depicted in Slide 5 below.

In the embedded chart, the vertical axis defines a "field of Y2K failures," meaning we're not going to offer any percentages or "hard numbers" here, just a rough notion of overall failure saturation.  Along the vertical axis we display the years 1999 through 2001, with the months of 1999 noted in solid-line marks and the months of 2000 noted in dashed-line marks.  The difference between the two markings is meant to suggest that while we may feel we have a firm grasp of appropriate time units for the timeline leading up to 010100, perceptions of time's passing once we pass through the 010100 threshold may vary greatly depending on locale.  For example, the subjective time unit of note for Wall Street at the beginning of January may be the first day of trading--a mere several hours' time, whereas the subjective time unit of note for a sheep herder in a less developed country may be as long as until the first time he brings his sheep to market--possibly several weeks.
 
 

The Ice Storm onset model offers the classic, TEOTWAWKI view of Y2K: it hits en masse on or about 010100 and strikes virtually every aspect of society.  To the extent that such a model may seem to hold true on a perceptual basis in any one locality or region (meaning, for all practical purposes, it seems as though all systems are impacted to some disabling degree), we posit that the Ice Storm's components are logically broken down into three categories:

• Direct Y2K failures
• Cascading system failures resulting from the direct failures
• Iatrogenic crisis management or social responses that exacerbate the cascading failures or trigger new threads.

While this model held implicit sway during much of the Y2K debate in 1998, it has receded in prominence over the course of 1999, as remediation efforts make clear that this is not a useful universal model.  Having said that, however, we believe the model retains great validity for understanding pockets of significantly damaging Y2K impact that may occur around the world, meaning those areas where--for all practical purposes--the TEOTWAWKI notion may well emerge among significant portions of a population battered by widespread network failures.

Of course, even here we're still talking only about the perceived onset, and not some sustained environmental status that would realistically drag on for months.  As such, the key question for the Ice Storm onset model is, "How fast can the society or economy in question recover by necking down the failure rate to some level commensurate with reasonably sub-optimal functioning (meaning, for many around the world, the return to "life as we know it")?"
 

Y2K Onset Model #2: The Flood

The Flood onset model is depicted in Slide 6 below.
 

The Flood onset model depicts a slow but inexorable bulge of network failures that first rises above the usual "background noise" level on or about 010100 and then expands for something in the range of the first six months of 2000, peaking near the end of the 2nd Quarter or at some point in the 3rd Quarter.  In some ways, we could suppose the same breakdown of elements (direct, cascading, iatrogenic) here as with the Ice Storm model, but because of the greatly extended timeline (thus allowing for more effective crisis management and network triage), we limit our description here to direct and cascading network failures, thus positing a peak failure rate somewhere in the range of 50 percent of all networks.

As such, the Flood model gets nowhere near the TEOTWAWKI pain range, but instead describes something more akin to a significant economic downturn, most likely corresponding to popular perceptions of a recession or financial market "correction."   In that manner, the Flood  model possibly describes a more profound economic impact than the Ice Storm, which, while it is a shock to the system, is probably of shorter duration.  So, like the Ice Storm, the Flood model involves an interrelated sequence of network failures, albeit with a far smaller immediate impact on the overall functioning of society.

In keeping with the weather analogy, the key question for the Flood model is, "What constitutes a 'low-lying area?'"  One example of a potential low-lying area would be manufacturing, whose network failures would not likely be centered on the 010100 threshold, but rather build up over time as production continued throughout 2000.  Another could be medical supplies, especially the production and distribution of key pharmaceuticals.  Still another might be the processing and distribution of clean drinking water.
 

Y2K Onset Model #3: The Hurricanes

The Hurricanes onset model is depicted in Slide 7 below.
 
 

The Hurricanes onset model presents a series of sectorally-limited (meaning unconnected across sectors) but relatively lengthy (meaning some cascading effect) constellations of network failures.  In effect, this model is a hybrid of the Ice Storm and Flood models.  The Hurricanes model packs the same immediate punch as the Ice Storm model, albeit in isolated "low-lying areas" (echoing the Flood model), thus limiting the overall impact on the functioning of a society.

The Hurricanes model speaks more to the "winners and losers" approach to thinking about Y2K's ultimate impact:  some sectors of society will seemingly get off scot-free, while others will seemingly suffer great damage.  The key difference with the Flood model is the lack of interrelation and simultaneity, so rather than employing the economic language of "downturns," we're more likely to describe "shake-ups" in one or another industry.

The same approach to identifying vulnerable sectors that one uses with the Flood model would apply here, although in an overall sense, the Hurricanes model is probably best used to think about countries whose remediation efforts have been weak, for here we run into the notion of over-confidence possibly leading to poor crisis management preparation.  If such "poor remediators" turn out to be far more vulnerable than they realize, then the key question becomes, "How can coordinated triage and crisis management avert the appearance of a critical mass of substantial--yet still relatively isolated--network failure clusters?"
 

Y2K Onset Model #4: The Tornados

The Tornados onset model is depicted in Slide 8 below.
 
 

The Tornados onset model refers to a "season" of sectorally- and temporally-limited Y2K-induced network failures.  This model is the closest to a null hypothesis of Y2K's overall impact, for, in many ways, it describes life as we know it, albeit with a higher-than-average failure rate.  The Tornados model can likewise be thought of as the "key dates" model, for the two go naturally hand-in-hand when one seeks real-world evidence of significant network failures that either produce serious disruptions of service or require extraordinary efforts at repair.  For if such key dates come and go without displaying any significant failures, meaning they're so big they can't be hidden by the service providers in question, then these Y2K milestones pass by without registering significant values on any sort of TEOTWAWKI scale, becoming the Y2K equivalent of a "tree crashing in the forest when no one's there to hear it."

The "key dates" approach does correspond nicely with the Gartner Group's predictions of Y2K failure rates rising and falling over the course of 1999 and through the year 2001, but the big deficiency of this model to date has been the lack of any stunning failures on key dates that have already passed.   For example, no failures featuring major negative impact occurred on 1 or 3 January, the first day and business day, respectively, of 1999.  The start of many fiscal year programs on 1 April also failed to reveal any serious disruptions for the governments involved.  The so-called "nines" problem that was slated to appear on 9 April likewise produced no failures of great societal value in any country around the planet.  Most recently, the 1 July threshold came and went with no apparent damage to the 46 U.S. states whose fiscal years began that day.

Meanwhile, Cap Gemini America, the computer consulting firm, declares on the basis of their recent survey of  Fortune 500 companies and a smattering of U.S. government agencies that close to three-quarters of the respondents report experiencing a Y2K-related failure through the first quarter of 1999.  But if these firms are having these failures and none are making any headlines, how is that much different from everyday life as we know it?  Aren't private firms and government agencies experiencing network problems on a fairly regular basis, and just as regularly keeping such failures under wraps?  The key missing data involve how much different 1999 is turning out to be compared to any previous year, meaning what is the "instability added" from Y2K?  And that's the data we haven't found anywhere yet.

Having said that, the key question for the Tornados model remains, "What constitutes good learning over time?"  For example, should our confidence grow due to the lack of Y2K headlines stemming from the key dates already passed?  Or should we ignore most if not all of that success, especially for a pure fellow traveler such as the "nines" problem?   After all, we can get fixated on Y2K key dates all through 1999, get through them all quite nicely, and still suffer significant tumult on 010100.  Uneventful key dates make that seem less likely, but don't rule out it out by any means.
 

Onset Models Leading to Generic Y2K Outcome Scenarios

Of course, none of the four onset models are likely to hold sway for any one region's entire Y2K experience, and in that sense, we are likely to see versions of all four models occurring simultaneously around the planet at various points in the Y2K Event.  As ideal types, the four models are designed to help the reader disaggregate the complexity presented by Y2K's myriad of possibilities, rather than provide a "pick one of four" analytical choice that would invariably prove false and pointless.
 
 

Slide 9 above arrays the four onset models on our X-Y axis, and the placement should seem fairly intuitive given our descriptions:

• Tornados represent the "Y2K events as discrete and episodic" and "systems are robust" quadrant, meaning a season of relatively isolated and concentrated damage that follows little rhyme nor reason to the extent that we can trace causality.
• Flood represents the "Y2K event as widespread and sustained" but "systems are robust" quadrant, meaning a rising tide or deluge of damage that follows the logic of systematic vulnerability, i.e., the low-lying areas analogy.
• Hurricanes represent the "Y2K events as discrete and episodic" but "systems are vulnerable" quadrant, meaning a season of somewhat isolated but wide-swath damage that follows the logic of either poor remediation or unforeseen vulnerabilities--basically one in the same.
• Ice Storm represents the "Y2K event as widespread and sustained" and "systems are vulnerable" quadrant, meaning a seemingly pervasive or all encompassing damage pattern that is inescapable, but one that at least reveals itself in its entirely with great speed, thus facilitating recovery.

Again, our rationale in presenting such onset models is not to encourage a "pick one" mentality, but rather to break down the abstract nature of the potentially universal problem set into a series of weather analogies that are far more easily understood by the average citizen--not to mention your average elite decision maker.

Slide 10 below presents a series of outcome-focused Y2K scenario titles arrayed along our X-Y axis.  By pairing them up with our onset models, we--in effect--offer a "coming and going" view of the Y2K Event (leaving the "guts" of our Y2K analysis for the section on Scenario Dynamics).
 
 

• Run of the "Mille" refers to the Best Case Scenario, meaning Y2K comes in bits and pieces and we prove far more robust than we give ourselves credit for.  So it's "run of the mill" in that we take Y2K in stride, but Run of the "Mille"  in the sense that the Millennial Date Change Event still exists at the core of the Y2K null hypothesis.  Thus, in this scenario, whatever social instability occurs around the 010100 threshold is more driven by millennial elements (e.g., apocalyptic-driven behavior, world's largest party, great religious feast) than by actual Y2K-driven network failures. Humanity emerges on the far side of this "crisis" wondering what all the hype was about.
• "Humans 1, Computers 0" refers to the Next Best Case Scenario, meaning Y2K is big and bad but we weather the deluge of failures and only the systematically weak are left with permanent damage.  This is the Nietzschean social scenario that says, that which does not collectively kill us, makes us collectively stronger.  Humanity emerges on the far side of this crisis with a renewed confidence vis-a-vis the invisible and pervasive information technology that "seems" to control so much of our lives.
• "Houston, We Have a Problem" refers to the Next Worse Case Scenario, meaning Y2K comes in bits and pieces but we are surprised to realize how fragile our systems are.  Like the Apollo 13 mission from which this quote was drawn, it seems as though relatively minor weaknesses--the IT-equivalent of an Achilles' heel--sequentially disable many sectors of society with ferocity, leading to cascading failures that can threaten the sum of the whole.  Humanity emerges on the far side of this crisis split into winners and losers, meaning--respectively--those who proved resilient and those whose weaknesses were exposed.  In some ways, Y2K will unfold something like a computer virus: those with sufficient immunity will survive just fine, while those with weakened immune systems will suffer catastrophically.
• "Y2 KO!" refers to the Worst Case Scenario, meaning Y2K is big and bad and we're far more vulnerable than we realized.  We are collectively "knocked to the mat," with the real uncertainty being, do we get back up before the "referee" finishes his "count?"  Or do we lie there prostrate, dazed and confused?  Of course, at some point we do get up, and how humanity emerges on the far side of this crisis is largely determined by the nature of the "knockout."  Is Y2K merely a "TKO," meaning a "knockout" attributed solely to "technical" failures?  Or is Y2K a genuine "whupping" where all our systems (political, economic, social, and network) fail us miserably?  In other words, are we merely embarrassed and so continue on as before?  Or are we truly humbled and thus serious changes result?

Potential Y2K Impact by Country Groups: Conventional Wisdom Has Changed Over Time

The conventional wisdom on which countries around the world are more vulnerable to Y2K has changed dramatically over the past year.  We display our interpretation of this changing debate in the following two slides.

First, a word on how we break down the world into four IT categories:

• We define an "Ultra-Modern" IT category as including only the United States, which, by all measures, stands head and shoulders above the rest of the planet in terms of IT adoption rates.  To put it bluntly, there's no way Y2K will be bad enough to derail the US's progressive adoption of IT.  There's simply no going back.
• A "Modern" IT category basically captures the rest of the OECD-type states (Organization for Economic Cooperation and Development) such as Japan, Germany, France, etc.  These economies tend to be relatively distributed in terms of networks, but not nearly as "New Economy" in outlook or practice as the U.S.  Like the U.S., these countries are unlikely to see the further adoption of IT derailed by Y2K, although it could greatly influence some of the choices they make in coming years.
• The "Modernizing" IT category corresponds to Jeffrey Garten's list of the "Big Ten" emerging economies, with the addition of Russia.  Garten's "big ten" are:
  • China
  • India
  • Indonesia
  • South Korea
  • Turkey
  • South Africa
  • Poland
  • Mexico
  • Argentina
  • Brazil.

  What's most immediately noticeable about this group is that you're talking about the bulk of the world's population, not to mention several that recently experienced serious economic tumult (or at least serious buffeting) in the Global Financial Crisis of 1997-98.  With this group, you're also talking about countries that have adopted IT in a huge way only in the past decade or so, so Y2K has some potential here to trigger a bit of a technology backlash if its overall impact is bad enough.

• The "Pre-Modern" IT category bundles up the Rest of the World (ROW).  Here we're talking about countries with low IT penetration rates.

Slide 11 above displays the conventional wisdom that we consistently bumped into when we began our research back in the summer of 1998.  In short, the broad assumption implicit in most writings about Y2K's potential impact was that there was a direct relationship between a country's development level and its potential vulnerability on Y2K-induced network instability.  Following this rule, an ultra-modern IT country like the U.S. was the most vulnerable, while Pre-Moderns like a Haiti or Somalia were least vulnerable.  On the face of it, this made perfect sense, because you can't be harmed by breakdowns in what you don't have--or so it seemed.  This thinking likewise tracked with much military strategizing regarding Information Warfare, which also posited that the more IT-intensive your society was, the more vulnerable it was to Information Warfare.
 
 

What a difference a year makes!  Or so it seems if you buy into the Gartner Group's estimates of likely Y2K network failure rates by country (see Slide 12 above).  Now everyone knows that the Gartner Group's data is heavily based on the self reporting of the countries in question (or the private firms within those countries), so taking this very rough estimate with a grain of salt, you're nonetheless faced with a stunning reversal of fortune that's apparently occurred solely on the basis of the remediation efforts each country has or has not pursued over the last year.  In short, from the perspective of failure rates, the U.S. goes from most vulnerable to least vulnerable, along with a host of like-minded states (e.g., Canada, United Kingdom, Australia).  On the other end of the spectrum, the countries looking at the highest failure rates are the modernizing countries, such as China and Russia, and the IT Pre-Moderns, such as a Vietnam and Zimbabwe.
 
 

While failure rates (the percentage of system failures) are expected to be much higher in the pre-modern and modernizing countries than they are in the U.S. or OEDC nations, failure rates do not, by themselves, describe the whole picture.  As noted earlier, IT is far more integrated into the economies and infrastructure of modern countries than those of emerging and modernizing nations.  Consequentially, 25 percent system failure in the U.S. is likely to be much more significant than a 90 percent failure in a small pre-modern nation.  In the most primitive of these, even 100 percent system failure is likely to be below the event horizon; while even 10 percent system failure in a modern IT-intensive economy could result in significant economic upheaval.  As suggested in Slide 13, when all the factors—remediation effort, dependency on IT, network maturity, distribution and redundancy of the architecture—are integrated, the nations that seem to have the most to fear from Y2K would seem to be those in the process of modernizing.  In general these tend to be increasingly dependent on IT, but have not been able to spend much money on remediation and have not developed the highly distributed and redundant networks of the U.S. and other modern nations.

So really, in the short span of about 12 months, the conventional wisdom on which countries are most vulnerable to Y2K has been dramatically reversed.  Like the original conventional wisdom before 1999, this one also makes eminent sense when you think about it: rich countries with a lot more to lose and a lot more disposable income to throw at the problem have succeeded most in remediating the Y2K threat into something more manageable.  Meanwhile, countries new to the IT scene, whose awareness of Y2K lagged significantly behind that of more advanced IT countries, tend to possess less resources to throw at the problem.  Moreover, they tend to pirate software more and, as such, pay less attention to system administration concerns such as Y2K or viruses such as CIH.  In that sense, the destructive path of CIH, the so-called Chernobyl virus, may well prove to be reasonably predictive of Y2K's ultimate impact--namely, more serious in Asia, Latin America, and the Middle East than in Europe or North America.
 

Matching Country Groups With Y2K Onset Models

So, to the extent that we're willing to go out on a limb regarding which country groups are likely to experience which Y2K onset model, our best guess would be as portrayed in Slide 14 below.
 
 

By arraying the countries across our X-Y axis, we're not so much predicting how we think Y2K will unfold for each and every country belonging to each grouping as suggesting that if any one of the onset models is going to be strongly associated with a particular development or IT-intensiveness level, they are likely to correspond as follows:

• We see the U.S., along with very similarly structured near Ultra-Modern states such  as Canada, Australia, and UK, probably experiencing the Tornados onset model, meaning that Y2K comes in bits and pieces and the countries are essentially robust.  Gartner predicts several other advanced European states, along with Israel, will fall into this category.  Correspondingly, this country group would likely experience the outcome scenario described as Run of the "Mille."
• To the extent that many important Modern states, such as France, Germany, Italy and Japan, have not progressed nearly as much as they might have in the time allotted, we expect that this country grouping may experience something closer to the Hurricanes onset model.  In short, we see the damage stemming from Y2K failures to be more significant than it might have needed to be because those countries enter into the situation more vulnerable than they realize. Correspondingly, this country group would likely experience the outcome scenario described as Houston, We Have a Problem.
• If the Ice Storm model actually occurs, we believe it's most likely to happen to a Modernizing country, such as a Russia, China, India, Poland, or Turkey.  Here, Y2K may hit with far more force both because remediation has been weak and because these countries' systems are--in general--more vulnerable to disruptions.  Correspondingly, this country group would therefore be more likely to experience the outcome scenario described as Y2 KO!.
• It is in the IT Pre-Modern category that we expect to witness the Flood onset model, or the slow build-up of progressive failures.  While these countries' systems in general tend to be more robust in the sense that they are more used to "doing without" or "working around problems," it may well be the slow but steady deluge of many small failures that causes Y2K to seem like a widespread and sustained event that drags out over several months.  Good candidates for Flood status would therefore be less developed states in Latin America, Africa, the Middle East, and South and Southeast Asia.  Correspondingly, this country group would likely experience the outcome scenario described as Humans 1, Computers 0.