Defining Goals and Objectives for System Development

Clearly expressed, goals and objectives are essential prerequisites for successful change management in complex environments. System design, implementation, and commissioning in the broadest sense are classic examples of this type of change. The only rational way of measuring the success of complex systems is to assess the degree to which goals have been achieved by measuring the attainment of objectives.

A commonly used mnemonic for defining meaningful goals is S.M.A.R.T. Which depending on who you read can mean many things including:

  • Specific Measurable. Acceptable. Realistic. Time Framed
  • Specific. Measurable. Attainable. Realistic, Tangible
  • Specific. Measurable. Action Plan. Realistic, Timetable
  • Specific. Measurable. Action Oriented. Realistic. Time Stamped

Taking a superset of these we get:

  • Specific. Measurable, Attainable, Acceptable, Action Oriented, Realistic, Tangible, Time Stamped. Time Framed.

These are all desirable attributes however I feel the S.M.A.R.T. approach confuses goals and objectives, which are actually different things. Furthermore it does not really provide much guidance on how to ensure these desirable attributes are present in any specific goals. I think a more rigorous approach is needed for system design purposes. I use the following framework when defining goals and objectives.


Goals can often be expressed in just a few words that describe a general aim. Goals are seldom directly measurable. They usually include a desirable quality and, if appropriate, an intended modification to that quality. Examples are: increased customer satisfaction, diversified client base, greater customer knowledge, improved cash flow, reduced hosting costs, improved staff moral, increased systems flexibility, and reduced cost of ownership. Often the goal merely specifies the quality that is to be accentuated, Profitability, Reliability, Maintainability, Security etc.


Objectives are tightly bound to goals. Each goal can have one or more associated objectives. Unlike Goals, Objectives are measureable and provide a direct indication that progress is being made towards the goal. By definition the goal is achieved if all its associated objectives are attained. Sometimes objectives represent steps along the way toward achieving the goal and are often called milestones. People commonly define goals without defining supporting objectives, this makes it difficult, and in many cases impossible, to assess whether the goal has been achieved.

Objectives are comprised of 5 parts – Description, Observable Attributes, Test Procedures, Test Values, and Derivation and Authority.


Every objective requires a succinct description that can be used to refer to the objective.

Observable Attribute(s)

Every objective defines a single observable attribute – a parameter, feature, or other characteristic that can be measured and is believed to be a good indicator of the desirable quality described in the goal. For Example; mean-time-between-failure (MTBF) is a common observable attribute for system reliability, days-sold-outstanding (DSO) is often taken as an indicator of client satisfaction and cash flow, and percentage voluntary turnover can be taken as an indicator of staff moral. There are many more commonly used observable attributes and I will try to ennumerate these in another article.

Test Procedure(s)

Every objective has a defined test that can be performed to quantify the observable attribute. The test procedure usually defines the conditions under which the test must be performed. For example; MTBF could be measured over a 6 month period of normal operation commencing 2 months after any significant change to the system, voluntary turnover could be measured every quarter and would exclude employees who were fired and whose total employment was less than 31 days.

Test Values

Each objective has one target value. This is the actual Objective. In addition several other values may be specified for comparative purposes. These are particularly useful when deciding if the target value is achievable and what the cost of achieving it is likely to be. When goals and objectives begin to change as the initiative or project progresses it is by tracking these factors that the system designer is able to stay in control.

  • Comparative Values
  • Worst Acceptable Value
  • Best Achievable Value
  • Current Value
  • Target Value
  • Planned Value

Derivation and Authority

This defines the origin of the objective and is useful when many changing goals and objectives must be managed.

A note on constraints: Constraints are merely objectives not to exceed a specific target value.

An Extreme Example. The Long Now Foundation‘s 10,000 Year Clock

I like extreme examples because they force us out of the mundane world and make us think about our approach as much as the solution we are designing. And as this article is really about the approach to defining goals and objectives and not a particular solution I think an extreme example is called for. The Long Now Foundation has decided to build a clock of monumental proportions that will operate for 10,000 years. They are entirely serious about this largely symbolic gesture for reasons that I will not go into here.

Below are the principles that Danny Hillis generated to guide the design of the Clock.

  • Longevity With occasional maintenance, the clock should reasonably be expected to display the correct time for the next 10,000 years.
  • Maintainability The clock should be maintainable with bronze-age technology.
  • Transparency It should be possible to determine operational principles of the clock by close inspection.
  • Evolvability It should be possible to improve the clock with time.
  • Scalability It should be possible to build working models of the clock from table-top to monumental size using the same design.

I believe these are not so much principles as condensed goals and objectives. Below is my attempt to restate these principles as goals and objectives using the framework I have outlined above.

Goal: Operational Longevity

Objective: Display the correct time for 10,000 years assuming occasional maintenance.

Observable Attribute: Difference between the time displayed by the clock and the time determined by another method.

Test Procedure: For 10,000 years after the clock begins operation an observer will examine the clock display at regular intervals of between 1 and 10 years. The examination will last for a reasonable period, not to exceed two days, sunrise to sunrise. The observer must not have access to another timepiece and can only use bronze-age technologies to take measurements. The time must be calculated from first principles without reference to pre-calculated tables or data.

Test Values:

Comparative Value 1: John Harrison (1693 – 1776) constructed a state-of-the-art mechanical clock called H4 between (1755-1759). Now kept at Greenwhich, it lost 5.1 seconds after a 43 day sea voyage in 1761/2. Extrapolating to 10 years gives a loss of 7 to 8 minutes.

Target Value: Difference between displayed time of the clock and time measured by another method is within the experimental error.

I can not find a description of how to calculate time without reference to astronomical tables or specific events such as transits of planets. I imagine it is hard to do with any accuracy in just two days. I would guess that +- 2minutes would be about as good as could be achieved. And even then the observer would need some information such as the latitude and longitude of the clock and the day of the year.

Objective: A reasonable expectation of long-term success assuming occasional maintenance.

Observable Attribute: Fitness of the clock design to support operational longevity

Test Procedure: The design of the clock, its intended housing, construction materials, methods, and location will be inspected by a team of 12 people with expertise in areas such as: the geology and meteorology of the clock location, Time measurement techniques using Bronze age technologies, Bronze Age construction techniques, The Archeology and construction of ancient monuments, Properties of the construction materials, Large scale chronometer construction and maintenance, survival and reliability of cultural institutions over long time periods, Cultural Anthropology, History and Large scale engineering. This team will inspect the design of the clock for a reasonable period and will vote on proposition that “with occasional maintenance, the clock should reasonably be expected to display the correct time for the next 10,000 years.”

Test Values:

Comparative Value 1: With No moving parts Stonehenge has survived for about 5000 years although it evolved over its first two thousand years. The current configuration is only 3500 years old and half of the stones are now missing or fallen. The purpose of Stonehenge is still a mystery but it appears to be some form of calendar.

Comparative Value 2: The Lighthouse of Alexandria , the only one of the seven wonders of the ancient world that was also a functional structure, lasted nearly a thousand years but eventually succumbed to earthquake damage.

Comparative Value 3: The Jingu Shinto Shrine at Ise is rebuilt every twenty years. It was first built in 04 C.E. The Long Now Foundation have identified this Shrine as an example of the kind of tight integration between institution and artifact that they are trying to create.

Target Value: Unanimous vote of confidence in the designs ability to survive 10,000 years and still operate accurately.

Objective: Occasional Preventative Maintenance .

Observable Attribute: Ratio of Expected Lifetime to Maintenance Interval (Mean Time Between Preventative Maintenance) = The number of expected maintenance events

Test Procedure: A 100 year rolling Mean Time Between Required Preventative Maintenance (MTBPM) will be calculated. This will not include repair of clock failures

Test Values:

Comparative Value 1: In her book Longitude Dava Sobel reports that the current curators of John Harrison’s clock H4 believe that if it were kept wound it would require maintenance, consisting of complete disassembly and cleaning , once every 3 years and could be expected to last at least 3 or 4 centuries, although after such a period so many parts would have been replaced due to loss or damage that it could not be considered the same clock. This gives a Lifetime to Maintenance Interval ratio of between 100:1 and 133:1

Comparative Value 2: The highest ratio I can think of for any machine is for an AK-47 which, I am reliable informed, could be expected to last 30 years if cleaned once a week. The remarkable thing about the AK-47 is that it rarely needs cleaning, no matter how badly it is treated. This cleaning regimen would give a Lifetime to Maintenance interval ratio of about 1500:1

Target Value: Lifetime : Maintenance Interval of 500:1 This target value implies a maintenance event every 20 years – Once a generation with enough overlap to support apprenticeship and knowledge transfer. Similar to the Jingu Shinto Shrine mentioned above.

Goal: Maintainability

Objective: The overall complexity of the clock’s design must be within the grasp of Bronze Age Engineers

Observable Attribute: Design complexity must not exceed Bronze Age capabilities

Test Procedure: A panel of 12 engineers from various disciplines should agree that the clock design does not exceed the capabilities of bronze age engineers implied by Bronze Age artifacts and where available contemporary written descriptions of Bronze Age engineering accomplishments.

Test Values:

Comparative Value 1: The Antikythera Mechanism. Built in 87 C.E. on Rhodes and lost in a ship wreak in 76 C.E. off the island of Andikythera was discovered in 1901 by divers. It consisted of a wooden box containing a corroded lump of bronze gears.

It is hard to exaggerate the singularity of this device, or its importance in forcing a complete re-evaluation of what had been believed about technology in the ancient world. For this box contained some 32 gears, assembled into a mechanism that accurately reproduced the motion of the sun and the moon against the background of fixed stars, with a differential giving their relative position and hence the phases of the moon. It is enough to know that there is no trace of anything like it until around 1000 A.D., and that when it was first published there were serious suggestions that it had been dropped into the wreck at a much later date or even that it was the work of alien astronauts.

Bill Casselmann

The Mathematician E. Christopher Zeeman claims The Antikythera Mechanism is an accurate model to 1 part in 40,000.

The existence of The Antikythera Mechanism implies that tantalizing ancient reports of more complex devices on Rhodes 2000 years ago may well be true. Although none is known to have survived.

Target Value: Unanimous vote of confidence that the clock’s design does not exceed the capabilities of bronze age engineers.

Objective: All parts must be within Bronze Age manufacturing capabilities.

Observable Attribute: Part dimension tolerances

Test Procedure:

A panel of 12 engineers must agree that the design of the clock does not require parts to be manufactured to tolerances beyond those achievable with Bronze Age technologies.

Test Values:

Comparative Value 1: The Antikythera Mechanism. Graduation marks etched on the a dial of device show an error of 0.25 degrees in a 45 degree line

Comparative Value 2: Charles Babbage’s Difference Engine No 1. In His book The Difference Engine. Charles Babbage and the quest to build the first computer Doron Swade states that measurements of the handmade parts from Difference Engine No1 established that Babbage’s engineer Clement was able to make repeat parts that differed from each other by no more than two thousandths of an inch. Obviously Babbage’s techniques were beyond Bronze Age capabilities however they are about the limit of what can be hand made.

Target Value: A unanimous vote that no parts tolerance is required by the design to exceed two thousandths of an inch.

Goal: Transparency

Objective: During reassembly the relative positioning and exact orientation of each part must be self evident.

Observable Attribute: Correct operation for 30 days (one lunar cycle) after reassembly.

Test Procedure

A mechanically competent person with no knowledge of the clocks design, applying only bronze age techniques, will attempt to assemble it from it’s component parts with no written or spoken instructions.

Test Values:

Comparative Value 1: Doron Swade reports that. Charles Babbage‘s designs for the Difference engine showed all the thousands of parts in their correct position relative to each other but gave no indication as to how they were to be physically oriented during assembly so their motions harmonized with the rest of the mechanism. When the engine was finally constructed it took many attempts to get all the parts in their correct orientation. Most of these attempts ended in a mechanism jam.

Target Value: The clock can be correctly reassembled from its component parts in less than a day.

Objective: The mechanism must support diagnosis of problems

Observable Attribute: Mean time to diagnose a problem

Test Procedure

Any mechanically competent person who is also able to correctly reassemble the clock from its component parts should be able to identify the source of any operational problem by inspection alone. This ability will be tested by introducing various problems into a working model of the clock and measuring the mean time to diagnose the cause.

Test Values:

Comparative Value 1: When the Difference Engine in the Science Museum, London jams it is almost impossible to diagnose where the jam has occurred. “There is no easy way of isolating one section of the engine from another so as to localize the source of the jam” All a hapless engineer can do is to poke around with a screwdriver.” The engines design lacks any way of localizing an error to a small section of the clock and identifying the problem without disassembly.

Target Value: The mean time to diagnose the cause of a problem and shall not exceed 15 minutes from problem identification. (Fixing it may take considerably longer!)

Objective: The procedure for resetting and restarting the clock must be intuitive.

Observable Attribute: Mean time to reset and restart

Test Procedure:

Any mechanically competent person should be capable of resetting and restarting the clock if it is stopped.(How they know what time to set it to is their problem!)

Test Values:

Comparative Value 1:

Target Value: The mean time to reset and restart the clock shall not exceed 30 minutes from problem identification.

Target Value: The mean time to reset and restart the clock shall not exceed 15 minutes.

Goal: Evolvability

Objective: The clock must be modular in design.

Observable Attribute: The mechanism must comprise many loosely coupled modules that can each be replaced with other modules that perform the same function but are of a different design.

Test Procedure: Functionally identical modules of different design will be constructed and switched between different model clocks.

Test Values:

Comparative Value 1: The Salisbury Cathedral Clock was installed in 1386 It has no face and only strikes the hours. It originally used a foliot balance with a verge escapement. At the end of the 17th century, these were replaced by a pendulum and a recoil escapement. Holes on the clock’s frame show that these were later replaced again. In 1884, a new clock was installed and the old one was forgotten. In 1956, the clock was restored to its original condition and made to work again. Its pendulum and recoil escapement were replaced by a new foliot balance and verge escapement.

Target Value: Each module of the clock should have two completely different designs that are interchangeable leading to 2 to the power n different configurations of the clock, where n is the number of modules.

Goal: Scalability

Objective: It must be possible to build a working model of the clock over a large range of sizes.

Observable AttributeSize of different versions of the clock

Comparative Value 1: 1/30th scale Corliss steam engine. Jerry Kieffer, a field representative for a Wisconsin power company managed to construct a working 1/30 scale Corliss steam engine model (1.5 orders of magnitude). This involved constructing miniature .009″ diameter hex bolts and matching nuts for use on the model. Every part of the model was to scale although some of the model parts were not made of the same material as the original engine.

Comparative Value 2: The great clock in the tower of the Place of Westminster, London, commonly known as Big Ben, was the largest clock ever built when it was installed. However size brings it’s own problems namely momentum which increases with the cube of distance. When one part of the mechanism failed in 1976 it took a year to repair the extensive damage.

At 3:45am on the 5th of August 1976 as the clock started to chime, metal fatigue in the shaft connecting the chiming train to its fly fan caused the shaft to break. Without the retarding and braking effect of the fly, the chiming mechanism, propelled by the 1.25 ton weight in the shaft, increased its speed of rotation dramatically. This led to the total destruction of the chiming mechanism, with various components and fragments of others being scattered about the clockroom. Some pieces of machinery were flung at the ceiling with sufficient force to penetrate to the room above. The cast iron frame was fractured and collapsed onto the winding motor below. The flying debris also caused damage to the going and striking trains.

It was necessary for the chiming train to be reconstructed from scratch. The magnitude of this task meant that other options, such as replacement with an electric motor, were considered. The reconstruction took almost one year to complete.

Target Value: The clock design must allow working models that scale over 2 orders of magnitude (1 to 100 meters)


In thinking about the 10,000 year clock I have come to several conclusions;

Bronze Age technology was more advanced than I ever knew. There appears to be one issue that Danny Hillis did not address in his principles. If the clock survives it could well become a victim of it’s own success. John Harrison’s H4 has become an historic object and is now considered too important to be wound. The “evolution” of Salisbury Cathedral’s clock has been reversed by removing the pendulum and recoil escapement, added to improve the clocks time keeping, and replacing it with a foliot balance with a verge escapement, thus “restoring” the clock to its original configuration. It would be a shame if The 10,000 year Clock of the Long Now were to survive only to be stopped to “preserve” it for future generations. Paradoxically it may be worth deliberately designing all the clocks parts to wear out and be replaced before they can become historic artifacts in their own right.

This diagram shows a food web, The nodes are species and the lines show predator-prey relationships between the species. Species at higher trophic levels eat those lower down in the web. This particular food web is for Little Rock Lake in Wisconsin and was produced by Neo D. Martinez of San Francisco State University, Romberg Tiburon Center for Environmental Studies.

Little Rock Lake food web

Enough food webs have now been cataloged and described in detail for them to be studied as a general class of system using the techniques applied to the study of other networks. There are only a few descriptions of these studies outside the scientific literature. The California Academy of Sciences quarterly magazine – California Wild had this article called Untangled Food Webs and the BBC ran this story Life’s not so complicated Web but beyond that there does not appear to be much out there.

Searching the Proceedings of the National Academy of Sciences (PNAS) for papers published since January 2000 with the search string “food web” in the title or abstract yields 104 results (2003-04-28) and after a visual examination of these papers only 5 seem to deal with the study of food webs as a distinct phenomenon.

Below I have attempted to summarize these articles and papers. All the errors and omissions are my doing. Please check the sources I have linked to above before you draw any conclusion based on my summary, I am not an expert.

All food webs seem to share similar properties

Food webs with more than around 12 species appear to have common properties that seem largely independent of the particular ecosystem they describe. These properties allow interesting predictions to be made about food webs and the species they contain. For example; the number of cannibalistic species, the ratio of species with specialized diets to species with broad non-specialized diets, and the maximum length of possible food chains can all be predicted.

Big fish tend to be less common than the little fish they eat

There appears to be a roughly inverse relationship between body mass and abundance for individual species within food webs. In one study body mass and abundance were each found to vary by at least 10 orders of magnitude but biomass abundance (average body mass times numerical abundance) varied by only 5 orders of magnitude. In addition species with small body mass tend to appear low in the food web and are numerically abundant.

Food webs are highly connected and easily disrupted

In one study of 16 food webs with 25-172 nodes from variety of aquatic and terrestrial ecosystems food webs were found to have higher connectance (the fraction of all possible links between nodes that actually exist in the network) and smaller size than other types of networks that were studied. The complex feeding relationships between species typical of high connectance food webs can mean population fluctuations of one species dramatically affect another. There is evidence to suggest this ripple effect rarely travels more than 3 links away from the source. However the high connectance of food webs (on average species in a food web are just two links apart and >95% of all species are within 3 links of each other) means that biodiversity loss and species invasion may have a greater effect on ecosystems than previously thought. The introduction of wild pigs to the California Channel Islands is thought to be an example of this knock on effect. Piglets provided a bountiful food source for golden eagles, that turned to the native island fox as a food source when the piglets were out of season, thus allowing the population of island skunks to soar. An attempt is being made to put the Channel Islands ecosystem back to what it was. A mass cull of pigs likely to take 6 years and a captive breading program for the few remaining foxes is now underway.

Food Webs interact with each other

In a study of contiguous food webs, a stream and the deciduous forest it ran through, the two food webs were shown to exchange insect prey during different seasons. During spring adult insects emerged from the stream and where eaten by birds and during summer insects fell into the river and were eaten by the fish. This exchange represented a reciprocal alternating subsidy.

Speedster from Traveler Guitar

I’ve been through so many airports recently I can’t remember which one this happen at. But I was waiting at the gate when this huge Viking of a guy came and sat next to me with what looked like a rifle case slung over his shoulder. My curiosity was aroused and as I checked out his case he caught me looking at it, so I felt it best to ask him about his Traveler Guitar to show I wasn’t just some loony staring at his baggage. He lit up with enthusiasm and proceeded to pull out the guitar and bolt on the arm support. Then he produced a digital sound processor the size of a cigarette pack, and some headphones. He plugged them all together and handed it to me. Cool! The thing was actually still in tune! He dialed in a nice funky sound and I was off, much to the amusement of the other passengers.

Some people get confused about quality. Quality is fitness for purpose it is not an absolute. This is the finest quality traveling guitar I’ve ever come across. But it’s not a patch on my Fender Stratocaster for sound quality. The point is it doesn’t need to be. The difference is all about fitness for purpose. The traveler guitar weighs in at only 4 lbs, half the weight of my Strat. It has a full scale neck with 22 frets. But the machine heads are buried inside the body so they can’t get damaged or knocked out of tune. As an added benefit the neck is shorter as there is no head stock. The arm rest has two positions one for playing and one conveniently tucked away for traveling. And the whole thing looks great which is essential for an electric guitar! But best of all when packed for traveling it fits in a small bag that can be easily slung over the shoulder, unlike my fender which requires a flight case the size of a small coffin. Necessity is the mother of invention which is why this guitar could only have been invented by a pilot. But the high degree of quality coupled with the simplicity of the design makes this guitar a masterpiece in my book.

In 1999 Ryan McCormack and I wrote a marketing piece on Globalization for Sapient Corporation. Aimed primarily at raising awareness of the issues involved in building global Internet systems it also touched on national market analysis and selection. I was reminded of this diagram showing income and connectivity for every country in the world from that piece while reading various articles on US Foreign Policy recently. These articles included this piece called the Pentagons new Map by Dr Thomas P.M. Barnett a US military Strategist on Globalization and US Foreign Policy.

Basically I think Dr Barnett is on to something when he claims that “disconnectedness defines danger”.

Show me where globalization is thick with network connectivity, financial transactions, liberal media flows, and collective security, and I will show you regions featuring stable governments, rising standards of living, and more deaths by suicide than murder. These parts of the world I call the Functioning Core, or Core. But show me where globalization is thinning or just plain absent, and I will show you regions plagued by politically repressive regimes, widespread poverty and disease, routine mass murder, and “most important” the chronic conflicts that incubate the next generation of global terrorists. These parts of the world I call the Non-Integrating Gap, or Gap.

Having said I think Dr Barnett is on to something I don’t think his conclusions are correct! He makes three mistakes;

  1. He confuses poverty with danger. Deliberate disconnectedness is dangerous for everyone, whereas poverty is only dangerous for the poor. Lumping the poor and the dangerous together is callous.

  2. He attempts to define a contiguous geographic region of the World that he calls the “Non Integrating Gap”.

The Pentagon's new Map

In this regard he makes a fundamental mistake. Conectedness is a network property and networks are fractal not contiguous. There is no contiguous region that is disconnected. Within each disconnected country there are islands of connection and within each connected country there are islands of disconnection. This is true at all levels, continents, nations, regions, cities, and companies, right down to individuals. There are terrorist cells in US cities fighting to disconnect the world and Journalists with satellite cell phones in remotest Afghanistan, Iraq, and Somalia working to connect everything.

  1. A corollary of the contiguous region hypothesis is the idea of seam states that buffer the integrated core from the non integrating gap. As there is no contiguous region there can be no border, or rather the border is infinitely long. Either way the concept of Seam States is meaningless.

Income and Connectivity

I believe a better analysis can be produced by considering relative income and connectedness. I used per capita income as calculated by the World Bank and the number of Internet hosts as counted by the Internet Software Consortium (see assumptions). By plotting these two factors against each other for every national economy a picture emerges. Not surprisingly there is a correlation between these two factors – richer countries tend to have greater connectivity. But countries also tend to cluster by geographic region. The graph below clearly shows clustering of Sub Saharan African, Middle Eastern and North African and European economies. Over time economies tend to move to the top right of the graph – they increase per capita income and Internet penetration. The rate at which they move seems to be determined by their per capita income. Wealthy countries can adopt the Internet more quickly than poor countries. But there does seem to be a limit to the speed with which a country of a given per capita income can adopt the Internet. Most countries seem to move at this limiting speed and so a wave front of countries seems to have developed all rushing as fast as they can afford toward greater connectivity and wealth. One interesting feature of this graph is the countries that are not in their expected place on the wave front. It is these countries that are the subject of rest of this article.

Income and Connectivity - Outlier Economies

The Outlier Economies

The graph above identifies several exceptional economies. These economies do not behave according to type. They are significantly different from their neighbors. They are outliers, islands of relative connectivity in a sea of disconnection like Israel, South Africa, and Kuwait or the reverse such as Burundi, Ethiopia, and Yemen. For example Israel, unlike it’s middle eastern neighbors behaves like a European economy with respect to income and connectivity, South Africa is significantly more connected and wealthier than the rest of Sub Saharan Africa. While Kuwait and Yemen show opposite extremes, one enjoys fabulous wealth while the other suffers crushing poverty. These outlier economies are similar to Dr Barnett’s seam states, however they do not surround anything in fact they are themselves surrounded. As a result of their isolated status they tend to get involved in conflict with their neighbors.

The Reluctantly Connected Economies

There are many economies that are unable or unwilling to achieve levels of connectivity that economies with similar incomes manage to sustain. These economies fall behind the wave front, they are the reluctantly connected. There is a significant risk that these economies will backslide and become disconnected. Preventing this backsliding and ensuring these nations realize the full benefits of connectivity should be a major objective of the connected world.

Income and Connectivity - Reluctantly Economies

The Disconnected Economies

Finally there are economies that could not be plotted on these graphs because the necessary data are unavailable. I believe these fall into three main categories. There are a few dependant territories like the Faroe Islands, Guam, and Greenland that can be discounted because their data are included in other economies. Then there are the economies disconnected by paranoia, fear and hatred and finally the economies disconnected by secrecy. The table below is ordered by population size and shows a fairly clear division. The larger states are the pariah states – dangerous, paranoid and in some cases anarchic states that refuse to publish economic information even if they have the capability to gather the data. The smaller states are the hear-no-evil, see-no-evil, speak-no-evil bankers, protecting their clients and their own economies. These offshore tax havens and centers of secretive banking often serve the pariah states, and their corrupt leaders. Both these types of economies are dangerously disconnected. The world would be a far better place if they became fiscally transparent, connected economies.

Economy Hosts GNI (Atlas Method $) Population (1000s) Status
Congo (Democratic Republic) 113 0 52360 Pariah State
Myanmar 2 0 48315 Pariah State
Sudan 0 330 31687 Pariah State
Afghanistan 2 0 27248 Pariah State
Iraq 0 0 23750 Pariah State
Korea, Dem. Rep.   0 22384 Pariah State
Syrian Arab Republic 0 1000 16593 Pariah State
Cuba 848 0 11222 Pariah State, hold over from the cold war
Senegal   480 9769 No Internet Host count available
Somalia 1 0 9089 Pariah State, Anarchic
Haiti 0 480 8114 Pariah State
Benin 0 360 6437 No Internet Hosts in 2001
Libya 59 0 5410 Pariah State
Nicaragua 1655 0 5202  
Puerto Rico 1667 0 3950 US Dependent Territory
Liberia 0 0 3216 Pariah State
West Bank and Gaza   1350 3091 Dependent Territory, Pariah State, Emerging Nation ?
United Arab Emirates 29029 0 2976 Fiscally Secretive
Oman 646 0 2452 Fiscally Secretive
Qatar 0 0 598 Fiscally Secretive
Equatorial Guinea 0 700 469 Pariah State
Brunei 4398 0 345 Fiscally Secretive
Netherlands Antilles 104 0 217 Dutch Dependent Territory, Fiscally Secretive
St. Lucia   3970 158 Fiscally Secretive
Guam 149 0 157 US Dependent Territory
Channel Islands   0 149 UK Dependent Territory
Mayotte 0 0 145 French Dependent Territory
Virgin Islands (U.S.) 58 0 122 US Dependent Territory
St. Vincent and the Grenadines   2690 116 Fiscally Secretive
Aruba 785 0 104 Dutch Dependent Territory, Fiscally Secretive
Isle of Man 116 0 75 UK Dependent Territory
Northern Mariana Islands 13 0 72 US Dependent Territory
Andorra 876 0 67 Fiscally Secretive
American Samoa 915 0 65 US Dependent Territory
Bermuda 4892 0 63 UK Dependent Territory, Fiscally Secretive
Greenland 2229 0 56 Effectively a Dutch Dependent Territory
Faroe Islands 1588 0 45 Dutch Dependent Territory
St. Kitts and Nevis   6880 41 UK Dependent Territory
Cayman Islands 533 0 35 UK Dependent Territory, Fiscally Secretive
Liechtenstein 762 0 32 Fiscally Secretive
Monaco 434 0 32 Fiscally Secretive
San Marino 673 0 27 Effectively an Italian Dependent Territory
Timor Leste 6 0 0 Emerging Nation, Destitute

Assumptions and Sources

All the data used was for 2001.

I assumed that the number of hosts that use the assigned Top Level Domain (TLD) for a given country is an excellent indicator of the connectedness of that country. I believe this because hosts (computers) themselves are tradable commodities that must be purchased from abroad and that connection to the Internet is a very real sign that someone in a country wants to realize the benefits of connection to a global communication medium. Data for the number of hosts per country came from the Internet Software Consortium July 2001 Survey. Assuming Top TLDs actually map to countries leads to errors. In particular for the US since very few US hosts actually use the US domain. I totaled the following TLDs to get values for the US (net, com, edu, org, mil, us, arpa, gov, unknown). This of course is incorrect but I assumed it would tend to boost the US values higher and reduce the values for all other countries proportional to the number of hosts in the country. Three TLDs were ignored (int, biz, info), they have so few host that would make no difference anyway.

Data from the World Bank was used for population values and per capita income.

I have not produced charts for all regions of the world. Here is the zipped excel spread sheet I used. Feel free to play with these data yourself. I’d be interested in seeing what you come up with.

L. Peter Deutsch first published the “8 Fallacies of Networking” internally while working at Sun Labs in 1991-92. This is a great list of the kind of wishful thinking that clouds so much system design.

Essentially everyone, when they first build a distributed application, makes the following eight assumptions. All prove to be false in the long run and all cause big trouble and painful learning experiences.

  1. The network is reliable
  2. Latency is zero
  3. Bandwidth is infinite
  4. The network is secure
  5. Topology doesn’t change
  6. There is one administrator
  7. Transport cost is zero
  8. The network is homogeneous