"Monoculture Considered Harmful
First Monday

Monoculture Considered Harmful by John S. Quarterman

Monoculture cotton crops and the economy they supported proved susceptible to a small insect in the early twentieth century. There may be parallels in the Internet in the early twenty-first century.


The Boll Weevil
Monoculture Communications Providers
History and Observation
All the World, All the Time
Most Effects
Looking Ahead
Ecosystem and Market
Show Me
Health and Prosperity
Fame and Fortune



The Boll Weevil

In the early part of the twentieth century, a quarter-inch, reddish-brown insect, Anthonomous grandis grandis, [1] spread across most of the southern United States. Its path included:

  • Texas, 1890–1904;
  • Oklahoma, 1904;
  • Alabama, 1915; and,
  • Georgia, 1920.

The boll weevil eats the cotton plant. It wiped out much of the cotton crop in the affected states. Severe economic consequences ensued, because cotton was the main crop, and only one strain of cotton was grown. Farmers faced bankruptcy. Towns failed.


Adult boll weevil
Figure 1: Adult boll weevil.
Note: Image used with permission, Clemson University Department of Entomology, Cooperative Extension Service.


Monoculture cotton was bad anyway, because cotton exhausted the soil. Together with bad weather, it resulted in the Dust Bowl. Depression.

Farmers had grown particular strains of cotton, such as long-staple cotton or sea island cotton, because even though it was hard to pick, it had a dependable market. Sea island cotton is supposed to be the best for making clothes; in Ian Fleming’s novels James Bond would only wear shirts made from Sea Island cotton. Bankers liked it because they had predictable farmers (usually going deeper into debt to the bank). It was a predictable economy. It was also an economy with problems. Let’s not go into sharecropping or pellagra here; suffice it to say there were human costs associated with the cotton economy. Because it was predictable, people put up with it [2].

But it was a fragile economy, because it was based on one crop. Any disease or parasite or predator or weather anomaly that could destroy that crop could destroy the economy it supported. It was going to happen sooner or later. As it happened, the damage was done by the boll weevil.

The boll weevil led to diversification: corn, soybeans, peanuts. A stronger national and world economy eventually resulted. Farmers, vendors, and bankers were wealthier.



Monoculture Communications Providers

On September 11, which worked better, the telephone system or the Internet? There was one main telephone company in New York City. Many people had great difficulty getting telephone calls through for many hours. There were multiple Internet providers. Many people exchanged electronic mail and instant messages with no difficulties. There were problems with some specific Web servers, and there were some routing problems that persisted in some cases for about 90 minutes, but the Internet in general survived quite well.

A single provider Internet

  • wouldn’t have survived September 11;
  • wouldn’t have carried full spectrum news about it; and,
  • wouldn’t have had access to primary sources such as the full text of Tony Blair’s speech to his party (almost all traditional news sources carried truncated versions).

The city of New York even requested Metronet to bring back their Richochet wireless Internet service for use by disaster recovery personnel. The telephone system was not dependable in this emergency. The Internet was.



History and Observation

Texas wasn’t the first place the boll weevil took its toll. In 1848 it had “practically caused cotton cultivation to be abandoned in the state of Coahulia [sic], Mexico” [3]. Apparently nobody remembered that when the bug reached Texas.

And when the weevil crossed the Red River into Indian Territory in 1904, the Oklahoma Experiment Station was not prepared to deal with it. No preparations were made. Data about the weevil had to be obtained by Oklahoma after the fact from U.S. Bureau of Entomology stations in Victoria and College Station, Texas.


Bollworm attacking boll
Figure 2: Bollworm attacking boll.
Note: Image used with permission, Clemson University Department of Entomology, Cooperative Extension Service.


There was apparently no organization assigned to look at current and potential effects of the weevil on cotton crops throughout the United States. If there had been, much economic damage could have been averted.

When our friends in Washington asked us to tell them what we could do about Internet events related to September 11, we found disturbances in various parts of the world, such as Italy, Germany, and Budapest. Timing indicated that they were related. Only extensive examinations by personnel looking at old and new data could produce any conclusions.

When bombs fell in Afghanistan, we needed to look at effects there. We were collecting data from a few destinations in that region, and we needed to add more. Packet loss peaks at 9 AM local time in Uzbekistan: what does this mean? We can try to work that out from contemporary comparisons with links and related regions. For that we need ready access to path information from regular traceroutes.

Also relevant is more extensive historical data on Uzbekistan. If those peaks have been going on for a year, they’re not remarkable. If they suddenly started, they are.

Our Matrix Insight Reporter service provides rapid response readiness, to address Internet performance events wherever they occur. That requires constant path and point observations of the whole world all the time so that we have baselines for comparison.



All the World, All the Time

When we detect a performance event, it’s useful to fire off traceroutes. But how do you know if those traceroutes indicate a change in routing unless you have regular previous traceroutes to compare to? We need automated comparisons and visible displays of them.

As part of our Matrix Insight Surveyor service, we recommend to our customers which ISPs they should use. To do that, we take into account many Internet performance parameters, both local to the customer and across the Internet to their customers.


ISPs by month
Figure 3: ISPs by month.
Note: Larger version of figure available here.


And we also take into account financial stability. We can do that because we look at many ISPs all the time.

In addition, capacity planning has aspects of contingency planning. Which combination of ISPs will produce best redundancy for our customer in the face of physical, financial, routing, or other failure? We collect and collate such information, so it is available for analysis and interpretation.

Similarly, we look at past performance for evidence of performance instability.

And for evidence of ISPs being more related than it might appear from their service offerings. Many ISPs buy transport from other ISPs; we should be able to tell that from similar performance of those ISPs.

These and other factors permit us to tell our customer which ISPs to select for best survivability.

It’s only the first step in Surveyor to look at current paths from a customer to their customers. We use comparison data about nodes on all those paths. And those nodes can be anywhere in the world, such as on the island of Mauritius, off Madagascar in the Indian Ocean. High net worth individuals can live where they please. That means we need to look at the entire Internet all the time: balanced worldwide coverage.

If you want to know if the boll weevil is likely to attack your cotton, you’d best be looking at past, current, and potential paths of the weevil worldwide. If we want to be able to recommend better crop rotation, we'd better have the past history, current alternatives, and stability factors lined up as evidence to convince our customers.



Most Effects

Which recent events had the most effect on the Internet?

  • Code Red worm (IIS);
  • September 11 attacks;
  • Nimda worm (Outlook); and,
  • September 26 cable cuts.

Few people know which of those events had the most effect on the Internet, nor on what part of the Internet; which applications, providers, cities, links. We do.

Few people know that almost all of the recent attacks on the Internet have been attacks on monoculture applications, namely Outlook and IIS, both provided by a single vendor. We can make that visible. Such visibility promotes diversity. Diversity will enhance corporate, national, and world security. Diversity should also be valuable to other providers.


World destination map
Figure 4: World destination map.
Note: Larger version of figure available here.


If people don't know that growing only one kind of cotton is bad for their economic health, they won’t grow anything else.

And cotton wasn’t the only monoculture crop that led to economic disaster. In the late nineteenth century an aphid-like insect, Phylloxera vastatrix, destroyed many of the world’s vineyards, so that that most French wines today are actually grown from American grapevine stock. The company that knows where to get disease-resistant vine stock will be in demand.



Looking Ahead

About five years ago I predicted in various talks that within about five years companies with deep pockets would own the Internet. Specifically I predicted telephone companies would own most of the big ISPs. That has happened. UUNET was bought by Worldcom and has now been assimilated. PSINet, the other of the two first commercial ISPs and the last independent Tier 1 ISP, is bankrupt.

At the time I forgot about energy companies, even though they had been early adopters, and for that matter I’d done work for them. They, too, have deep pockets. Enron became a player. Of course, I also didn’t predict that Enron would self-destruct. No prediction is perfect.

Nonetheless, we need to look ahead. We can look at potential protocols, such as MPLS, IPv6, streaming video, etc. We need to do that, but we also need to avoid getting buried in minutae. Being tied to specific protocols such as SNMP can radically limit the scope of the viewing instrument. That would be a platform-centric approach. A more network-centric approach can work better: Look at the big picture, and zoom in on customer-relevant details.

We can also look at deeper philosophical questions.

  • Still end-to-end?
  • Still decentralized?
  • Still no settlements among big ISPs?
  • All VPN?
  • Cryptography used, required, or prohibited?
  • W3 consortium and Web patents?
  • Government control?

Given the state of the financial market, and especially the failures of so many ISPs and Internet content providers, everything about the Internet is now open to question. Given overbuilding of fiber optics, reallocation of spectrum, and legal manipulations of media companies, everything about communications in general is also open to question.

We can and should examine all these and other issues. Many companies do. Someone needs to look at the big picture.

Some issues can’t be avoided, however. A centralized Internet would be a monoculture Internet, with little room for diversity, not to mention for survivability. Networks are political.



Ecosystem and Market

The Internet is an ecosystem. It is composed of many interacting parts, ISPs, datacenters, enterprises, end-users, each of them drawing sustenance from the others and from raw materials. Each of them need to make informed decisions. This is an ecology. And this ecology whose life forms are corporations and people is also a market.

A market needs price differentiation. Perhaps via some form of service ratings?



Show Me

An ecology evolves by natural selection. Internet evolution has been hampered by difficulty of selection.

Many people find it difficult to understand how an ISP between their company and their customers can be of interest to their company, since neither they nor their customers are directly connected to it or paying it. Talking about it seldom convinces anyone. Drawing it often does. Animating such drawings to indicate the problem and how it would improve if the company changed ISPs can be very convincing. ISPs and bandwidth traders and their customers need Internet performance made visible, so they can differentiate their services. That way buyers can choose and suppliers can compete.


World destination map
Figure 5: World destination map.
Note: Larger version of figure available here.


Visibility enables differentiation, and thus selection. Visibility can enable evolution of the Internet ecology. Visibility can make this market.



Health and Prosperity

An Internet with visible performance will be an Internet in which providers continue to leapfrog one another. It will be an Internet in which buyers can select suppliers rationally. It will be an Internet of stronger providers. It should also be an Internet of continuing diversity. Diversity that will manifest itself as robustness in the face of disasters and attacks.


Customer path performance map
Figure 6: Customer path performance map.
Note: Larger version of figure available here.


That way lies a healthy and prosperous Internet, a healthy and prosperous country, and a healthy and prosperous world.



Fame and Fortune

A monoculture Internet would be like the old pre-1995 telephone system, in which there were only a few suppliers (AT&T and GTE) and performance was not an issue.

The boll weevil never went away, you know. It is merely kept at bay through crop diversity, more resistant cotton varieties, different planting methods, pesticides, and constant monitoring. A visible and rapidly evolving Internet will be an Internet in which continually more sophisticated performance analysis is needed.

That’s a market in which fortunes can be made. End of article


About the author

John S. Quarterman wrote The Matrix: Computer networks and conferencing systems worldwide (Bedford, Mass.: Digital Press, 1990.), a comprehensive book on the history, technology, and people of computer networks worldwide, as well as six other books. He is a founder and Chief Technology Officer of Matrix NetSystems, which is the most experienced Internet analyst, established as Matrix Information and Directory Services (MIDS) in 1990, and since 1998 an Internet startup with Internet performance products. Quarterman published the first maps of the whole Internet; conducted the first Internet Demographic Survey; and started the first continuing series of performance data about the entire Internet in 1993, on the Web since 1995 in the Internet Weather Report, plus comparisons of ISPs.



1. “Cotton Insects,” Entomology at Clemson University, Cooperative Extension Service, Entomology Department of Clemson University, at http://entweb.clemson.edu/cuentres/cesheets/cotton/, accessed 1 February 2002.

2. Personal recollections, David S. Quarterman.

3. G. A. Bieberdorf, 1926. “History of the Distribution of the Mexican Cotton Boll Weevil in Oklahoma,” at http://www.obweo.org/histmexbw.html, accessed 1 February 2002.


Editorial history

Received 21 December 2001; accepted 25 January 2002.

Editorial history

Paper received 21 December 2001; accepted 25 January 2002.

Monoculture Considered Harmful
by John S. Quarterman.
First Monday, Volume 7 Number 2 - 4 February 2002
doi: https://doi.org/10.5210/fm.v7i2.932

A Great Cities Initiative of the University of Illinois at Chicago University Library.

© First Monday, 1995-2019. ISSN 1396-0466.