Home / Blogs

You Paid to Join; You Can Leave Anytime

Once upon a time, around 1998-1999, three of us were hired by APEC-Tel to study “International Charging Arrangements for Internet Services”. APEC-Tel is a regular meeting of Pacific-nation telecommunications ministers. The impetus of the study was their consternation that connection to the Internet was being charged (paid for) in an entirely new way. The template of the old telephone settlement scheme had been overthrown. Those wishing to connect to the Internet, which was centred in the United States, were being forced to lay lines across the Pacific, pay landing rights in the United States or Canada, and pay further to connect to the Internet at the nearest negotiated peering or transit point.

Gone were the old arrangements of telephony, where state-owned institutions split the costs of interconnection, and where traffic settlements were so arranged that North America subsidized telephone usage around the world to a considerable extent by the differential charged between costs and price, on the one hand, and the share of revenue accruing to the trans-oceanic PTT, on the other.

In retrospect, it was astonishing how quickly the peering and transit system was put in place. You “peer” with those who carry equivalent amounts of traffic, and do not charge for the exchange traffic. You pay upstream providers if your traffic volumes are not large enough to qualify to join the club. The system was put in place nearly spontaneously, so far as we could figure out. It made access to the Internet a matter of paying, and the United States Internet carriers made it clear that they had no interest in paying for access to Australia, and that Australia (or China) had better pay for access to the Internet sites of the USA. This was the source of the grievance that gave rise to the study.

The Internet grew at the physical plane by private negotiations among carriers, probably carried out with imperfect information on traffic volumes, by carriers seeking access to North America and its trove of websites. The situation may have changed today, insofar as traffic volumes shift frequently, so that payment patterns may vary. At the time we could obtain no real information about these traffic flows from carriers. Yet the principle was and remains the same, that the world was connected to the Internet by private financial arrangements among large carriers, and that the United States held the superior negotiating position, because the users in China, Japan or Australia wanted access to the nodes in the United States, and the users in the United States were not as interested in reaching those countries. The negotiating advantage was held by US carriers.

You may be wondering about the relation of the payments system to Internet governance. It has to do with bargaining position. Who wishes to connect to whom?

Much of the usefulness of the DNS root lies in the convenience of having a single authoritative look-up system. It is easily conceivable that the root could fracture, and that look-ups would have to go to several different roots, rather like searches through parallel universes. One could easily conceive of new software that would be installed in every computer to systematically search root zones for the actually wanted website, and special recognition software that would tell you whether you have reached the trusted website you were looking for. One could easily imagine states that would have an interest in breaking the root for the purposes of better political control at home. All this is merely to say that the unity of the root is a convention, and that states may overturn this convention.

Accordingly, it is conceivable we will end up with a set of websites unreachable by one look-up mechanism, but available to another. What precisely would have been accomplished?

So far as I can tell, the original impetus would persist to connect to the United States, and the universe of websites reachable by top level domains. The separation of the Internet into various islands of separate address spaces might serve the interests of some states, yes, but the re-unification of those Internet islands by more elaborate look-up systems would probably follow swiftly. On this matter software experts will have more to say than I.

Thus we arrive back at the Internet governance debate. The contention of the opponents of ICANN must ultimately rest on the bargaining power of non-US governments to force a separation of the addressing space (breaking the root), if the United States does not comply with their wishes for a more compliant (and politicized) ICANN. For several reasons I think this bargaining power is illusory.

1. Countries paid to join the Internet in the first place, based on the relative strength of the attraction of US websites and resources compared to their own.

2. The relative attraction of sites reachable through address spaces under ICANN supervision has not fundamentally altered.

3. Governments can cut off the access of their citizens to addresses not approved of by those governments, without any institutional change. As China has demonstrated, censorship can be made to work.

4. Software work-arounds will probably evolve to unite any separate islands of address-spaces.

There was an old joke about England in its prime. The headline of the Times read “Fog in the Channel. Continent isolated”. In the case of Internet governance, separation from the root isolates the separatists, not the United States and the rest of the world using domains which remain under the purview of ICANN.

The United States is perfectly aware of this and will bargain, or not, in the light of these realities. Or as Clint Eastwood once said in the Dirty Harry film: “Go ahead, punk, make my day”.

By Timothy Denton, lawyer, former national regulator (CRTC)

Filed Under


wayne  –  Nov 2, 2005 7:43 PM


This article seems to completely confuse the differences between inter-ISP routing policies with DNS root policies.  I went off to Mr Denton’s website in hopes of finding a clearer explanation, but instead found a bunch of links to “file://C:/Documents…”.  This seems to completely confuse the differences between Mr Denton’s home computer with the WWW.

Pardon me for ignoring any thing Mr Denton says until such time as someone can show me that this isn’t just a bunch of confusion on his part.

Howard Li  –  Nov 3, 2005 2:05 AM

I guess I have to agree with Wayne. This article really got me sidetracked from many different things. Though I tend to agree with the last part that the US awares its premier position and will try to retent the control of the root, I doubt it is due to the interconnection.  And I guess Mr. Denton is ignoring the fast growing in all the ccTLDs, especially the countries in Europe. Anyway, Mr. Denton has mixed up the interconnection infrastructure and the name routing mechanism. And

The Famous Brett Watson  –  Nov 3, 2005 3:34 AM

Although the technical theme of the article is a bit of a blend (tier 1 peering + DNS), the social theme is quite unified, and can be summarised thus.

We are the USA. We hold all the cards and call all the shots. Those who aren’t happy about it can suck it up, because they’re not realistically in a position to force the issue. When the rest of the world comes whining with their “governance” issue, the solution is simple: flip ‘em the bird, and say, “go ahead, punk: make my day.”

Only when viewed as a social comment of that tone does the title of the article make any sense. Don’t be confused by the references to technology: this isn’t a technical article. It is, rather, a quintessential example of modern American arrogance. Re-read it with that perspective, and be educated.

Suresh Ramasubramanian  –  Nov 4, 2005 5:07 AM

The article is staggeringly naive, but does highlight one of the main grouses foreign ISPs have - almost all their traffic (even for local websites) is outgoing to the USA, for various reasons

* Lack of local peering and/or incompetent local network design (aka “static routes upstream”)

* Lots of local websites hosting themselves stateside (besides the usual hotmail yahoo google etc etc)

ps to Timothy - please let me know what you think about anycasted root servers (and large websites that enable akamai to get an effectively local copy of the website anyplace there’s an akamai cluster) and their impact on what you are talking about.

The Famous Brett Watson  –  Nov 4, 2005 2:50 PM

Suresh, you may note that there’s a positive feedback situation in effect here.

I’m in Australia, and our Internet access prices tend to be on the high side thanks in part to the one-way transfer of money situation described in the article. This is a cost borne by the Australian end of the link regardless of the underlying access pattern: client in Aus, server in USA == Aus pays; client in USA, server in Aus == Aus pays.

This means that not only are our Internet access rates a bit on the high side (thanks to us subsidising the USA), but our hosting prices are also on the high side (thanks to us subsidising the USA on that front as well). Consequently, my own personal virtual servers are hosted somewhere in the USA, because that’s the cheapest place for me to rent a server—partly thanks to all the international subsidy they receive.

And so the hegemony reinforces itself.

wayne  –  Nov 4, 2005 3:16 PM


What you describe really has little to do with AU vs US, but more to do with being a Tier-1 ISP vs not being a Tier-1 ISP.  The exact same arguments about subsidizing others can be made by anyone in the USA that isn’t a Tier-1 (although the cost shifting isn’t always as high).  Witness the recent SBC CEO rant about how Microsoft and Google should be paying SBC in order to get to SBC’s customers.  Those customers, of course, have to also pay SBC to get to Microsoft and Google.

If people in AU want to stop paying, what they need to do is create enough content and eyeballs, buy enough cross ocean links and force your way into the Tier-1 club.  I suspect that AU doesn’t have a large enough population to do that, and needing cross-ocean links to get to the rest of the world, your costs will always be higher.  The EU and CN will likely be able to break even with the US.

Anyway, your summary of the anti-US rant is much more honest and technically accurate than the original article.  It is also *MUCH* shorter.  I doubt it would have been accepted as an article here on circleid, so I have to wonder why this article was accepted.

Suresh Ramasubramanian  –  Nov 4, 2005 4:20 PM

* circleid is not moderated - it is open posting, with the circleid admins apparently maintaining a loose and reasonably hands off policy except in rare cases

* circleid is not slashdot and nor is it j.random left/right wing blog for long and pointless rants - i keep reading circleid because the average content on circleid is far better than what I see elsewhere.

The Famous Brett Watson  –  Nov 4, 2005 6:22 PM

Wayne: the main thrust of my second comment wasn’t “US vs AU”, but that the situation has a tendency to reinforce itself. Also, it’s not really about “content” and “eyeballs”—at least not from the perspective of the tier 1 players. It’s more a question of “who hurts worse if our networks don’t communicate?” The party that hurts worse pays. Not exactly noble, but fairly simple.

Ram Mohan  –  Nov 6, 2005 6:48 AM

The same imbalance in access pricing is generally true for India also.

The Indian end of the link generally pays more as long as either the client or the server are in the US.

Hosting prices in India used to be almost 5 times that of the US.  It’s now _only_ 2 to 3 times compared to the US.

Suresh Ramasubramanian  –  Nov 6, 2005 7:37 AM

Well Ram, you probably know this already, but for the sake of completeness, I’d attribute the reason for what you noticed to

1. Increased competition

2. The earlier complete lack of peering + static routes to upstream type of rudimentary routing common at most ISPs (partly because of vsnl’s refusal to allow peering and partly because of lack of adequate bgp knowledge).

Even the limited effect nixi is having (most ISPs still haven’t learned how to use it optimally, and nor do they - or at least their management - seem to trust each other enough to peer fully) seems to have helped a bit - but the driver is increased competition especially in international links, with many many more pipes coming into india now

Not at all like the old days where vsnl, then government owned and the incumbent ILEC, now privatized, had a monopoly on the FLAG cable that was india’s only international connectivity, and kept 90% of it dark while reselling the rest at grossly inflated rates while, as I said, refusing to allow downstreams (most if not all ISPs in india, all competitors to vsnl’s consumer access / hosting etc businesses) to peer at vsnl.


Timothy Denton  –  Nov 7, 2005 6:20 PM

In response to Wayne’s comment, that I have confused intercarrier routing policies, and DNS root policies, the argument I present is an analogy. An analogy is defined as a process of arguing from similarity in known respects to similarity in other respects. My argument about peering versus intercarrier settlements of the telephone world analogizes from the peering situation in the mid-1990s to the ICANN situation now. How much bargaining power do the opponents of ICANN have? I suggest rather little.

As for the rest of Wayne’s comments, they reveal bad manners, the condescension of the ignorant towards arguments beyond their educational breadth, and a generally foul disposition.

Matthew Elvey  –  Nov 9, 2005 12:44 AM

As long as we’re discussing the USA being able to call all the shots:

1)To what extent has this improved today?  I understand that there are lots of peering points around the word, but I’d say at least half of all intercontinental traffic where the USA is not an endpoint still goes though the USA.

2)Does BGP—and do typical peering agreements—allow, for example, a colo facility in Hyderabad with its own connection to a USA peering point (and to local peering points) to say we will only accept traffic over this peering point that terminates in the USA?  Can it at least say it will not accept traffic with both endpoints in India?

I can suck down a steady 3Mbps of traffic all month, for $22+fees/month.  I’ve done so.  I’m sure there’s no way I could do that outside the USA.

Comment Title:

  Notify me of follow-up comments

We encourage you to post comments and engage in discussions that advance this post through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can report it using the link at the end of each comment. Views expressed in the comments do not represent those of CircleID. For more information on our comment policy, see Codes of Conduct.

CircleID Newsletter The Weekly Wrap

More and more professionals are choosing to publish critical posts on CircleID from all corners of the Internet industry. If you find it hard to keep up daily, consider subscribing to our weekly digest. We will provide you a convenient summary report once a week sent directly to your inbox. It's a quick and easy read.

I make a point of reading CircleID. There is no getting around the utility of knowing what thoughtful people are thinking and saying about our industry.

Co-designer of the TCP/IP Protocols & the Architecture of the Internet




Sponsored byDNIB.com

New TLDs

Sponsored byRadix

Threat Intelligence

Sponsored byWhoisXML API

Brand Protection

Sponsored byCSC

Domain Names

Sponsored byVerisign

IPv4 Markets

Sponsored byIPv4.Global


Sponsored byVerisign