The IP Development Network Blog

I was contacted this week by Stefano Quintarelli, President of the Italian Internet Service Providers Association. He was interested in my views on IP as a natural monopoly which is a subject I have often thought about but from a different angle to that proposed by Stefano.

My point is that a change in the mix of the load of backbone/access on IP
networks might well evolve in a monopoly of access. The basic considerations
are:
80% of traffic (or nearly so), is P2P
the real-time requirements of P2P traffic are likely to increase in the future (think P2P real-time re-transmission, a video-skype)
2 users in the same building connecting to the same ISP will route their traffic in the best possible point (as near to CPEs as possible)
2 users in the same building connecting to different ISPs will likely peer in a remote point
the overload in the backbone and worsening of latency and throughput will become more noticeable as we evolve towards synchronous services
perception by the users of different quality of service (in a non-technical sense) will increase leading to concentration on one single ISP
People in different communities might choose different ISPs, but in a situation like Europe, where incumbents own majority of the markets, they would benefit the most.
In conclusion: due to the fact that IP traffic is routed at the nearest point whilst operators peer at distant points, in the medium term, the operator that has the majority of users will provide to users a better experience grabbing even larger market shares.

Previously I had only really thought fibre networks as a natural monopoly, but Stefano's points make clear that IP as a protocol must be considered too. I had not properly considered the impact of peer to peer, which as Stefano stated promises to further concentrate the benefits of scale. "On-net" performance is superior - we know this from the messaging of corporate ISPs for many years now.

I have long held the view that telecoms is a natural monopoly. This starts for me with the economics of laying fibre. One company laying fibre can provide capacity far more cost effectively for an entire population that could 2 or 3 or more, sharing the customers. This is because the major cost is in digging the road and laying the ducts. Once you have done that, you can provide additional capacity and additional customers for a small marginal cost.

Problems come when more than one network is built. In these cases more is spent building the two networks than would be spent building just one with double the total capacity. So, multiple networks are fundamentally uneconomic.

When you have multiple networks you also see over supply and a price war. Once you have built a network and found that you have competitors in the same position, it makes sense to try and recover something, even if it is a small percentage of your total investment, rather than letting your competitor win all the customers with a cheaper price. While a price war is good for consumers in the short term, for investors in particular it is very bad indeed. Just ask the people who paid for the Jazztel, Versatel, Verizon networks for example...

In the long term, consumers also lose because those same investors will shy away from later investments. I think this is where we are now. There is a lot of "once bitten, twice shy" money sitting on the sidelines gravitating towards private equity (to extract value from earlier investments), rather than going into next gen networks, products and systems.

The rumoured private equity buy outs of NTL (and even BT) would be intended to extract the value of past investments by maximising short and medium term cashflows. This would be very bad for consumers who would see prices rise and investment stop. My point is that because of past failures, it is more likely that investors will take this approach than support new wave investments.

IP as a protocol has always encouraged monopolistic approaches, the first example I can identify clearly was that of UUNET in the 1990s. Without regulatory intervention, which happened when parent company WorldCom acquired MCI, UUNET would have denied peering to anyone. Indeed, such was its market strength at the time (60% share) that it could have demanded fees from its Tier 1 "peers" simply because they would not have been able to offer a universal service without interconnection with UUNET. That the company chose the opposite route and opened up to more peers, including some major European carriers, was due to the regulatory pressure.

In other related communications industries we see the same monopolisation - software for example with Microsoft or routers with Cisco. In other communications activities, like language we have the same phenomenon - in business, we communicate in English even. The point is that we need a common platform for communication and interconnection, which yes, leads to a monopoly.

The problem with monopolies is where they are either stagnant - avoiding investment because they don't have to invest - or artificially constrained by regulation. PTTs have been constrained so as to create a market, but what we have found is that now there is a market, it is the companies that were created by the artificial intervention that are most at risk from the price competition and unnecessary infrastructure that regulatory action caused them to build.

So what is the solution...? It strikes me that markets that have been extremely successful have been those that left the network as one entity and created competition in the sales and marketing of products delivered over the common framework. Examples include CPS and IP Stream. These have led to lower prices for consumers while protecting the investment return of those who built the network (in the examples, BT). For sure, CPS companies and ISPs have gone broke, but none have done so leaving the spectacular trail of destruction wrought on investors by the fibre network builds of the 1990s.

The problem then is innovation and whether long term investment is made in the one network. The solution there might be where regulation has a role, mandating year on year investments, technology rollouts or whatever is necessary to take the common platform forwards, rather than competition per-se.

We don't want the situation where the smart money is on milking the assets, but that is where we are heading unless policy can be developed to protect those that invest. If LLU leads to another wave over over supply and price competition (why are ISPs all building enough ports for twice the population of the same city exchanges leaving rural communities with nothing???) Where you have multiple networks, you have oversupply and investors risk losing their shirt, trousers and underpants too.

# posted by Jeremy Penston @ 10/17/2006 04:26:00 PM

Tuesday, 17 October 2006

IP as a Natural Monopoly