Tuesday 25 March 2008

OECD broadband infographic using Google Motion Chart

Over the weekend I've been fiddling with the OECD Broadband stats to get them into a Google Spreadsheet and then be able to publish the results in their moving charts. I've also made a heat chart. It's not really perfect, but it shows the possibilities. One of the conlusions I do draw is that there seems to be a strong correlation between household PC usage and Broadband penetration. Unfortunately there are only limited numbers available for PC usage

I've found that the apps are not consistent in the way the work. A heatmap works with rows, a moving chart with columns. Also I'm missing tracking options to follow one particular dot (you can click them, but they don't leave a trail unfortunately.) You can find the sheet here.

If you really want to know how it is done, than have a look at http://gapminder.org/ or look at the Youtube movie of Prof. Hans Rosling, the inventor, at the end.

Moving Chart


Heat Map



Hans Rosling

Thursday 20 March 2008

Mobile Growth in India is something different

Yesterday I wrote something about how Bharti Airtel sees one network for all as a viable option. I also did some speculation on growth in India, without actually really knowing the numbers. Today Om Malik linked to Unstrung where they show the latest growth figures in India. Prepare to be shocked! (at least I was)

8.44 Million new customers per month! -> Close to 100 million per year?
This is actually down from 8.77 million in January, because one operator is facing serious supply problems for mobile equipment
Bharti Airtel by itself did over 2.25 million new subs for a total of 59 million.
There are currently 246 million mobile phone users. By the end of the year there will be more mobile phone users in India than there are people in the US. In two years more than the entire population of the EU.

However I do wonder how they will get mobile broadband as this would require an even larger investment in networks and fibre than is currently taking place.

China probably does similar numbers...

Wednesday 19 March 2008

Ms. Reding, the industry is ready for structural separation (in India)

Tim Poulus linked to an interesting interview of Telecoms.com with Don Price, CTO of Bharti Airtel on how they have grown from hating outsourcing and one single network for all to loving it. Could you imagine Deutsche Telekom arguing for one network for all? (actually you could, but that would be in the traditional monopolist way, not in the lets all share the same network and compete on access and services)

This kind of heresy can only come from a country where the telecommunications game is played on different scales with different budgets and margins. Rolling out the whole of India is a daunting task, evenmore when taking into account that the 1 billion Indians have much less to spend each month but all want to have a mobile phone. (Yes all of them have a need to communicate, so I actually expect mobile penetration to reach 100% before 2020)

Greatest quote
As a network person I can do very little do drive the top line but I can certainly do a hell of a lot to drive cost out of the middle. So therefore I believe things like passive network sharing, site sharing, co-building of sites and ultimately even the active network sharing is the right path to go down. If you and I are competing in the same market, it doesn't make sense for both of us to do the build out. We have an expense that we can't reduce, you're left with an expense that you can't reduce. You're on the left side of the highway, and I'm on the right side of the highway. What did we do? We crossed the finish line six weeks apart. So as we go forward in the industry, as a network community, let's build one highway, one common infrastructure and let the sales and marketing guys compete on the cars."

Quote of the day

This just typifies blogging (including my own) on so many levels:

Nothing can be so amusingly arrogant as a young man who has just discovered an old idea and thinks it is his own.
- Sidney J. Harris

Monday 17 March 2008

No Broadband means no Mobile Broadband

Brough Turner brings up an interesting point how most backhaul from mobile cell sites are only 1 or 2 E1 or T1's and that just doesn't do the trick for mobile broadband. Nyquist Capital made a quick remark that this would be booming times for mobile backhaul. My opinion is that in many countries where broadband in the rural areas hasn't arrived yet will also not benefit from mobile broadband as it takes broadband to the cell site just to be able to make mobile broadband.

Broadband btw is not defined as 256kbit or 2mbit/s but at least 10mbit to the cell site in order for it to maintain adequate capacity for modern usage. WiMax sounds interesting for backhaul until you realize that the advertised 40mbit is half duplex and only reached within short range of the antenna. So for Wimax or LTE to fly in Rural areas at advertised speeds you need a minimum of VDSL/SDSL to the cell site, but a strong preference for VDSL2 or Docsis 3.0 as a minimum. However VDSL2 requires being closer than a 500 metres from the cell site. At which point you could say... why bother and just dig all the way. Docsis is tricky for different reasons most of them being that Cableco's don't do SLA's as a general rule. And there we have it... mobile broadband over any technology requires a finely distributed fixed network. Gentleman start digging!

Two FTTH projects fail and more trouble in The Netherlands

Two projects supported by Reggefiber have had to face defeat. Both were unable to reach the required sign-up rate for the project to commence.

The first one is a progression of the succesful Lijbrandt Telecom FTTH project in Hillegom. In Hillegom market share has been way over 50%. Trying to build the network in neighbouring Lisse (of Keukenhof-fame) was a failure. The treshhold for building the network was at 40% and wasn't met. This being a non-subsidized, private sector initiative the investors cancelled the project.

The same thing happened in the Dutch town of Arnhem, where XMS wanted to build a fibre network in the neighbourhood De Laar. Here too not enough people signed up for the project to start.

BBNed has troubles with ISP's defecting the FTTH projects where it manages the layer 2 infrastructure. This is both in Amsterdam and Rotterdam. Complaints are that BBNed doesn't fulfil orders in time and that it doesn't fix technical problems fast enough. Whether or not this is true is hard to see from the outside, but it does show that running an open fibre network requires quality on many layers and not just the fibre layer.

This shows the difficulty of entering the market. In order for it to be financially viable a new entrant needs to become an instant incumbent overnight and that is not easy. Everything needs to be perfect from the get go, both the technology and the companies that deliver those services have to deliver perfection. Price is also an issue. In the Netherlands the cable companies have started a fierce competition against any new entrant into the market. UPC for instance dropped the price of its offer in very specific neighbourhoods to counter the introduction of fibre. Just to make matters worse, lawyers are part of the problem. Legal fights over Citynet's fibre in Amsterdam have been well documented.

Thursday 13 March 2008

Free announces stellar numbers, but why is FTTH capex so high?

Iliad, the mother company of the French ISP free.fr posted its financial results this morning. Based on these numbers you can only say: Man, there is a load of money in providing internet access. Free's business model is quite simple. You pay €30 and you get 28mbit/s unlimited access, with free calling to fixed lines in 70 countries, and HDTV and 70 tv-channels, and a personal tv-channel, Digital TV Recorder, remote access to stuff you recorded on your home DVR etc etc.

Ok, here are the numbers.
KPI 2005 2006 2007
ADSL Subscribers 1,595,000 2,278,000 2,904,000
Unbundling Ratio 70.2% 75.9% 81.5%
ARPU (at year end) €32.2 €34.5 €36.3
Churn = 1% / month less than 1%
month
much less than 1%





Column1 2006 2007 difference
Revenues 935.1 1 212,4 +29.7%
EBITDA 328.5 443.6 +35.0%
EBITDA Margin 35.1% 36.6%
EBIT 180.5 213.8 +20.4%
EBIT Margin 19.5% 17.6%
Net Income 120.6 150.2 +24.4%
Proposed Dividend €0.27 €0.31 +14.8%


626,000 ADSL net adds in 2007 and 24% of all net adds in France.
€33.3 M of FTTH CapEx in 2007
€21.4 M of FTTH equipments & NROs under leasing agreements at year end
€1,500 CapEx per existing subscribers confirmed in Paris
2008 / 2009 CapEx between €300M - €400M

This last one surprises me a bit. The Capex per sub in Paris seems to be rather high. I have no clue where that comes from as it seemed that Paris would be cheaper than €500 per sub because of not having to dig.

Sunday 9 March 2008

What Submarine Fibre costs

Pretty soon I will be the Submarine fibre blogger if my posts continue the way they do. However I found a very interesting quote throught the ITU's/Strategic Policy Unit Blog. They link to an article in Business Daily Africa, which shows the costs of running the TEAMs project (wiki entry), to run fibre to the east of Africa. Those joining in the project expect an Internal Rate of Return of 32.71 per cent with a pay back of 2.4 years. Not a bad return for a fibre that's supposed to last 25 years. Current prices of connectivity are $5000 dollar per month (satellite only I believe) and the government hopes to go to $500/month in a couple of years. Though UUNET says this can't be achieved.

Another similar project is Eassy, to which TEAMS is a reaction. The Kenyan government seems to have gotten fed up with Telekom South Africa. According to the news I read Telekom South Africa is not being cooperative as it wants to keep prices high. From what I heard here and there Telekom South Africa is one of the leading reasons why Africa has such expensive bandwidth. It's the major network in the region and it lays down the rules, one of them being that prices need to be high and competition is not allowed.

“We have structured the pricing in such a way that even after factoring in the cost of onward connectivity the highest an operator may charge is $700. Anyone who will be charging consumers more than that in the first year of operation will just be serving the cause of greed,” he said.

Dr Ndemo said the cost of connectivity was expected to continue dropping gradually in the subsequent years as the demand for the bandwidth picks up.

“The reason we are working on these figures is that we do not plan to make money from the pipes but only from the services rendered and by taxing the jobs that will be created,” said Dr Ndemo.

The Kenya government has a 40 per cent stake in the project with Dubai’s Etisalat holding 15 per cent. A 45 per cent stake has been reserved for private telecommunication companies.

Shareholders will have access to large amounts of capacity with a 13 STM1 (Synchronous Transport Module ) — basic rate of transmission of fibre optic network — for a minimum of five per cent equity investment.

Capacity will be allocated at cost with TEAMs investors paying $400,000 per STM1 per year. This translates to $2580.645 per megabit per year up to Fujairah and $ 215.00 megabit per month.

TEAMs shareholders are expected to operate on an Internal Rate of Return of 32.71 per cent with a pay back of 2.4 years.

Friday 7 March 2008

Mein Gott, Europeans may get unlimited phone calls!

The great thing when journalists react to brilliant papers is that they generally summarize the myriad of pages and the fruits of years of work back to one sentence: the headline. This week, this piece of brilliance on the basis of termination fees in the context of European regulation and the development towards IP-networks was headlined as: "Should Europeans pay to receive phone calls?" and "Mein Gott!" My headline for this piece could be: "Mein Gott, Europeans may get unlimited phone calls!" or "Telecoms and interconnections for dummies!"

Both articles mentioned show a complete lack of clue on how the telecommunications industry works. Raving on about how we will pay for incoming calls and don't forget the poor children in Africa. They also show us why it will be hard to change the industry, since the misconceptions are so entrenched ever 'hack' will instantly write them down and not know the depths of their incompetence. Abolishing terminating fees is good from a public welfare and market economics point of view. Abolishing terminating fees is bad from the private perspective of a telecoms operator. If I was a regulator, I would promote it. If I owned a telecoms company I would oppose it. I am a consultant, an arms merchant, I therefore supply both sides. (Typical Dutch thing, we supplied the Spanish their guns in our war of independence 1568-1648)

The problems starts at the root: Knowing what a telecommunications network is. Economics/financials quickly make people scream Adding interconnection to the equation complicates stuff way beyond the knowledge of most people in the industry. The business models of interconnection complicate stuff even more Looking at it from the point of a regulator/market economist adds another layer of problems. So here in my arrogance I write down my For Dummies Guide.

Misconceptions about the network
The biggest misconception is that telecommunication is about communication. It's not. It's about enabling communication across distances using some form of technology. Communication is between end-points (people/machines) Telecoms is just about carrying the data from a to b. (Correlation, switches just connect wires, they don't connect and change data. Only end-points do. Therefore there is no intelligent network, knowing how to handle data is limited to putting it onto the right wire. That's fast, simple and efficient.)

Data is carried by modulating a signal onto a wavelength. In the case of digital communications this is represented by 0's and 1's, the strings of 0's and 1's together form the data.

Carrying data from A to B requires a network. For real-time communication a wired or wireless network is best. Fibre is used in the core, copper is still used to get close to the end-user and wireless is used to hop to the end-user and allow the end-user to move around and maintain mobility.

The costs of sending bits over a network is zero. The costs of a network is in building the infrastructure (60-90%) and maintaining it (people, electricity etc.) Whether you send no bits or a million bits per second doesn't matter costwise. 10% full is just as expensive as 88%.

When the network is full. (the amount of data reaches the maximum throughput) You need to upgrade the network. QoS mechanisms won't work.


It's not good business to build multiple networks. Has to do with all the networks offering the same thing, high capex, low marginal costs, mutually assured destruction and stuff. That's why we only have one gas, water and electricity network. Even Telco's like Vodafone and Orange now understand that they should share infrastructure and endorse structural separation. Much to the delight, no doubt, of mrs. Reding.

Telecommunictations networks are not expensive. They are cheap. The costs per year of a high speed fibre optic network to the home, including wireless service is only one quarter ($600 of the average electricity bill $2200).

You can get very richt from just building infrastructure. See Reggefiber and Free.fr. Building networks can be a profitable business as long as you follow the simple business rules of the local grocery store: Don't spend more than you can afford. Make sure people want to be your customer. Make sure the price is right.

Bandwidth is not scarce. Our current appetite for online video still doesn't bring the network down. Andrew Odlyzko sais that the rate of growth of traffic has declined (and if he sais so, it's true).

There are no bandwidth hogs. Every month different people are in the top 10% download group (cho et. al.) There isn't a Pareto distribution where the top 20% are incurring 80% of the network upgrade costs. And with 50% growth per year, there is no use in delaying the increase in network capacity.

The flow of traffic is mostly incoming for consumers. Not outgoing. Cho et al. show a factor 2 to 1.

Misconceptions about the economic/financial side of networks
Usage is a very bad way to determine how costs should be allocated over the network. The amount of data you send/receive may determine a bill, but this says nothing about how costs are incurred. Capital expenditure and operational expenditure is mostly determined by having a a connection tot the network. Not what you do with it.

The type of service you use is an even worse way of charging for network use. An SMS is only 160 bytes. A telephone call at standard GSM quality 1200 bytes per second. The price of an SMS is however higher than the per minute charge for a voice call. As said, bits are bits, there are no bits that incur higher costs than other bits.

Paying per connection is the fair way of paying for a network, separate of the network. Being connected means you're able to use the network in any way you please and therefore enjoying the full benefits of the network. So an owner should pay his/her fair share of the network. Now I know that economic theory allows for differential pricing to get more out of the market, or to offer the basic service at a lower price to those who are not able to pay. In this case i do think that the market is better served with universal access and a single price and some government fund for the poor than by differentiating prices.

The flow of money in a transaction says nothing about the perceived value between two parties, since it is met with an equal flow in goods/services/information.

Price fortunately says nothing about value. Selling below value is very good for consumer surplus.

Time is one of the ultimate resources. A person spending time on a voice communication values the communicaton more than the time involved in the communication.

European customers have absolutely no problem with paying for incoming traffic. If you are European and you read this blog, you've paid for incoming traffic. I don't hear you complaining. You can also observe that when being called some Europeans may offer to call you back if they are on a cheaper plan or the boss is paying. If you offer them an all you can eat model they will go for it.

Little Children in Africa don't get telecommunications because we call them and pay for interconnection. Telecoms in Africa is best served by direct investment and liberalisation of the telcoms market. The economies in Africa are best served by cheap incoming and outgoing data traffic. 10% more mobile phones means 0.6 point extra growth.

Misconceptions about interconnection
The costs of an interconnection between two networks is the cost of a dedicated switch for an exchange and two network ports on each of the networks. The AMS-IX currently switches the equivalent of 5 million parallel VoIP conversations at 80Kbit/s. All for €800 per port per month. Couple of thousand dollar per network per month for multiple interconnects (let's say $100000 per "day, month, week, year) for interconnects on multiple places for 1 million users). The cost per user is about 10 cent per user. Could you spare a dime?

Paying to deliver traffic for an interconnection sais nothing about the value of the interconnect, nothing about the value of the communication and hurts consumer surplus. If you factor in Metclafe's law and the improvements made on it by the aforementioned Andrew Odlyzko, than the act of interconnecting itself generally adds enough value on its own to obliterate any specific value calculation for a communication.

There is no economic rationale for paying more from a fixed line to a mobile network and/or vice versa. The costs of a network are not in the interconnection or use, but in the existence of the network and should therefore be born by those that want the network to exist. If you disagree, then please tell me by dialling +31900XXX which connects to my super-duper-self-designed-gold-plated-QoS-enabled network(€2000 per minute) It's so expensive, because I wanted a network that is completely based on fiction, pidgeons and elfs. Elfs are expensive, but because you want to tell me I'm wrong, you now have to contribute to that network. It's only fair you see.

Misconceptions about Calling Party Pays and Mobile Party Pays/Peering and Transit
CPP is not the reason why Europe is ahead of the US in mobile calling. MPP or Bill and Keep models are used in countries like Singapore and China without any problems in proliferation of mobile technology. Europe is ahead because their was one standard that everyone converged on. In the US different technologies, different handsets, spotty coverage, weird roaming plans, corporate incompetence etc all fought against proliferation of mobile technology. Add to this the fact that areas like Alabama are third world nations in effect and the national income in the US is distributed unevenly and you have some interesting reasons. (BTW could someone do an analysys of distribution of broadband vis a vis income per family etc. might explain too why the US is lagging there)

CPP rates have to be high in order for companies to be able to pay for the networks. Nope, in India they make due with half a cent per minute.

Peering and transit can be profitable, provided that it is done right. So whenever you provide a link between two points you will have to make sure that your price is low enough and dropping quickly enough to make sure that no one will enter the market. When traffic has grown from 100 to 150 your price will have to go down from 100 to 66.

CPP pays for handset subsidies. Nope it doesn't. In some countries handset subsidies are even prohibited by law. Finland and Belgium come to mind, but somebody should check. What really happens is that all income of various sources is poured together. No money is specifically earmarked for handsets. For competition authorities the rule should be: A waterbed effect from a non-competitive area (CPP to keep handsets) to a competitive area (no CPP and cheaper handsets) is a great thing. This would give consumers choice instead of making them feel stupid for not getting a new phone after two years, because they do pay for it.

CPP is more efficient, because it limits calls to the essential calls and therefore makes that we need to invest less in networks. As said before there is no basis for this reasoning. From a market economics perspective the consumer surplus in the United States is much higher than in Europe, because of them calling 4 to 10 times as much.


The European telecommunications market is not competitive as some people think. Both Roaming and Interconnection have such rules that it becomes impossible to compete on price. If you have a 20 euro per month mobile bundle, you can only call 200 minutes at a 10cent interconnect rate. Anybody wanting to offer 400 minutes is committing commercial suicide

All in all it's late and I have to go to bed :-) I have ranted enough, but I hope I got the feeling across that its not just fine to belief that its normal to pay termination charges. That actually there is no reason to do so and that by allowing the system to continue we're actually allowing competition on price to be non-existent.

Goodnight.

OECD Workshop on Fibre Investment and Policy Challenges, Stavanger, Norway, 10-11 April 2008

The aim of the Workshop is to examine fibre investment across the OECD and look for best practices across a range of investment scenarios. The Draft agenda of the Workshop is available in PDF format. Yours truly will also give a presentation with the title: Dig or wait: Is now the best time to roll out fibre? (Funny thing; they already have the title. I haven't even started making the presentation yet) Oh well, I don't need to find a catchy title ;-)

My session will be: Investment opportunities and challenges
The investment session will look at the opportunities and challenges facing existing communication operators and new market entrants (e.g. utilities) as they work to expand fibre coverage. The session will evaluate situations where investment makes economic sense for private firms and other situations where governments may need to play a more active role. It will look at the track record of various public/private sector partnerships and attempt to find some effective recommendations.

First Post

After years of using Lunatic Thought, it just started to sound really lame. I'm no idiot.. no raving lunatic and not really funny either. I just do internet and telecoms on multiple levels. So here it is: Internet Thought! Everything is the same, just the name is different, so update your RSS feeds! (I'll crosspost for a while and take some interesting posts here.)