Communications

US Mobile Internet access prospects looking up

This is a very good week for the mobile Internet in the US.  Our best prospect for open mobile Internet access is not legislation or regulation, but having four or more competing networks that are technically able to offer mobile broadband access.

We have three such networks today — Verizon, AT&T and Sprint — but three is not enough to break the walled garden mentality.  What's changed?

1.  T-Mobile USA has launched their first 3G service using the spectrum they won in the 2006 AWS auctions.  For now, it's only New York City, but Reuters reports that T-Mobile plans to launch in 20 to 25 new markets by the end of the year and T-Mobile's stated intention is a full national HSPA network. In 2009, this will be our fourth national 3G network fully capable of multi-Mbps down and multi-hundreds Kbps up.

2.  Clearwire has cut a deal to take over Sprints WiMAX network.  As the Wall Street reports (subscription required) today:

Sprint Nextel and Clearwire are close to announcing a $12 billion joint venture that plans to roll out ultra-fast wireless Internet access for cellphones and laptops in coming years, with the backing of an unlikely alliance of technology and cable companies. Sprint has agreed to merge its wireless broadband unit with Clearwire, a Kirkland, Wash., firm founded by cellphone pioneer Craig McCaw. The new company has raised a total of $3.2 billion in outside financing from several heavyweights -- $1.05 billion from cable provider Comcast, $1 billion from Intel, $550 million from Time Warner Cable and $500 million from Internet giant Google. Smaller cable provider Bright House contributed $100 million. The investments value the new company at more than $12 billion.

This also reduces Sprint's financial exposure and hopefully reduces the likelihood they will be taken over or their network consolidated, at least in the short term.  I've been negative on the prospects for WiMAX in the past, but if anyone can make this go, Craig McCaw is good bet.  So Clearwire represents our fifth national network capable of delivering mobile broadband Internet access.

Assuming all this holds together, we will see affordable flat rate open mobile Internet access in the US by 2010.

Models for Muni WiFi completely neglect technology evolution

Modern travel means interminable waits, but it's a good time for reading. I finally read Wireless Pittsburgh: Sustainability of Possible Models for a Wireless Metropolitan-Area Network by Jon M. Peha, published in February as a working paper of the New America Foundation.

The good news — it’s full of interesting cost estimates and projected subscriber take rates based on specific demographics in Pittsburgh, Minneapolis and Philadelphia.  The paper also examines a range of business models, in detail, from complete monopoly to structural separation (wholesale–retail) to let-the-market-take-care-of-it.

The bad news — all of the models turn out to be extremely sensitive to revenue assumptions, i.e. to the estimates of subscriber adoption and willingness to pay.

The flaws in this study

All of the models compute a net present value based on five years of stable operations, but there is no mention of technology evolution or adoption rates of competing broadband services, i.e. cable and telco (DSL or FiOS) services since this is a US study.  Technology is evolving at a great rate.  You can’t bet on stability.

During the past five years we saw WiFi go from 11 Mbps to widely deployed 54 Mbps systems and bleeding edge (pre-standard 802.11n gear) systems doing well over 100 Mbps.  The last five years also saw costs decline to the point where we see widespread deployment of WiFi by individual consumers, a significant percentage of which are running open WiFi hotspots. 

On a recent drive through three different neighborhoods in Portland Maine, I was interested in looking up real estate information on the web.  On each of a half dozen occasions, I was able to find a open WiFi hotspot within one city block of deciding I wanted to connect.  In January, I was in north New Hampshire and spent two nights at small motel (not part of any chain) in Littleton, NH.  They had no Internet on offer, but a quick check for WiFi signals revealed two within range of our room.  If you don’t like my anecdotal information, look at the WiFi hotspots that have been mapped by Navizon.  It’s very different than five years ago.

No matter how much it simplifies the analysis, you can’t bet on stability.

What might happen with WiFi in the next five years?  The latest WiFi specifications add multiple input, multiple output (MIMO) support, additional modulations and other goodies.  As low cost WiFi routers incorporate these advances we’ll see speeds go over 300 Mbps, but more importantly MIMO technology increases both range and directionality.  This means overlapping systems work better (despite their overlap) and the signals from isolated systems reach further.

If you’re worried today’s open systems will be locked down, then spend your time thinking about schemes like FON which offer more secure ways for consumers to share WiFi bandwidth.

If you’re worried consumer solutions won’t reach the inner city, perhaps someone needs to relook at where WiFi has already been deployed, and then forecast what might happen over the next five years, given the cost of a basic PC is approaching that of a television and the Cable and Telco duopolists both push triple play bundles. 

Don’t short change technology evolution. 

NY Times grossly misreads WEF report

Today's New York Times includes an article by John Markoff entitled "Study Gives High Marks to US Internet."  But either John Markoff is fuzzy about exactly what the Internet is or he didn't actually read the report.  His title is way off base.  He did interview a few people who are quoted in the latter part of the article, so there is some information in the article.  But he's done a major disservice for the many who read only the title or perhaps first paragraph.

The study in question is the Global Information Technology Report recently published by the World Economic Forum.  While the printed report costs money, the summary is on-line and an interactive version gives access to all their data.

This is a report on information technology not specifically on the Internet (just look at the title).  The researchers measured 68 different attributes of information technology, only a few of which directly pertain to the Internet, e.g.

Attribute	US Rank
Internet users (per 100 pop.)	7
Internet access in schools	12
Broadband Internet subscribers	17
Internet bandwidth	19

To his credit, the first section of the report is about "network readiness" and the US is ranked #4, however the study's definition of network readiness includes all sorts of features of the broad business environment, regulatory aspects and computing infrastructure.

Prior to this, I remember John Markoff only for the book Takedown, which I enjoyed.reading even though my sympathies were with Kevin Mitnick.  :-)  So this article is a major disappointment.

Here are the 68 attributes the World Economic Forum study examines:

WiMAX in the US, a limited window of opportunity?

Their first service launch has been delayed, but Sprint Nextel CEO Dan Hesse repeated his vow to blanket the US with a WiMAX network in his talk at CTIA this week.

I sincerely hope he succeeds.  We need as many competing data networks as possible, if we're to see any measure of open mobile Internet access. However, I'm worried.

WiMAX has been quite successful in emerging markets, providing fixed wireless Internet access in countries as diverse as Pakistan, Bulgaria, Nigeria, Georgia, Ethiopia and Georgia (the country!).

The US is another story.  The only significant US deployment is Clearwire's with roughly 400K subscribers, but on a mostly “pre-WiMax” network.  Again, the application is fixed wireless Internet access. 

It’s one thing to use fixed WiMAX to provide Internet access in Pakistan.  It’s another thing to compete for fixed access in the US.  Yes, our telephone & cable duopoly is moving slowly, but they are going after all the more profitable neighborhoods with broadband offerings substantially faster than what fixed WiMAX provides.

What about mobile WiMAX?  Mobile WiMAX is in trials today, using technology and providing performance that the 3GSM community will only see 2–3 years from now.  Sprint clearly hopes to use WiMAX as a springboard past its competitors and past concerns about its declining user base.  Presumably, in the longer term, they hope to converge their dissimilar networks (Sprint EVDO and Nextel iDEN) on mobile WiMAX.  But can mobile WiMAX build a large enough market soon enough? 

Volume drives down cost and cost advantages win in the end — witness Verizon’s announcement that they are jumping ship on Qualcomm’s CDMA evolution in favor of the 3GSM community’s long term evolution (LTE).  Today, GSM networks (GSM/ EDGE/ W-CDMA/ HSPA) have 80% of the world mobile phone market with billions of subscribers and a billion or so handsets manufactured each year.  Right now the entire Sprint-Nextel customer base is ~54 million subscribers.  Perhaps emerging markets will also adopt mobile WiMAX, thus driving up volumes?  Unfortunately emerging markets are even more price sensitive with the high volume application being basic voice telephony plus SMS.  GSM is the lowest cost solution by a wide margin. 

I hope I'm wrong.  I hope mobile WiMAX's performance lead (over LTE) is enough to carry the day.  And, in particular, as I’ve written before, I would very much like to see Sprint succeed, with or without Clearwire, because their WiMAX network would represent yet another source of wireless Internet access.  With four of more networks capable of mobile broadband Internet access, competitive pressures alone should give us what the FCC is currently ignoring, i.e. mobile data plans that are both open and reasonably priced.

 

Broadband penetration vs. broadband capabilities

I'm listening to the first session in the VON pre-conference workshop on Competition Policy and getting increasingly depressed.  Everyone on the panel is entrenched in traditional policy infrastructure — no thinking out of the box here.  What's more, at least two of the five people on stage seem to think the US is doing well, among OECD countries, with our deployment of fixed line broadband access.

The problem is definitional.  They are comparing broadband penetration rates.  I am comparing the relative performance of readily available broadband services.

The FCC's defines "broadband" as at least 200 Kbps in at least one direction.  The OECD uses 256 Kbps, while the ITU definition is “transmission capacity that is faster than primary rate Integrated Services Digital Network (ISDN) at 1.5 or 2.0 Megabits per second (Mbits).”  Using the OECD's definition, the US comes in 15th for broadband penetration by country.

But that's not the point!

Speeds of available broadband services

In Stockholm, dark fiber reach to every city block and an enormous number of buildings.  This dark fiber is available for lease by anyone.  As a result, there are many ISPs offerring services in greater Stockholm (and elsewhere in Sweden).  Just about anyone in Stockholm can get 100 Mbps symmetric Internet connectivity for 98 Kroners pe rmonth — that's $16 per month.

Similar service (100 Mbps) in Tokyo costs ~$26 per month and in Seoul it was closer to $30 the last time I checked with friends there.

In the Boston suburbs I have Verizon FiOS Internet access.  Downstream capacity is 20 Mbps most of the time and upstream measures out at 2.7 Mbps.  This costs $50/month.  I'm extremely fortunately to be able to get service this fast.

So my friends in suburban Stockholm have 5 time the downstream capacity and 30 times the upstream capacity for 1/3rd the cost.

Why don't people talk about broadband performance?

The impact of cellphones on the price of grain in Niger

On the plane from Amsterdam to Kuala Lumpur I caught up on some more reading, this time a paper by a Berkeley economist,  Jenny C. Aker,  "Does Digital Divide or Provide? The Impact of Cell Phones on Grain Markets in Niger" (January 15, 2008).

Long time readers may recall my post about Robert Jensen's 2006 study of the advent of cellphones in the Indian state of Kerala and the positive impact they had on fish markets.  Aker's study doesn't have the beautiful graphs that Jensen provided, but the quantitative results are similar.

Due partly to costly information, price dispersion across markets is common in developed and developing countries. Between 2001 and 2006, cell phone service was phased in throughout Niger, providing an alternative and cheaper search technology to grain traders and other market actors.  ...we use a unique market and trader dataset from Niger that combines data on prices, transport costs, rainfall and grain production with cell phone access and trader behavior.

The results provide evidence that cell phones reduce grain price dispersion across markets by a minimum of 6.4 percent and reduce intra-annual price variation by 10 percent. Cell phones have a greater impact on price dispersion for market pairs that are farther away, and for those with lower road quality. This effect becomes larger as a higher percentage of markets have cell phone coverage.

Niger is a landlocked country in western Africa.  It's extremely poor and most of the population are subsistence farmers.  Among the interesting conclusions of this study is the introduction of cellphones produced the equivalent of a 3% to 8% decrease in the price of grain together with an increase in grain traders income.

Particularly notable was the largest differential (8%) occurred during a food crisis in 2005 when people with access to cell phones fared better than those without.

More on fishermen with mobile phones in Kerala, India

Two years ago, Robert Jensen presented work he had done on the economic benefits of mobile phone adoption among fishermen of the Kerala state in India.  Jensen showed that mobile phone adoption reduced price volatility in the fish markets and was able to show quantitative benefits for fishermen and for the fish consuming public in Kerala.  This blog post has a summary, some graphs and the pointers to his initial presentation.

Jensen's paper on this topic was subsequently published by the Quarterly Journal of Economics, Volume 122, Issue 3, August 2007.

Now there's new data from Kerala on the transition that occurred with the advent of mobile phone service.  In Volume 4, Issue 1 (Fall 2007 ) of Information Technologies and International Development, Reuben Abraham reports (PDF, registration required) the results of a series of focus groups conducted at 12 locations in Kerala India and interviews with nearly 200 local people associated with the fishing industry.  Using this data, he provides a good view of the fishing supply chain in Kerala, details on what happened when mobile phone service became available, and a nice summary of the implications for policy makers:

This article has provided an example of the value of communications technology, especially mobile phones, in making markets work more efficiently. This research also firmly establishes the role of ICTs in rural markets as one that reduces transactions costs. Investments made with the aim of reducing transactions costs are more likely to succeed than amorphous, ill-defined attempts to bridge the “digital divide.”

What I found interesting was the complexity of the fishing supply chain with separation between fisherman, boat owners, fishing cooperatives, brokers/commission agents and wholesale and retail fish merchants.  Apparently the economic benefits measured so far have come with little change in this complex supply chain.  However, Abraham notes:

... the power the middleman holds over the fishermen due to the monopoly on price information has lessened somewhat. The free flow of information ensures the fishermen get the opportunity to drive a harder bargain than before.

It will be interesting to see what changes happen over the next 5-10 years as ownership of capital assets (like boats) evolves and the market adapts to the new flow of information.  I'll bet that, increasingly, middlemen will be cut out and consumers will see even lower prices.

Similar motivations for mobile and Internet usage in rural Zambia

Over the years I've uncovered a number of serious studies of the social and economic impact of mobile phone adoption in emerging markets.  In fact, my list of links (embedded in various blog posts: 1, 2, 3, 4, 5, 6, 7, 8) has become a reference for several people, for example here. I've been somewhat less organized regarding the social and economic impact of Internet access, but I've resolved to track such references as well. 

Here's something I stumbled on this weekend -- a study by Paula van Hoorik and Fred Mweetwa of TNO Information and Communication Technology in the Netherlands, entitled "Use of internet in rural areas of Zambia."

What struck me in this study was the parallels between the social and economic benefits they found and those conventionally associated with the adoption of mobile phones.

Based on our study, the most important social benefits are:
• Internet enables people to keep their network and enlarge it by communicating with friends, family and others
• Internet enlarges the world of people in rural areas by giving access to information
• Internet brings knowledge and supports education

The most important economic benefits are:
• Reduction of commute time
• Saving money in a lot of different ways, such as only traveling to pick up something when you know it is actually there instead of having to return multiple times
• Bringing new opportunities and using them, such as learning new farming methods or opening an internet café to make a living.

Most studies of mobile phone adoption in emerging markets, especially amongst the poor, show that social reasons - communicating with friends and family - come first, economic advantages come second.  And among both social and economic advantages, the ability to eliminate journeys, i.e. save time by not having to walk to the next village to ask a question, is extremely important.

I suppose I shouldn't be surprised.  At root, they are both communications technologies and our use of such technologies is driven by basic human needs that are remarkably similar, for everyone, everywhere. 

How tiered Internet pricing could actually facilitate P2P

Time Warner Cable's planned experiment with tiered charging for Internet access has generated a flurry of coverage in the blogsphere, but no new insights (at least that I've seen).

The primary problem ISP's complain about is that 5% of their customers use 90% of the available bandwidth and when they examine this traffic, it's mostly peer-to-peer file sharing.  A reasonable question is how to allow as much of this traffic as possible without increasing an ISP's variable costs or slowing down their other users.

This may not be as difficult as it appears.   Indeed if Internet access was as competitive as mobile telephony, we might already have seen what I'm about to propose — a combination of bundled pricing equivalent to mobile's "free nights and weekends" and "free on-net calls" with a way to facilitate P2P traffic that leverages exactly these "free" periods.

An ISP's costs

ISPs have some costs which are relatively fixed and others that are tied to usage.  A network is a relatively fixed cost and when it's not full, the incremental cost of adding traffic is zero!  This is the reason mobile operators give away free nights and weekend.  They've built their mobile network for the peak daytime traffic, so it costs them nothing to run promotions that add incremental traffic at off hours.  Peak hours and off hours may be different for an ISP, but the concept is the same. When a data pipe is lightly loaded the ISP's cost of adding incremental traffic is zero.

On the other hand, some ISP costs are usage based, for example "IP Transit" or more properly, Internet Transit.  This is the ISP's upstream cost to send and receive traffic to/from the rest of the Internet.  However, even here, usage-based costs occur at heavy usage.  Light usage periods don't save money.  To understand what's happening, it's worth a digression on Internet Transit.

Internet Transit

Internet access is monopoly or duopoly or a heavily regulated industry.  The middle mile connections from the local network to the Internet backbone may or may not be competitive depending on where you are.  But the Internet backbone itself is extremely competitive.  If you can get to a major Internet Exchange Point in the US or Europe, there are many providers offering extremely competitive rates for Internet Transit.  Typically these services are priced on a megabit per second per month basis (Mbit/s/Month) with lower rates for higher volume commitments.  The other key idea is that charges are based on the 95th percentile of all the five minute data rate samples taken during the month.  So an ISP can have a few bursts above their typical rate, as long as they represent less than 5% of the sampled intervals.

But this also means there is no extra cost to run at or near the typical rate at all times.

Local traffic

Even more important, if file sharing is done with other computers on the same ISP's network, then there is no need to pay for Internet Transit at all.  The question is how to figure out which potential peers are "on-net" and which are "off-net."

Sending signals to P2P software

Most P2P file sharing software has relatively little knowledge of locality.  Some P2P software practices "prefix awareness," for example, Joost gives preference to peers in the same /24 IP address block when they are available.  But if a major operator provided an automatic way for P2P client software to determine whether a prospective peer's IP address was currently reachable "for free", it seems likely the file sharing community would leap on it, and if there's money to be saved, active file sharers would download the new clients immediately.

A standard way to present such information might be via an extension to the XML-based response codes in one of the whois information exchange proposals, e.g. from ICANN or from APNIC.  Also, while what I'm proposing might start as a pricing plan rather like a mobile operator's "free nights and weekends" and "free on-net calling," it's not hard to see extensions where an ISP could offer dynamic access to underused capacity to those programs that were prepared regularly interrogate an ISP's server and use just the advertised off-hours capacity.

In closing

People liked fixed price deals.  Unlimited is great, but there's plenty of experience with bundles of minutes and the idea of data bundles has already showed up in 3G mobile data plans.  The combination of several tiered data bundle prices with the availability of "free" connectivity for "on-net" peers and during off peak intervals is likely to appeal to file sharers and produce better results for both the sponsoring ISPs and file sharers alike.

Good News: Pan-African Roaming following US not EU model

Last year Celtel launched a multinational mobile phone service for Celtel subscribers in Kenya, Tanzania, and Uganda under the name One Network which

"allows customers to move freely across geographic borders without roaming call surcharges and without having to pay to receive incoming calls."

Now I see they've extended the service to include twelve countries. This is very good news, as there are many tribal and cultural groups in Africa that span national borders.

So it's encouraging to see Africa following a path that should lead to roaming competition and long term consumer benefit.  To understand what's at stake here, let me refer to an excellent paper by Scott Marcus on Europe's "New regulatory framework" which compares the US & EU regulatory approaches to roaming.

"In 2000, Vodafone Airtouch merged with Mannesman.  <EU> Commission competition authorities were concerned that the merged firm would be the only entity able to offer pan-European mobile telecommunication services...  <under pressure the> merging parties resolved competition concerns by agreeing to provide roaming tariffs to affiliated and unaffiliated mobile operators on a nondiscriminatory basis.  As a practical matter, this eliminated the merged entity's incentives to offer pan-European service packages..." 

Scott compares this with the US where AT&T Wireless's introduction of Digital One Rate service led to a wave of mergers, alliances and joint ventures as competitors were forced to respond.  Because of that competition, US mobile subscribers now get large buckets of low cost minutes for which there are no long distance and no roaming charges.  As a result, US average minutes-per-subscriber-per-month is four times that of Europe.

Both the US and the EU have well intentioned regulator's, but as far as I know, neither can claim credit for understanding what they were doing or not doing.  Luckily in Africa, there doesn't appear to be an African regulator with any interest in pan-African roaming, so hopefully African roaming will follow the path of competition.  The initial signs are positive.