Investment and demand in fiber to the home

There are a couple of threads that, juxtaposed, are interesting. First, Australia moved ahead with its ambitious fiber-based Internet project, and second, this report gives one pause as to whether it would make sense in the US.

So is the US foolish in not making larger investments in infrastructure like our trading parters, is it wise?

The answer is not so simple and certainly not obvious because of the enormous delay involved in rolling out infrastructure. AT&T witnessed this when it signed on as the exclusive carrier of Apple's ground-breaking iPhone. Indeed, they are arguably still trying to make the investments due to the runaway success of the device.

Would the same thing happen with some as-yet unknown device (Device X), except with end user bandwidth? We don't know if Device X exists, or, if it does, what the impact on existing networks would be. The iPhone has shown us that the potential exists, but how can carriers (and their shareholders) be convinced to make large investments based on revenue streams that are highly uncertain at best? Should the government take these risks?

AT&T and the Internet of Things

This item over at GigaOm is interesting. To me, beyond the numbers, the most interesting observation was this:

The irony here is that M2M connectivity in many ways represents the dumb pipe future that AT&T is so worried about — it’s not providing anything to its partners but the bits. On the call, AT&T executives explained that the number of bits sent via the network are high-margin bits and the machine-to-machine clients have very low churn.

Customer acquisition costs in wireless are quite high, so low churn is a rational cost management strategy. The observation is right that M2M clients would not turn over very fast, if at all.

Economics of Wireless ISPs

In the FCC's Electronic Comment Filing System, you will find this contribution from the Wireless Internet Service Providers Association (WISPA). What is notable about this comment is that it contains useful traffic engineering and cost data. For example:

• "Middle Mile" capacity needs range from 50-260kbps per user and should be 5% of the aggregate bandwidth supplied to last mile business customers
• Second mile capacity needs range from 100-300kbps per user.  This is higher because it is averaged over fewer customers.
• Rural ISPs pay more than $200 per megabit for OCn connectiveity
• Capital expense for a Part 15 microwave is up to $10K and for a Part 101 (licensed), this cost goes up to $15-$20K.
• In the second mile, a DS3 router interface costs $8K and a fiber interface module $200.  
• One rural ISP reported paying $4K/mo for a burstable DS3 to support 250 rural customers.  This translates to $12/mo per customer for middle mile transport and limits how cheap service can be.
• For Internet service, WISPs pay from $2 to $300 per megabit per month to Tier 1 providers, depending on where they are.

There is a lot of info embedded in this paper, though it isn't in tabular form.  Instead, it is in more of an anecdotal form.  In any case, if you're interested in the economic foundations of wireless internet provision, you will find this an interesting read.

Separately, this Wiki, which describes a research project over at KU, is worthwhile reading.

Internet censorship as restraint of trade?

I found this item interesting, though I have yet to read the full report. It reminds me of how telecom reform in the EU was developed and pushed under the guise of competition policy rather than communications policy in the early 1990s. In similar fashion, the US FCC introduced competition into telecommunications not as a common carrier matter but as a frequency allocation matter (the "Above-890" decision in the late 1950s). It clearly matters how you frame the question!

"Censorship is the most important non-tariff barrier to the provision of online services, and a case might clarify the circumstances in which different forms of censorship are WTO-consistent," said the study by Brian Hindley and Hosuk Lee-Makiyama.

Economics of SIP

I found this article over at GigaOm interesting. The article argues that Skype's "Global Index" technology is much cheaper than SIP, which is a significant factor in its (i.e., Skype's) success. I only wish they had provided real numbers. I wonder if anyone has done a careful cost analysis of VoIP signalling?

NANOG 47

There are some interesting presentations up at the NANOG 47 website. In particular, this paper is an interesting global look at network traffic, and this one seems to be a nice presentation on LTE.

Update: Here is an article that discusses the first presentation.

Open source and marketing

This article over at Ars Technica is interesting. Apparently, ipoque, a manufacturer of "Deep Packet Inspection" (DPI) equipment has released the code for parts of its key inspection engines to reassure the public that it does not include the ability to store personal information associated with users, which has been a concern voiced about these technologies.

The interesting thing about this article for me was the tradeoff made by ipoque between the value of their intellectual property and the value of the PR they would gain, followed, they must have presumed, by increased sales.

As the article in Ars points out, that privacy isn't the only concern; others are concerned about DPI from the perspective of bandwidth caps for certain applications, which would be enabled by this technology.

Economics of content on the web

I only follow this topic in a casual way, but I found this article to be interesting, especially given the challenges being faced by the traditional news organization. Quoting the article:

The vast majority of the value gets captured by aggregators linking and scraping rather than by the news organizations that get linked and scraped. We did a study of traffic on several sites that aggregate purely a menu of news stories. In all cases, there was at least twice as much traffic on the home page as there were clicks going to the stories that were on it. In other words, a very large share of the people who were visiting the site were merely browsing to read headlines rather than using the aggregation page to decide what they wanted to read in detail. Obviously, this has major ramifications for content creators’ ability to grow ad revenue, as the main benefit of added traffic is the potential for higher CPMs.

So, as always, the big question is how you get the incentives right so that people can be compensated for creating valuable content?

Broadband carriers and government funding

This article is interesting. According to the atricle

As the Aug. 20 deadline nears to apply for $4.7 billion in broadband grants, AT&T, Verizon and Comcast are unlikely to go for the stimulus money, sources close to the companies said.

Their reasons are varied. All three say they are flush with cash, enough to upgrade and expand their broadband networks on their own. Some say taking money could draw unwanted scrutiny of business practices and compensation, as seen with automakers and banks that have taken government bailouts. And privately, some companies are griping about conditions attached to the money, including a net-neutrality rule that they say would prevent them from managing traffic on their networks in the way they want.

While it is quite possible that some of the rules, such as "network neutrality" may affect them anyway, it is clear that the carriers felt that the cost of participating in this program outweighed the benefits. A significant part of their concern is related to uncertainty about the consequences of an irreversible commitment. Thus, it seems an apt subject for a real options analysis.

Doing such an analysis rigorously would be challenging since the uncertainty is not easily quantifiable. But clearly carriers have concluded that the high probability of a modest upside does not outweigh the uncertain probability of a potentially large downside.

Phone line shrinkage at AT&T

I have blogged before about the decrease in access lines (see this, for example). While it is not surprising given the increase in wireless only households, this article over at GigaOm shows that the decline has been in the 6% per year range for AT&T, not the 3% range.

As the article correctly points out, this is one of the reasons that the large ILECs have been aggressive in rolling out their broadband infrastructures. Since consumers are increasingly opting for wireless for voice, the only way that the ILECs have to continue receiving a share of the consumer's communications expenditures is to build out broadband, which enables them to compete with cablecos for television and internet access expenditures.

If they don't they have to depreciate their infrastructure at a faster rate than consumers are leaving it, else investors (the company owners) will be left holding the bag. Of course, this is an end-game that they would only play if they decided to cede the marketplace to other access providers. There is no sign that ILECs are interested in that strategy!

Whither Internet traffic growth?

In case you've never seen it, this site is interesting. The MINTS project's objective is to collect and study public data about the growth rates of Internet traffic. Interestingly, they conclude that traffic growth, while substantial, is far below the hype ... typically 50-60% per year.

Sprint and Cogent

If you have followed the news, you may have come across this item last Friday. A more detailed technical description is here. The world of interconnection in the Internet is cloaked in non-disclosure agreements (NDAs), so when a story like this breaks, it is useful to look at it in more detail. Not surprisingly, both Sprint and Cogent were pointing fingers (see Sprint's version here, and Cogent's here).

As of this writing, the link has been temporarily restored, but you can expect that this will make waves again. It can be quite difficult for a carrier to maintain the traffic levels needed for peering, especially with a large ISP like Sprint.

Bandwidth caps

For the past several months, there has been much discussion about carriers such as Comcast imposing bandwidth caps on their users. The stated motivation for this is to rein in the highest bandwidth users, and perhaps to set the stage for price discrimination, where users would pay more for higher caps. To this end, there are a few items that I would like to bring to your attention.

This item over at BusinessWeek takes a look at the rationale for bandwidth caps. Dispite the claims of congestion, the article does not find evidence of the feared growth in traffic.

Secondly, this item over at GigaOm points to a recent white paper on the subject. It then points to some possible unintended consequences of this approach. By the way this is a critique of the cited paper over at the Tech Liberation Front. Tim Lee is a fan of metered usage.

Internet transit prices

David Clark's paper on the incremental cost of IP service at the recent Telecommunications Policy Research Conference was quite interesting, so this item over at Telegeography caught my eye. According to their research, transit prices have been declining over the past year and vary significantly across the world. The figure below captures some of this data.

State of the Internet according to Akamai

I found this report interesting (free registration required). It reports data as seen by Akamai, which serves many of the Internet's web sites. While the data are interesting, there are some notable countries missing ... Russia, for example ... so, while of interest, it may of limited value.

Autsralia's National Broadband Network

In an earlier post, I had discussed Australia's plans for construcing an (apparently) subsidized broadband network. According to this site, proposals for this network will be due in November 2008. Do you think this is a reasonable approach, or should broadband be privately provided by market incentives?

Peering and Transit article

Interconnection in telecom is one of the more challenging aspects of telecom. This item from TLF is an article/anecdote that captures some of the dynamics of this problem.

Regulation of cyber cafes in India

I found this article over at BusinessWeek interesting. The article shows how regulation can stifle business activity, even if it is not price regulation.

There were some useful facts buried in the article that I'd like to pull out, though:

... India is trying to increase household PC penetration, which is currently at just 2 PCs for every 100 households, says the technology trade group NASSCOM, and broadband connectivity, an abysmal 4 million connections, vs. China's 3.2 million new connections every quarter, according to BNP Paribas. Even Vietnam, with a population of just 84 million, is signing up 120,000 new broadband users per month, according to IDC.

Mobile Internet usage

This item reports on mobile Internet use. The authors found that approximately 15% of mobile users in the US access the Internet from their phones on a regular basis, leading the 16 nations tracked by Nielsen (UK was second at 13%). Not surprisingly, these users prefer "unlimited" Internet packages (though there are often limits in reality). Yahoo Mail was the top channel accessed by the mobile web users, followed by Google search in a relatively distant second. The authors believe that the US mobile Internet market has reached critical mass, with approximately 40 million users.

Recent OECD reports

In advance of the Seoul Ministerial Meeting on the Future of the Internet Economy, two useful reports were released:

• This report on convergence is a nice summary of the implications of NGNs, with ideas for regulators to consider.
  
• This statistical profile to support discussions at the meeting.
  

In addition, this report was released as an outcome of the ministerial meeting, as well as the Seoul Declaration.