Why “Basic Access Broadband” won’t work

18 March 2006

 

The Annenberg Center Principles for Network Neutrality introduced the notion introduced “basic access broadband”, a minimal level of neutral Internet connectivity. It falls in the middle ground between net neutrality purists and free market purists. 

 

The Principles are a worthy attempt to advance the debate. (Disclosure: I was a participant in workshop where these ideas were discussed.) They bring together some commonly held values (win/win for operators and consumers, light touch regulation, transparency, competitive entry) in a way that clarifies common ground.  However, I’ve realized that the key innovation – Basic Access Broadband – is attractive in principle but unworkable in practice.

 

I’ll assume that the Basic Access Broadband means a fixed minimum amount of bandwidth on which transport neutrality is assured; it also goes by terms like “basic tier” and “basic neutral access”.  Let’s assume that “neutrality” means no enhancement (eg paid prioritization) or degradation (eg blocking) of access to any sites, for any traffic types. Within this definition, many variations are still possible:

 

  • The minimum bandwidth can either be an absolute speed (eg 1.5Mbps), or a percentage of the total bandwidth offered (eg 10% of 20Mbps capacity per user)
  • Basic Access Broadband can be offered as service tier which a customer has access to in addition to any other services, or as “floor” as part of a larger IP service (eg all sites will get at least 1.5Mbps neutral access, but can buy prioritization at higher speeds)
  • Other parameters beyond speed will need to be defined, eg minimum latency (to allow viable VOIP service) and data volume per month

 

To summarize the problems, which I’ll explore in more detail below:

  • One can’t set a meaningful “national minimum absolute speed”
  • One can’t set a meaningful “percentage of bandwidth”
  • Basic Access Broadband as a tier leads to price control, which the FCC can’t deliver
  • Basic Access Broadband as a floor will not prevent bias in favor of some content providers
  • A mandated minimum precludes socially beneficial offers
  • A plethora of parameters will have to specified, leading to endless politicking and litigation
  • Basic Access Broadband doesn’t prevent bias in favor of rich players, but comes at a high social price

 

In short: while Basic Access Broadband sounds simple, its implementation will be prohibitively complex.

 

One can’t set a meaningful “national minimum absolute speed”

  • Network performance varies widely from place to place, depending on the quality of the plant.  It also varies within a particular system; DSL clients a long way from the Central Office (CO) will get much lower speeds than ones close by because speed drops off with distance.
  • Hence, any minimum speed will be very much a lowest common denominator.  For a DSL customer a long way from the CO the best possible speed might be 784kbps; requiring a 1.5Mbps “basic tier” is impossible. On the other hand, 1.5Mbps is nugatory in a fiber system where per-customer speeds of 20Mbps are available.
  • Offered speeds are not absolute, since network elements are shared among many users.  For example, hundreds of customers share a single access channel in cable systems. While it might be easy to get 1.5Mbps in the middle of the night, it might be impossible to deliver at peak times when all users are on the system.  Defining the meaning of a “minimum speed” would degenerated into baroque and arcane rules that would place an intolerable political burden on the process.

 

One can’t set a meaningful “percentage of bandwidth”

  • The problem with a minimum absolute speed leads us to a percentage of bandwidth. However, differences in infrastructure implementation make this an apples-to-oranges proposition.
  • Consider a telco with an all-fiber, all-IP plant.  In order to offer IPTV (5Mpbs/channel, say) they build out a 20Mbps plant. Let’s say the percentage is 10% of their total capacity - they have to reserve 2Mbps for Basic Access Broadband. However, they may choose to devote 75% of their capacity to IPTV, leaving only 5Mbps for Internet access (both Basic Access Broadband and proprietary).  The 10% of IP capacity (2Mbps) is therefore effectively 40% of their total Internet offer (5Mbps).
  • A cable operator chooses to offer digital video service without using IP as a transport.  They choose to reserve 10 Mbps per node (shared among 500 customers) for Internet access.  A 10% requirement here would mean that Basic Access Broadband would only be 1Mbps/node.  Even leaving aside the fact that this 1Mbps is shared among 500 customers, the same percentage requires them to deliver only half as much Basic Access Broadband to customers as would be the case for the telco.
  • Defining the meaning of a “minimum percentage” would degenerated into endless arguments and litigation about the technical and commercial differences between systems.

 

Basic Access Broadband as a tier leads to price control, which the FCC can’t deliver

  • Let’s assume that Basic Access Broadband is mandated to be delivered as a service tier.  If the network operator is free to price this as they choose, they are free to make it so expensive that there will be effectively zero uptake by customers.
  • Prices would have to be set on a local level since market conditions vary widely from place to place (eg degree of competition, and capacity of infrastructure).  The FCC has neither the expertise nor the capacity to do this.

 

Basic Access Broadband as a floor will not prevent bias in favor of some content providers

  • Let’s assume that Basic Access Broadband is a floor, that is, that neutrality is required for (say) the first 1.5Mbps of bandwidth.  
  • Since the client software just sees a single IP connection, there is no way for it (or the user) to distinguish between the treatment that different sites receive. Let’s say that Google Video has paid a network operator for preferential access, and thus it will download and/or stream at 5Mbps; YouTube (say) did not pay, and will have to make do with 1.5Mbps.  The user/client only sees that Google performs better than YouTube, leading to the “tilted playing field” that net neutrality purists are trying to avert.

 

A mandated minimum precludes socially beneficial offers

  • One can easily imagine a service that uses prioritization to partially fund an entry-level service that would be attractive to low-income users. For example, Google could partially fund a $10/month 1Mbps service (ie below market rate) where queries to any major search engine always return Google results.
  • This offering could not be made if the Basic Access Broadband rules applied, since it falls below the speed floor specified.

 

A plethora of parameters will have to specified, leading to endless politicking and litigation

  • I have referred to some potential complications above, eg defining “minimum speed” for shared access systems, and defining “minimum percentage” across different technologies. There are more:
  • One might want to define a minimum latency in order to allow viable VOIP service.  But what should the latency number be, and between which points of the network?
  • P2P applications dramatically change bandwidth needs, since P2P clients effectively use up all available capacity.  P2P file sharing also offers an alternative way to get video programming that could compete with network operators’ offerings.  Network operators therefore have an incentive to limit P2P applications.  If they are precluded from blocking access to sites or P2P traffic flows, they may choose to set monthly download limits.  In this case a Basic Access Broadband definition that aimed to level this playing field would have to define a minimum per-month download allowance.
  • The Basic Access Broadband solution assumes that “neutral” and “proprietary” traffic will run over the same infrastructure. Network operators already manage traffic, and will continue to do so.  Preventing traffic shaping from discriminating against “neutral” traffic partition will require complex and debatable rules.

 

Basic Access Broadband doesn’t prevent bias in favor of rich players, but comes at a high social price

  • Since the Basic Access Broadband approach allows prioritization in the non-neutral part of an offering, it doesn’t prevent content providers buying preferential access.  It thus does not address one of the complaints of net neutrality purists.
  • Since it is defined as a minimum bar, the quality of service on the Basic Access Broadband partition is likely to be poor.  Some form of access is guaranteed, but it’s not particularly good.  
  • On the other hand, the complexity of rules that will be required to implement Basic Access Broadband will impose a heavy social burden, and is likely to be subject to regulatory capture by well-funded players in any case.