N. American Internet Policy: Tough Choices Ahead

You've identified the challenge in both fixed and mobile broadband.  We want great broadband but we don't want to pay a high price.  It's like wanting a BMW 700 on a Civic budget at one level; buyers only get what they want if there are sellers willing to meet the strike price.

At another level, technology improvements can make "average" broadband at an "average" price better, I think.  One example would be a program to require operators offer roaming to competitive customers at a reasonably low price.  This would reduce the amount of overbuild in towers and the backhaul costs, and it would free spectrum for other applications.  This is one example of something I think regulators could look at that could have a major impact on mobile costs.

In the wireline space, I still feel (as I blogged over a year ago) that the best approach to "subsidizing" broadband would be to apply a tax credit to the investment.  The tax credit could be based on the length of the fiber run that was created--the longer a run you deploy the larger the percentage that can be written off.  This focuses investment on operators who can leverage access fiber because they already have other facilities, and it also rewards direct invesment in longer-loop rural links.  The current FCC plan really doesn't address rural empowerment, it only subsidizes the operators and thus could well be used to improve broadband in towns where there's already a natural market without subsidies.

Technology could help in some cases, but I think we'd make the biggest difference by just looking at the problem from a business/regulatory perspective.

Tom

But I don't see ISPs surviving without doing it because usage isnt going to stop, and an all-you-can-eat plan is only sustainable after you build a big enough infrastructure, first was fixed lines, now mobiles but broadband has a long way to go, and mobile broadband even more.

That's true, I think when we can presume that there is some root causal factor that's generating the packets and so similarity among them would be likely.  Multicasting is an example of this, as you know.  The problem with mesh routing in a combinatory way is that the drivers of the individual traffic flows are independent human behaviors and so with the richness of possible sites growing, the chances of the behaviors creating any combinatory opportunities would be, I think, fairly low.  The chances would be further reduced if we presumed a need for "hot-potato" handling; move packets through a cell as quickly as possible.  Not only would the need to reduce latency make it hard to look for packet matches, the low latency of the cell would mean few packets would be there at the same time to be compared.

Tom

Tom:
May be we need to re-think how we treat data flow. In most of current communication scenario (like routing e.t.c) data flow is considered equivalent to commodity flow, which is not true. One can not add two cars to get a single car, but in case of data flows one can add two packets each of k bits to get another packet of size k bits.

That's always been my concern, because the only chance for any sort of wireless network to provide significantly better economics than wireline is if the number of backhauls is considerably lower than the number of wireline connections to the same users.  As cell size decreases, user bandwidth consumption increases, or cell-hopping to reach a backhaul increases you face the choise of contaminating performance or increasing the wireline backhauls, which then kills the economy of the wireless network.

Tom

Tom:

You are right; a lot of questions are still need to answered in case of multihop scenarios. Even the capacity of a general relay network is still unknown.

There is a big difference between taking a limited-scope research network that's really a backbone without access, then making it commercially available in a time when dial-up modems were the way to get onto the Internet, versus today with broadband.

The PSTN started as a regulated monopoly with a guaranteed rate of return for providers and a mandate for Universal service, which is what we have now.  That let people run copper wire to most (but even then not all) locations.  Broadband over copper loop works (DSL) but it's limited by the distance.  If you want good broadband you're limited to less than a mile.  When rural density is too low, you can't get copper that close unless you deploy a fiber remote for every house.

I'm sorry but the private sector doesn't do any of those things you say unless the market can generate a suitable ROI.  We have VCs today jumping into investments in everything but basic telco infrastructure--they run screaming from that and I know that for sure becuase I work with them.  Why?  Because they have no realistic chance of a big win there.

Opportunity creates innovation and competition, not just goals, sad to say.

Tom

That all sounds very sensible, but did any of that happen when the govt. let the web go commercial in the first place? 

Yes, there are initial and ongoing costs, but if the govt. announces a benefit to any company that can deliver x mbps, then it's up to the private sector investors to decide whether they can make a profit with that boost. Germany said home-produced solar energy would have to be purchased by utilities for X dm -- above the going rate, and suddenly industry figured out how to put panels up everywhere. When the subsidy ends (next year or the next, I think) then you're right, you may see a shakeout of the companies, but some will survive, and the benefit to the public will endure.

How did the POTS companies manage to deliver rural service? They lobbied for and got a tax on everyone's service. Now that they've covered those costs, there's an effort to roll that tax back, i believe, but it could be converted to this purpose.

By announcing a goal and an inducement, but not involving the govt in the details of how to deliver, we'll let private sector do what it does best -- innovate and drive costs down. If no one bites, then, yes, the goal or the benefit might need to be adjusted, but very little will have been spent or risked, so it's a cheap experiment.

We'll have to agree to disagree on that one!  Even if it were possible to build a network at zero unit cost of bandwidth there would be no providers of equipment at those prices.  The marketplace funnels investment to where a return can be earned, and there is never a return for free stuff.

Tom

I agree with you, Modza, but I think that government subsidization of a real market in any form demands understanding the natural dynamics of that real market.  The taxpayers at the least have a right to understand just what would happen naturally, what the objectives of the government program are, what their benefits are, and what the cost would be.  That means knowing what the natural starting point would be.  A good example is that if we were to say "We'll provide 8 Mbps of broadband to everyone for $20 per month" there will be a natural cost to that, and the difference between the cost and the funding the user provides with the twenty bucks has to be approved by taxpayers.  We could also say "We'll prove 100 Mbps" and the cost would scale up literally like a hockey stick.  To many, saying "8" or "100" doesn't seem earth-shattering, but the basic cost and the ROI that would be provided by what the user has to pay would be radically higher.

This is particularly true when you are considering the population curves in rural areas, and also the issue of whether rural users would in the main be adopters of broadband even if offered.  Remember, more than one of every three households offered broadband elect not to take it.  Are whatever factors that cause this rejection more likely to be present in rural areas?  We could roll out broadband and have nobody come.

Finally, you need ROI consideration on government-subsidized plans to insure that the ongoing operation of the new network is within the reach of the provider.  It doesn't help to build out a broadband network to have it drop dead in 2 years because the company can't service the debt because the ROI is too low.  Remember FairPoint.

Tom