Fast Net News, DSL Prime, DOCSIS Report

Telefonica totally dominated Spanish broadband until a few years ago but actually lost 23K DSL customers in February. Since January 2010, alternative carriers have added 667K while Telefonica gained only 138K. The cause is no mystery; the CMT finally revised the unbundling rules to give the competitors a better deal.

    In the first four months after launching 30 meg and 50 meg DOCSIS 4.0, Ono Cable signed up 128K subscribers (bandaancha.eu.) Some were upgrades of existing customers, but Ono has net adds of 65K since last September and Telefonica only 13K. Ono, like Numericable in France and Virgin in Britain, has severe financial constraints and is limited in how effectively it can compete.

       Spain is finally getting some fiber home. 9,000 signed up in January and February, to a total of 66K. Telefonica has made some extravagant promises to fiber the country, including all of Madrid in a few years. The actual build is much slower as Telefonica has diverted investment funds to the faster growing - and more competitive - markets in Latin America. 

       Before we praise CMT, the Spanish regulator, they need to explain why they want to raise the access charge by 7 percent, to €8.32 a month. (NYT) If regulators want fiber, the right move is obviously to reduce the return on copper. Matthias Kurth got that one right, cutting Germany's access charges.

    Update 14 April Telefonica presented to investors a dismal prospect for its business in Spain and up to 6,000 layoffs. That's one way to make the numbers work for a few years but only a temporary solution. 

As DSL growth ends in developed nations there are losers as well as winners. Sky, on the other hand, had a record quarter with 204K net adds and BT added 188K.  CEO Dido Harding has a colorful description of what went wrong. TalkTalk, she said, resembles a “teenage octopus. Like companies that have grown very fast . . . we don’t have control of our arms and legs, like a gangly teenager. And we have a lot of arms and legs. The teenage octopus will take a while to grow up.” (FT) 

    One primary problem was severe customer service problems after they took over Tiscali UK. "Customers ... have had a pretty rotten experience," Harding says. "We have described buying Tiscali as being like a snake that has eaten a goat. We are busy digesting an enormously complicated business." 

    One promising fact is that they have 85% of customers on their unbundled network. They are well along on a plan to unbundle 93% of the British population. Their network is now dark fiber and Gig-E, dramatically lowering their cost per bit. That allows them to include "unlimited" bandwidth in their $25/month service, which typically is 8-15 megabits (up to 24 megabits max.) Voice line required.

   The problems at TalkTalk have been pretty bad. They achieved "the dubious distinction of receiving the Daily Mail's wooden spoon as Britain's worst organisation for customer service. Our readers complained of rude, insensitive staff, calls not being returned and emai ls unanswered. Some TalkTalk customers were even pestered by debt collectors  -  after bills had been settled or contracts cancelled." Dow Jones adds OFCOM has launched a second investigation into their customer service. Harding insists it will get better, but that won't be easy after she just fired 590 people. 

Your mileage will almost certainly differ, but one lucky home saw their speed go from 1 megabit downstream to 8 megabits. More typical is a doubling from perhaps 1 meg to perhaps 2 meg. The current unit is aimed ay low speeds beyond 12,000 feet. It's logical for Actelis to also offer units designed to bring 2-3 megabit lines up to the new FCC broadband definition of 4 megabits.

     Essentially, you plug in a small box around the middle of the line and let it do some magic. It's sized for 1, 2, 5, 25 or more lines and works off the phone line power. Once installed, it's designed to work indefinitely without maintenance. There's little opex after the initial installation and pole-climbing. The components are inexpensive, so the price of the unit itself can be very reasonable if purchased in large quantities.

    Actelis purchased the product IP of Phylogy’s Triplestream, line conditioner/amplifier which has been on the market since 2005. They tweaked the product to reduce interference and are re-launching it as the "Actelis Broadband Accelerator." "We've solved the spectral compatibility problem," Actelis' Eric Vallone tells me. "That's our secret sauce. Most customers getting 1 meg will see a 100% improvement."

    The earlier design has sold over 100,000 lines, many in South America. There are plenty of places this can be very helpful, even if it's not ideal everywhere. Earlier distributor Charles Industries has an excellent 15 page product brief with installation and troubleshooting advice. 

    The first version of this story said "The unit is polarity-sensitive. Don't plug it in backwards." But Vallone writes me the new model has minimized the problem.

Something like 1/3rd of the homes that can't get broadband were ignored by the Broadband Map. At least a million homes and possibly three million have very local problems that prevent them getting service such as out of date remote terminals. NTIA recognized this and the NOFA required bidders "to gather data at the address-level on broadband availability, technology, speed, infrastructure, ARPU, and, in the case of wireless broadband, the spectrum used, across the project areas."

    Everyone involved in broadband mapping realizes that specific data by address is required for accuracy. The major carriers have fairly accurate data in their "pre-qualification" database based on actual line tests. Somehow in the NTIA process this was lost and instead they worked with aggregated data known to have serious errors. 

    The giveway was how many areas on the map showed 100% coverage. There are almost no areas in the U.S. where 100% can get broadband. For example, Verizon couldn't offer DSL to about 10% of New York and AT&T had to say no to about 10% of Chicago. New York and Chicago are older, dense cities where nearly all homes are within DSL reach of a central office. If you make a DSL map based on distance/block data, nearly everyone would be served. It turns out a meaningful number of apartments in both cities are served by remote terminals that block DSL unless upgraded. The same is true in many suburban and rural pockets developed between 1975 and 1995. Other technical issues (pairgain equipment, bad copper) similarly prevent serving selected homes.

     Similarly, the cablecos do not offer service in the parts of their territory that have fewer homes. Residences within business and industrial areas also often aren't served. There's no way to accurately know the overlap without the data by address. 

     This was an obvious problem in the early maps such as Connect Kentucky. CK told me the prequal data wasn't available for their early efforts but they would get it from the carriers from then on. Because they clearly understood the problem and said it was solved I backed away from my criticism of CK's mapping.  Every expert in broadband mapping knows the issue well. 

    When you buy a computer from Dell or Best Buy, the seller's computer checks with the carriers prequal to see whether they can earn a commission by selling broadband as well. It's awkward and expensive to find out later that a customer sold service can't be reached, so the carriers make sure to have that data available. I'm guessing how many homes are involved. It's shared daily with thousands of resellers, but they've never provided an aggregate number of how many homes are involved.