Broadband in America – We’re doing it wrong

For the last six months or so, the focus has been on Netflix and net neutrality. Should ISPs be allowed to rig their high speed networks, through various means like traffic shaping, peering and interconnection decisions, etc., to purposefully disadvantage or advantage an internet-based service or experience. The majority of the users on the internet seem to think that no, they shouldn’t. But corporations and corporate apologists seem to think that without giving high-speed internet monopolies or duopolies the freedom to do whatever they want will mean an end to broadband investment, and a loss of competitiveness as a whole.

First its important to note why regulation is so heavy-handed in this market. Its all about competition, or lack there of. The barriers to entry for building out a wireline or wireless high speed data infrastructure are immense. And those high barriers to entry mean there is a lack of competition. And that lack of competition means that the competitors who are established in the existing market environment need to be regulated to ensure that the consumer is protected, especially with a product especially as important as internet access.

The current broadband situation in America isn’t great, its not even good. We have more than one problem, even if the Netflix/Comcast issue is the most salient issue right now. Our issues are…

  • Speed: our speeds are slower than most other developed countries around the world, and even some developing countries have faster, cheaper internet in their cities than we do in the US
  • Usage Caps: most large ISPs that offer fast speeds have or will soon have data caps to prevent users from downloading a lot of data (read: protect their expensive TV packages from over-the-top competition)
  • Open access: companies like Comcast and Verizon have decided not to increase their network interconnect capacity with certain companies to keep Netflix traffic slow. Level3, a tier 1 ISP, even illustrated the habit of ISPs creating congestion to incent data providers to enter paid-peering arrangements.
  • Strategy Taxes/Internal conflicts-of-interest: companies that provide internet access have their own vested interests (television, phone) so they have internal conflicts of interest to providing an open, unfettered, fast internet experience
  • Lack of competition: it costs a lot of money to build a broadband network from scratch, so there is not a lot of incentive to enter a market to create an additional competitor, especially if a regional or nationwide competitor can create a price war locally until you run out of cash and go bankrupt.

How can we solve all five of these problems? New rules for Net Neutrality, or regulating ISPs as Common Carriers under Title II don’t solve all the problems. We’re still left with the monopoly/duopoly with companies we have now. They’ll begin work immediately to lobby for loopholes in the rules or legislation, while working to undermine the enforcement through lawsuits, and finally by getting people friendly to them in the regulator’s chair to keep the rules from being enforced.

It turns out our approach to internet access its entirely wrong-headed. Infrastructure should only be built once and from there, upgraded over time to meet demand. Otherwise we’re just wasting money. Let the market compete on top of the infrastructure, not by building separate infrastructure networks and selling proprietary access to that network.

But we can’t just throw it all out and start over.

Infrastructure

The internet is most often referred to in analogy as a highway network. Billions of miles of fiber optic cable criss-cross the globe carrying your request to watch that funny cat video for the 100th time.

There is robust, healthy competition at the “Tier 1” level – that is, the worldwide backbones that carry traffic all over the world. But when it comes to your metropolitan area, that last few miles of connectivity are often dominated by one or two competitive ISPs ((I moved recently, and actually lost my phone company – when I moved in, Centurylink wasn’t able to provide DSL at my house, so my only option is Cox)), in the same way your local roads are maintained by your local municipal agency (whether its a city or county).

This infrastructure is expensive. Its expensive to build – which is why we have so few options in the first place. Its expensive to maintain – everyone keeps using more and more data, and it costs money to add capacity to the network. So why are we building redundant infrastructure in many places?

Does it make sense to build two power grids to provide some fig leaf of “choice” for consumers, while the fixed costs of building and maintaining two separate power grids are factored into everyone’s monthly bill? Does it make sense for your community to build two sets of roads for the same purpose ((By this, I mean having two driveways, two street addresses, two non-interconnecting local streets that both connect to the highway network)).

But with our current state of broadband, its not as straightforward as that since the cable and phone networks were originally built for other purposes, and were re-purposed for internet access. Phone lines are inherently more handicapped than their coax sibling – DSL speeds are pretty much always slower than cable. But phone lines are more ubiquitous, usually required by regulation. Cable companies can pick and choose who they serve – if you’re not sufficiently urban or suburban, I hope you like DSL and Dish/DirecTV!

Wireless-only options are a very distant third place in the broadband game, encumbered by very low monthly data transfer caps, high cost, and questionable signal strength ((Masayoshi Son’s fevered dream of using Sprint’s 2.5GHz spectrum to compete with wireline home broadband is laughable – its a simple matter of physics, even 120MHz of shared wireless communication at QAM64 isn’t able to compete with 200MHz of coax bandwidth operating up to QAM4096 (the DOCSIS 3.1 spec) to offer 10Gbit speeds to your home)).

So we are left with different kinds of monopolies in the two different situations – in exurban and rural areas, DSL is your only option because you are unserved by cable; elsewhere cable ISPs can offer much faster speeds than DSL can manage and its really a not much of a competition ((My Cable ISP’s two main tiers are now 50Mbps and 100Mbps, while the fastest the DSL company can go in most locations its 10Mbps)).

Even in the best case (fast cable and fiber-to-the-home from the phone company), we’re left with duopolies that have similar products (triple play bundle of TV, phone and Internet). This means that, in the case of Verizon and Comcast, they have no incentive to help Netflix or future over-the-top video providers eat away at their revenue for entertainment services. Which means we still lack a competitive environment, as long as the ISPs have something else to sell you that could easily be provided over the internet in general? ((Cable and phone companies have started to get into the home automation and alarm markets lately, does that mean that AT&T should be able to handicap their internet traffic from ADT and alarm.com to harm competition?))

Loop Unbundling

One option for solving the problems would be to force loop unbundling on the cable and phone companies, requiring them to lease their own lines at competitive rates ((Which probably would not be competitive, but what are you going to do? Sue them and their lawyer army in court? How profitable is that? Its just cheaper to put up with their uncompetitive rates.)). This would allow other companies to sell internet service over the hosting company’s coax or fiber-optic network. These companies would be responsible for their own transport out of the network to whatever Tier 1 ISPs they purchase transport from, and all of the technical support, billing and usage information, etc.

Loop unbundling works well in Europe, but their business environments are drastically different than ours in the US (PDF), and I fear that between the relentless lobbying budgets of the cable and phone industry, the bad-faith dealings and legal shenanigans that would occur against those who want to come in and resell transport, it would not be a profitable business. In the same way we see cable and phone companies working hard to prevent municipal broadband, we would see an even more rigorous offensive campaign against loop unbundling before, during and after its implementation.

Heavy Regulation

Another option would be heavy regulation of internet service provided by wireline companies. This regulation has been happening at the national level with the FCC, but maybe that’s the wrong place for it. The best option may be with the local and state agencies like Corporation Commissions or Public Utilities Commissions. In the same way electricity rates or phone companies are regulated by a PUC or Corp Comm., so would the internet access. The Department of Energy in DC doesn’t regulate your electricity rates, so why does the FCC want to regulate broadband with the goal of making it more affordable and ubiquitous? The local solution may be much more effective because local officials are much more responsive to citizen complaints. If everyone is complaining that Netflix is running slow on Comcast, or that Verizon is intentionally letting its copper-based infrastructure degrade, the PUC may be in a better position to force Verizon or Comcast to deal with it than a gridlocked FCC or Congress.

This would put some burden on the telecommunications companies because its more government they have to deal with. In my case with my ISP (Cox), they seem very receptive and integrated with the local community ((Even though they raise my cable and internet bills by 10% a year, 3-5x the rate of inflation)), so I don’t believe that for companies like Cox it would put an undue burden on them, but for large companies stuck in that unaccountable monopoly mindset like AT&T, Verizon and Comcast, it would be more difficult.

Municipal Broadband Networks

One option that has had mixed results so far is municipal broadband networks, where the local government agency owns and runs the ISP. Unfortunately, approximately 20 states have legislation that prevents or has halted the growth of these types of networks. The FCC is looking into creating regulation that would overrule states’ abilities to pass laws against them.

The existing companies have one legitimate issue with municipal broadband networks as competitors – these private companies have invested billions and billions of dollars in infrastructure and they don’t want the government to offer a subsidized competitor with the community’s tax dollars making up for the financial losses. However, that is not an excuse for the banning or restriction of municipal broadband networks. The community’s side is that broadband is (in their opinion) more expensive than it should be – see Chattanooga’s successful $70/mo Gigabit internet service – and nationwide ISPs are unresponsive or just don’t see the return on investment needed to build out or improve service. Ultimately, the benefits of municipal broadband networks is that, as taxpayer-owned agencies, they are ultimately responsible to the people they serve through the feedback look of their board of directors being locally elected officials. Short-circuiting the dysfunction in the marketplace and in Washington DC through local elections.

Municipal broadband networks aren’t always successful though, either through internal mismanagement or being undercut by a national competitor (who can sustain losses in one region) for a long enough period of time to lead to financial ruin for the publicly funded network.

Building a True 21st Century Broadband Network

The goal would be to build a network that combines the advantages of the methods described above, while trying to do this as cheaply as possible and without building a redundant network.

Eminent domain

Eminent domain is essentially creating the municipal broadband network while removing the threat of a private broadband company attempting to undercut and destroy the effort ((This isn’t socialism or communism — as long as the assets are purchased in the public interest (to provide faster internet at cheaper prices) for fair market value)). Customers can continue to use their existing ISP under a temporary management agreement until the necessary infrastructure upgrades can be made to support Loop Unbundling.

Network Migration & Loop Unbundling

Eventually, all devices on the network (Internet, TV, Phone) will need to be IP or have a set-top-box for incompatible devices. This would be the most difficult aspect – you would need either affordable devices that can authenticate themselves to the network and then broadcast the subscribed set of channels over the customer’s home coax, or you would need even cheaper boxes for people to hook up to individual TVs and output the broadcast signal over RCA, S-Video or HDMI to the TV.

Management

The new network would be managed by a regional “Broadband District” for a metropolitan area. That BD would issue bonds to purchase the physical infrastructure from the cable company or phone company (whoever has the best infrastructure) by force at market value under eminent domain. Once the network migration is complete, it would be opened up to anyone who wanted to run their own ISP.

Eventually the goal would be to use the local fiber loop purchased by the government agency to provide everyone fiber to the home, and then allow everyone to have a competitive list of broadband providers to choose from.

Successful Business Models

The unbundled nature of the network would provide for real competition, ranging from little value-add (here is your internet service, have a nice day) to high value-add (on-location tech support, home networking, security, phone, TV, whatever else they can think of) at various prices.

Companies like Google or Facebook could offer subsidized service that, in exchange for your browsing habits, phone records, TV watching habits, dignity, and whatever shreds of privacy you have left, give you a discount on service.

For economically disadvantaged areas, instead of providing fiber to the household or multi-tenant building, offer WiFi services and prepaid broadband cards for a fixed amount of data transfer (1GB, 5GB, 10GB) sharable amongst a number of devices. This way, even those with only a basic smartphone could still use the internet for essential tasks like applying for jobs or requesting benefits.

Conclusion

First, we need to remove redundant investment. We don’t have competing road networks, power grids, or water systems ((Even in the great free market of Texas, there are multiple energy companies, but only one energy transmission company – why? Because infrastructure is expensive!)). Why do we have these expensive, redundant last-mile communication networks? Because of some historical legacy? Time to ditch them and have one broadband infrastructure for all Americans.

Next, we need to figure out where competition works – clearly its not working in its current state as prices climb and the US lags in broadband penetration. Letting companies compete over a shared infrastructure will work – it works every day as companies compete with each other while driving over our shared roads, utilizing the power grid and water and sewer networks.

Finally, costs need to be managed as we convert the proprietary networks of today to the open networks of tomorrow. We can’t have gold-plated equipment blowing up the budget. Any agency willing to do this would need to be responsible and disciplined in its conversion to eventually opening the network up to other companies to compete on.

Apple WATCH battery math update

One the things I mentioned during my Apple WATCH post was the estimated battery capacities. Well, thanks to iFixit, the iPhone’s battery characteristics have been revealed and using the same type of batteries in the WATCH would stand to improve on my estimates somewhat. Here are the figures for the iPhone 6 and 6 Plus batteries

So what does this mean for the watch? That the battery capacity would be slightly higher than my original estimates of 1.9Wh or 520 mAh at 3.6V. Its more likely that the battery will end up around 600 mAh or 2.1Wh at 575Wh/L.

Tesla Model S AWD, Autopilot and Model 3…

Model S AWD

The Model S AWD models are impressive. We knew they were coming for a while now – there is a space in the frunk that would be perfect housing for the motor for the front wheels. The performance is the headline here – 691 HP (combined) and 3.2 seconds for the 0-60 time launches the Model S AWD into supercar territory. But its not a supercar that just sits in the garage and looks nice – its a great daily driver, highly efficient, very low total cost of ownership per mile relatively speaking, and very user friendly. Beyond that, the way the two electric motors are tuned to work together help improve the overall efficiency of the car, allowing it to, despite the added weight of a second motor, increase the electric range from 265 to 275 miles for the performance model and a whopping 30 mile increase from 265 to 295 for the non-performance 85kWh model (numbers provided are Tesla estimates, EPA numbers are presuming to be pending certification before deliveries in December).

Autopilot

The new Autopilot functionality of the Model S seems eminently more practical than the hyped up Google self-driving car of the last few years. The great news is that instead of having to wait until 2018 or 2020, we can get highway autopilot several years earlier than expected.

The sensors include sonar around the car (forming a bubble around the car) as well as a forward-looking radar and camera system. These allow for active safety systems – automatically braking the car if an obstacle in front is detected, preventing you from steering into a car in your blind spot, etc. These features have existed in cars for a few years now. Beyond the active safety systems is the new Autopilot software.

Autopilot is a very fitting name for this feature, as it mimics overall idea of autopilot on a airliner – the pilots control everything until the plane is at a comfortable cruising altitude and can be turned over to an automated system. Same with cars – if your commute is a long drive on a highway, once you’re on the highway, you can manage the car with just the turn signal. Cameras read the speed limit signs, slow for cars in front of you, and perform actions to keep you safe.

Though after watching the official Tesla video and reading the press release, I wasn’t quite sure what features are being delivered today and what will be made available in future over-the-air updates to the autopilot software. What’s important is that the hardware necessary for autopilot is being delivered today. Software improvements can come in time, but its prohibitively expensive to go back and retrofit this hardware on existing cars (Tesla has stated they won’t retrofit, so you’d have to buy a new Tesla and trade in your old one). As Tesla adds features to the autopilot software over time (the ability of the car to park itself in a garage without you in it), the car will evolve to the self-driving ideal, though it won’t make it all the way there.

The only negative is I don’t think there are enough sensors – that in the future, rear facing radar sensors or cameras will be added to the package to help the car switch lanes when there are high differences in the rate of speed between the two lanes. And making sure the sensors are redundant enough to withstand a failure.

Model 3

One of the interesthing things about the new AWD cars is that the smaller electric motors (188 and 221 HP) seem to be a perfect fit for a Model 3-sized car – one for the standard model, and one for a “performance” Model 3. Tesla should be able to re-use the motors with small adjustments in the firmware to optimize it for single-axle drive.

Beyond this, we’re able to get a better idea of the specs of the Model 3. One thing Elon has stated is that the Model 3 will be about 80% of the size and weight of the Model S. The Model S originally (2012) had a curb weight of 4,450 lbs. However recent statements have indicated they’ve taken “hundreds” of pounds out of the car, I’d estimate the current curb weight for the single-motor model is around 4,200 lbs. A 20% reduction would put the car around 3,400 lbs. A 188 HP motor should be able to propel the car with respectable (certainly not supercar) 0-60 MPH times. By comparison, my Chevy Volt has a curb weight of 3,700 lbs and only a 160 HP motor. While 188 HP might not sound like a lot, the fact that its electric and instant torque will compensate for the relatively small HP rating compared to gasoline engines.

The battery for the future Model S will end up around 45kWh using these smaller motors, reduced vehicle weight and improved efficiencies (an improvement from 300 Wh/mile in a RWD Model S to 225 Wh/mi for the base Model 3). This reduction in pack capacity, combined with the reduced costs of the pack through the Gigafactory increase the chances that Tesla will be able to hit the $35,000 price with a base 200-mile model. The conservative estimate for packs out of the Gigafactory is $196/kWh (down 30% from Tesla’s early 2014 baseline of around $280/kWh), and the aggressive estimate is around $168/kWh (down 40%), which would put the pack price between $7,600 and $8,800. This is 22-25% of the price tag of the overall vehicle, which should leave plenty of room for the rest of the car (50%, or $17,500) and a gross margin of 25%. A longer range 60kWh version could be made available with the beefier 221 HP motor for a range of just over 250 miles. The only issue with a battery pack that small is how fast (or slow) it can be supercharged.