It was fantastic to hear Mark announce his ambition to accelerate universal access to the internet. I strongly agree with this goal and wanted to suggest some admittedly crazy ideas for the effort. We’ve all got to dream very big to help realize this goal and hope others will share their ideas.
As context, there is a huge disparity in Internet access between developed and emerging countries. In the U.S., 81% of the population has Internet access, and in Europe, it’s 75%. However in Asia, where there are more than 4 billion people, only 32% have Internet access, and Africa, where there are a billion people, only 16% have access. There is no country in the world that provides its citizens with universal free access, which I strongly believe would increase quality of life for all residents and skyrocket a country’s GDP over a relatively short time (back of envelop estimate would be within two generations).
How we provide access is the big question. The recent formation of Internet.org by Facebook, Ericsson, MediaTek, Nokia, Opera, Qualcomm and Samsung is a good start, and they laid out some lofty goals: decreasing the cost, increasing network efficiency and developing new business models. But they were short on specifics, because this is a very tough challenge within a complex ecosystem. These aren’t the only players making an effort – Google, which is not part of the coalition, is working on several different initiatives to provide free access, including unorthodox ideas like Project Loon, and there are other global initiatives such as those being spearheaded by The Internet Society.
Here are my top ideas for addressing each area internet.org has enumerated as key levers to universal internet access:
Fund free access by charging carriers royalties for the most popular apps. Right now, people buy devices and service, to get access to great apps. Amazingly, the carriers pay nothing for apps (and in fact sometimes get paid!), despite the fact that without apps, there is limited value in paying for internet access. Compare this to the basic cable TV model in the U.S., where users buy the hardware and service, but the cable carrier pays for access to the channels (whose parent media companies also make money off of advertising without sharing it to the cable companies).
Let’s extend the Cable TV model to the mobile app ecosystem. If any of the massive top apps were to pull out of the app store and require the carriers to license their apps to provide access for their subscribers, we would see a massive shift in value capture. Concretely, imagine if each carrier had to pay something on the order of a $1-3 per user per month to preload apps like Facebook, Google Maps, Instagram and Gmail. Of course the software companies would be tempted to pocket the money, but I think their long-term interests could bend their short-term incentives to ensure they used the funds to enable affordable universal access. Worse case scenario is they could be taxed for profiting off of national resource of wireless spectrum. Think about the economics: Facebook and Google each currently have about one billion mobile users. At $1 a month per user, that’s at least $24 billion dollars a year for subsidized access right there.
Why would the carriers agree to pay? Imagine if you could not use the Facebook apps on AT&T or the Google apps on Verizon but could use both on Sprint. Users would still be able to use mobile web versions (which, by the way, are the most popular versions), but the carriers would have little choice if any major mobile app company implemented this strategy. Once one did, all of them would be able to do it because apps & OTT services are now driving customer subscription revenue and where differentiation (or lack thereof) is most impactful.
We could improve utilization of existing and future network infrastructure by changing the way data moves across the network. Today, the primary data transmission protocol (TCP/IP) finds a path on the network, then push bits through that path as fast as it can until the transmission is completed. Often the path that was found gets overused, or congested in network terms, and thus the transmission has to be slowed down. And just like highway traffic patterns, there are times of day and certain geographic corridors where everything crawls along. This is a big reason why it’s still hard to send huge files (on the relatively small order of a terabyte) in a timely way. Network congestion makes it impractical.
Continuing with the highway analogy, one of the more expensive aspects of managing highway congestion is adding new lanes and building new highways. Unfortunately, data network infrastructure has similar cost characteristics. So one way to visualize the goal of getting everyone in the world on the Internet is to imagine giving everyone in the world a car & then telling them to always drive their car while they are awake. DISASTEROUS!
So this idea for free Internet access would take advantage of the fact that roughly half of the world is asleep while the other half is not. Thus there are zones of times where you can assume the vast majority of people in an area are offline and networks built to support them are underutilized. The way the system works now, if you want to send a terabyte of data over the Internet from Japan to the U.S., it likely travels the direct route West-to-East via trans-Pacific cables regardless of time of day. If those cables are congested with other data, the transmission slows. But if we were able to send it via the most efficient and least congested path, that data might flow East-to-West at certain times, completing the overall transmission much faster.
If we rewrote the transmission protocols so that we used the most efficient path by throughput to move data, two things could happen: quicker data movement around the globe, and lower aggregate transfer cost per megabyte. We would be increasing network utilization while decreasing cost. This would allow us to offer Internet service at a much lower aggregate capex cost, which would lower price to the point were access could be free for people depending on time of day or what resources they are accessing.
We could give people free Internet access in exchange for using their mobile devices for data collection. Since mobile devices are chock-full of sensors we could create data sets that have never existed before. What if we gave away free Internet-enabled cellphones in return for real-time weather information collected by sensors on the phones? Yes, this is possible! I suspect this would quickly displace traditional expensive data gathering techniques and also would have very nice network effects that could also decrease backhaul costs.
An analog that comes to mind is electronic toll collection devices. In addition to their primary purpose – collecting tolls – those devices transmit real-time traffic data that is aggregated and reported out as average traffic speed and freeway congestion. I bet that could be a billion dollar idea if it was done on mobile handsets via a clever consumer app. Oh, shoot, that specific idea has already been done…though I suspect it would probably would work again.
Those are my current favorite ideas to spread free Internet access more quickly around the world. If you can run with any of these or have another idea to make this happen, you will likely be remembered for eternity. I’m excited to see it happen and would be thrilled to help make it so if opportunity come arises.
Bubba, I love the highway analogy you use when discussing efficiency. I have used the same analogy but with a different emphasis: the best way to avoid traffic congestion is to leave your car at home in the garage. The implication for the Internet: make client devices smarter and have them cache more data so they don’t have to request so much data from the cloud. As Ilya Grigorik at Google puts it, “The fastest request is a request not made.”
Great post. On section #3, and I know you can’t answer this (!!!), but I’ve always felt FB has a huge opportunity to help the carriers get out free plans and phones to the under-phoned in exchange for data collection.