Why you subsidize Google's Soviet-style Net
Inefficiency creates hidden costs for punters and taxpayers
Google certainly likes to throw a party. Its annual, roving Zeitgeist conference sees politicians and policy-makers beating a path to the door for inspiration. As a result, Google and policy makers have got very, very cosy together. But how much are ordinary net users and taxpayers subsidising Google?
The answer is a lot, and it's certainly getting more from us than it's paying for.
One estimate I've seen comes via analyst Scott Cleland, who certainly has a horse in this race: one of his jobs is chairman of NetCompetition, the telcos' lobby group. But let's use this as a starting point, and then go on to see why it's important in ways the business pages of the kebab wraps haven't clocked yet.
According to Cleland:
Google’s search bots regularly copy every page on the Internet, some as frequently as every few seconds, and Google’s YouTube streams almost half of all video streamed on the Internet. The study estimated Google’s payment to fund just the U.S. consumer broadband Internet segment to be approximately $344 million in 2008, or 0.8% of U.S. consumers’ flat-rate monthly Internet access costs of $44.0 billion. Thus Google’s 16.5% share of all 2008 U.S. consumer bandwidth usage, is ~21 times greater than Google’s 0.8% share of U.S. consumer bandwidth costs – or an implicit ~$6.9 billion subsidy of Google by U.S. consumers.
With traffic trends, that invisible subsidy may have doubled by now.
In effect, you're paying for inclusion in Google, because the bots use a lot of your bandwidth, although the cost is hidden. Is this fair? To stretch the highway analogy, vehicles that use the road more (heavy trucks) pay more than light users (scooters).
That's worth bearing in mind, as Google seeks taxpayer subsidies in the US for its 1Gbps fibre network (fiber if you're in the US) - far faster than anyone needs today.
Now the telcos' own behaviour - pocketing billions in subsidies - has been less than shiny. But two wrongs don't make a right. If we're to discuss what "shape" the internet will be in the future - with each shape having associated costs and technical advantages and weaknesses - we should do so on the basis of honesty. There are no "goodies" and "baddies" in this story, only sets of hidden costs that will be sprung on us. There's no such thing as a free lunch.
Google's preferred "shape" for the network of networks we call the Internet certainly isn't the best or most efficient, technically.
Imagine the content of UK's FilmFlex cached at the cabinet at the end of every street. Video on demand services would then be instant. The network would save money, and if we did what decent consumers should do and insist the savings are passed on to us as consumers, we'd be quids in. The ISPs could do it themselves, or they could use a CDN, such as Akamai, which has its servers inside the ISP's POPs (Points of Presence). Not as efficient as a streetside cabinet, but not far off.
By contrast, Google's technical architecture is incredibly inefficient. Huge data centres [sample plan], which the company has described as "our Manhattan Project", are built far from consumers. They must transmit data thousands of miles, which naturally requires high bandwidth, and for this bandwidth we now see, Google wants a subsidy. Google alreadygets subsidies from states and municipalities keen to have one of these honking great data centres built in their neighbourhood: running into many millions of dollars per facility.
It would be a peculiar way to run a railroad company to insist passengers went thousands of miles on horseback first, and then drove round at high speed round a little loop, before continuing their journey on horseback again. It may be how a Soviet-style planner might like it, but it's no way to design the internet.
Neither Google's vast data centres nor gigabit speeds are needed for us to enjoy a well designed internet, with lower costs. But both happen to be perfectly aligned with Google's heavily subsidised ideal.
Just fancy that!
It's left as an exercise for the reader to judge whether Google settled on a technical architecture optimized for taxpayer subsidies, or whether the subsidies are optimized for the peculiarly inefficient architecture. Or both - you tell me.®
Although Google now gains more in revenue than ITV, it pays no UK tax. It's entitled to do so under EU Law, so you'll have to blame the single market for that, not Google.
Sponsored: Hyper-scale data management