Friday 26 August 2016

Google fiber : technicolour bunny no longer in the pink

Google Fiber has been an iconic presence in debates over superfast broadband and FTTH. Initially announced in 2010, the Google Fiber bunny lolloped into its first market, Kansas City, in 2012. It has since expanded to five other cities in the US, offering symmetric 1 Gbps service at low prices.

"No new internet product has generated as much excitement in the technology world as Google Fiber", said the FTTH Council in 2014. Indeed, for fiber enthusiasts, Google Fiber has been emblematic of what should be possible. Here was a preeminent technology company, backing fiber all the way to the home. The fusty old telcos and cablecos simply didn't get it.

What a difference a couple of years make.

The Wall Street Journal recently reported [$] that Google is "rethinking its high-speed internet business after initial rollouts proved more expensive and time consuming than anticipated". Today come reports that Google Fiber is laying off half its staff. The company is also changing its strategy, with a shift to using wireless technologies to connect homes, rather than fiber.

I suspect we'll hear rather less about Google Fiber from the likes of the FTTH Council. But actually there are some valuable lessons here:

  • It's way too glib to say FTTH is 'future proof'. New technologies (such as the milimeter wireless Google are now moving to) are proving quicker and cheaper to deploy, while delivering similar performance to FTTH
  • A further technology problem for FTTH is that cable can now deliver speeds of 1 Gbps and more. This means that there's less reason for consumers to switch to a new entrant (where cable's available), and also puts the new entrant at a significant cost disadvantge. Upgrading the cable plant is far cheaper than deploying new fiber. Cable operators in Google Fiber cities have done exactly this, almost certainly reducing Google's uptake
  • It's easy to underestimate the 'friction' involved in doing any work that touches the last few meters to people's houses. Google has been accused of causing flooding in Austin, for example. Less dramatically, issues of home access, damage to gardens and so on all drive up costs
There's no reason to scorn the Google Fiber bunny. Google is a company that proceeds by experimentation, and not all experiments work. In the light of their experience, and of developing technology, they are revising their strategy. Utterly sensible - and a useful case study for those who continue to believe that it must be FTTH or nothing.

Monday 11 July 2016

What do apps require? Shall we ask Marketing or Legal?

The new net neutrality guidelines currently being considered by BEREC (the European regulators group) propose that ISPs be required to publish:
"an explanation of speeds, examples of popular applications that can be used with a sufficient quality, and an explanation of how such applications are influenced by the limitations of the provided [service]"
In other words, they're going to need to make a formal disclosure of which applications are possible with which bandwidths.

Similar documents are already provided by a number of ISPs in the US, and they make interesting reading, not least because they often have a rather different emphasis than the language used in marketing materials. Here for instance is the legal disclosure of  United Communications (a Tennessee telco). They say - accurately - that their 25 Mbps ADSL service can support email, browsing, social media, HD video streaming, video calling and multiplayer online gaming.

For all their fibre products (up to 1 Gbps) they simply say "All the applications listed above". That is, no matter how fast the fibre is, it doesn't actually enable any applications not possible over ADSL.

Contrast this to the same company's marketing,which says:
"United’s fiber internet lets you stream movies and TV shows with no stutters or slowdowns... it gives you the power to handle the next generation of awesomeness: 'massively multiplayer online games'"
The Marketing Department's language isn't false. Fibre does let you do those things. But the Legal Department's filing shows it is incomplete. You don't need fibre to do those things.

Tuesday 24 May 2016

NBN and the missing superfast customers


Australia’s national broadband network (NBN) is the Kim Kardashian of government broadband plans – bigger, bolder and more exposed than any other. (Though to be fair, it has left ‘breaking the internet’ to Telstra)

The NBN represents a renationalisation of the access network, as well as a significant upgrade to fibre - initially FTTP, now mixed technology. Whether or not government ownership is a good idea, it does at least bring a degree of openness on operational detail. This is very handy for broadband watchers around the world.

For instance, NBN publishes the mix of speeds its customers take, which are as follows:


Source: Various NBN co publications
Note: Excludes plans above 100/40, with trivial take-up

The most popular speed tier is 25 Mbps down, 5 Mbps up. At the end of March, 46% of customers were on this plan. Certainly such speeds would be beyond an ADSL connection, so NBN has provided some benefit.

However, a moderate uplift vs ADSL was never the goal for the NBN. When it was announced in 2009, its primary ambition was to offer “up to 100 Megabits per second”. In this regard, consumers seem to have been less impressed. Only 15% of those on FTTP are taking the 100/40 product, a percentage that has been steadily falling since 2012. Despite the fact that the premium for moving from 25/5 to 100/40 is just £10/US$14, the great majority of consumers simply aren’t willing to pay the extra. (The premium from 12/1 to 25/5 is half this).

Based on the European 30 Mbps threshold, just 19% of NBN FTTP customers (or 155,000) are on superfast. The original NBN plan expected roughly double this percentage (and a far higher absolute number).

One consequence of this change in speed mix is that the average capacity  of an NBN line is actually falling over time – from 36 to 34 Mbps between June 2014 and December 2015, for example. However, traffic per line has grown appreciably in the same period – yet more evidence that traffic and bandwidth growth are two very different things.

Source: Author's analysis of data from various NBN co publications


Oh, and demand for speeds beyond 100 Mbps, striving into a gigabit future? Just 65 customers have taken them – 40 on 250/100, 3 on 500/200 and 22 on 1000/400. Clearly Australians are not bitten by the gigabit bug.

So what does this mean for policy makers? I’d suggest:

  • Don’t overestimate willingness-to-pay for higher speeds
  • Leave room for price differentiation of fibre products – lots of customers are happy at the low end, and you need to enable appropriately low prices for them
  • Report equivalent data, so the debate about demand for bandwidth can be anchored in reality


(Note that all the above figures are unconnected with the change of the NBN plan to a mixed technology model. They are for customers who are on the FTTP portion of the network).