Among the more alarming claims about the troubled National Broadband Network is an assertion that Kenya now has faster internet than Australia.
When this was put to Prime Minister Malcolm Turnbull in an October 6 radio interview he replied: "Complete rubbish. You know, like, one-and-a-half per cent of people in Kenya have access to broadband. In Australia, it's 90 per cent."
Mr Turnbull went on to say that the comparison with Australia was "not real".
"You might have a handful of wealthy people with apartment buildings that have got first-world telecoms, in a country where the vast majority of people have got no access at all," he said.
Is the comparison Mr Turnbull is making a valid one when comparing Australian and Kenyan broadband? And what are the facts around both countries' broadband penetration? RMIT ABC Fact Check investigates.
Mr Turnbull's claim is not the full story.
Only about 1.75 per cent of Kenyans have access to fixed broadband. However, as is the case elsewhere in Africa, millions of Kenyans rely on mobile services for their internet, through either 2G, 3G or 4G subscriptions.
Although there is some disagreement among experts, by some definitions 3G and 4G networks are seen as fast enough to qualify as "broadband", and are marketed as such.
Data compiled by Kenya's Communications Authority shows more than one-third (34.2 per cent) of Kenyans have access to broadband, after factoring in 15.4 million mobile subscriptions.
Alternative data supplied toFact Check by the NBN suggests the figure is about 10 million.
Either way, it is clear that a far greater proportion than 1.5 per cent of Kenyans have access to speeds that might qualify as "broadband" when mobile subscriptions are included.
However, it is unlikely that the millions of Kenyans using mobile broadband services would get faster speeds than the majority of Australians.
In addition, using the most credible measure for comparing fixed broadband speeds, Australia remains faster than Kenya, despite some reports to the contrary.
Australia also has slightly faster mobile broadband than Kenya.
These factors put Australia ahead of Kenya, but Mr Turnbull is not entitled to suggest that the scale of Australia's advantage is in the order of 90:1.5.
Nor can he say that all but a handful of wealthy people in Kenya have no access to first-world telecoms.
Assessing the claim
Mr Turnbull claims that comparing internet speeds between Australia and Kenya is "not real" because only a small proportion of Kenyans have access to broadband, while the "vast majority" have no access.
There are two variables that need to be examined to assess the claim. First, how Australia compares to Kenya on speeds and second, how the two countries compare in terms of access.
The task is complicated by a critical difference between internet services in Australia and Kenya.
In Australia, internet access is predominantly via fixed networks such as DSL, cable and fibre, but also via mobile networks and satellite.
In Kenya, the most common access is via mobile networks, while fixed broadband is available to some.
Mr Turnbull used the phrase "access to broadband". He did not mention either fixed or mobile internet access.
Radio host Neil Mitchell asked about "internet speed" and said in Kenya "they're still doing it a lot faster than we are".
Listeners were entitled to take Mr Turnbull to be talking about broadband in general.
Fact Check considers that both fixed and mobile access should be taken into account in assessing the claim.
Comparing internet speeds
The notion that Australia has fallen behind Kenya on broadband speeds is based on the quarterly State of the Internet report, released by content delivery business Akamai.
Fact Check has previously considered in detail the reliability of the Akamai report.
In June last year, Fact Check scrutinised a claim by Labor's then communications spokesman Jason Clare that also relied on the Akamai data.
Mr Clare said Australia had slipped from 30th to 60th in the world on internet speeds since the Coalition was elected.
Experts contacted by Fact Check at the time said the data contained in the Akamai report was a credible and widely used indicator of internet speeds. Fact Check concluded that Mr Clare's claim was "correct".
The State of the Internet report provides two main measures of internet performance: "average connection speed" and "average peak connection speed".
As set out in the fact check of Mr Clare's claim, Akamai's Senior Director of Industry & Data Intelligence, David Belson, explained in a February 2015 blog that "average peak connection speed" was a better measure when comparing countries, "more representative of internet connection capacity".
Experts likewise advised Fact Check to use average peak connection speeds.
Akamai's country comparisons for average peak connection speeds are for fixed broadband connections. As the Akamai report explains, traffic from known mobile networks is, where possible, removed from the data and analysed separately.
How does Australia stack up against Kenya on internet speeds?
While Australia's speeds have steadily improved, our international ranking has slipped further since Fact Check last examined the issue because other countries' speeds have improved at a faster rate.
The latest Akamai report, covering the first quarter of 2017, ranked Australia 64th using Akamai's preferred peak connection speed measure, with an average of 55.7 megabits per second.
On the same measure, Kenya ranked 101st, with an average peak speed of 38.5 megabits per second. That suggests that Australia is performing better than Kenya.
However, Kenya is performing better than Australia using the "average connection speed" measure. Australia ranks 50th, with an average of 11.1 megabits per second, while Kenya ranks 43rd, with an average of 12.2 megabits per second.
The relative performance of the two countries using both measures is shown in the two graphs below.
Akamai's separate analysis of mobile connection speeds shows that in the first quarter of 2017 Kenya achieved an average mobile connection speed of 13.7 megabits per second, only slightly behind Australia's average of 15.7 megabits per second.
What about access?
In a blog published a day before Mr Turnbull's 3AW radio interview, the NBN's Chief Network Engineering Officer, Peter Ryan, also challenged the idea that Kenya has faster internet than Australia.
Like Mr Turnbull, he did this by arguing that only a small proportion of Kenyans have access to fixed broadband services, compared with 90 per cent of Ausralians.
"Is it actually true? Does Kenya — a country with a GDP per capita of $US1,455 per year — compared to Australia's $US49,900 — really have faster broadband than Australia?" Mr Ryan wrote.
"The answer, to put it bluntly, is no — unless you happen to live in one of the 180,000 lucky, perhaps wealthier residencies receiving Fibre-to-the-Premises (FTTP) or Hybrid-Fibre-Coaxial (HFC) services. For the other near 9 million Kenyan homes there is no fixed-line broadband."
He adds: "Kenya has a total fixed-broadband penetration rate of just 1.75 per cent. To be quite clear, that means 98 per cent of Kenya's households — that's around 8.8 million premises — don't even have a fixed-broadband connection. I wouldn't like to be telling those people how great their internet was compared to Australia."
At face value, Mr Ryan's argument would appear to roughly mirror Mr Turnbull's, albeit with more detail.
There is, however, an important qualification contained in Mr Ryan's blog that was not used by Mr Turnbull — Mr Ryan uses the phase "fixed-broadband", whereas Mr Turnbull does not.
Both Mr Turnbull and Mr Ryan fail to mention the millions of internet users in Kenya access broadband through mobile services.
Access in Australia
Internet access in Australia is also improving. According to the Australian Bureau of Statistics, there were 7.7 million households with internet access in 2014-15, representing 86 per cent of the total, up from 83 per cent in 2012-13.
These figures, which refer to the internet generally and not to broadband specifically, are the most recent available.
During the 2016 Census, respondents in 6.8 million occupied private dwellings, representing 83 per cent of the total surveyed, said that they had internet access.
And the Australian Communications and Media Authority's (ACMA) Communications report 2015-16 estimated that approximately 15.8 million Australians (85 per cent) had an internet connection in the home.
The report said that "85 per cent of Australians also had a home broadband connection ... [which] reflects that almost all home internet connections are now broadband".
Other figures from the ABS show there were 13.7 million broadband subscribers in June 2017, including 6.1 million "mobile wireless" broadband subscribers and 7.4 million subscribers getting broadband either by DSL, cable or fibre connections.
The ABS defines broadband as an "always on" internet connection with an access speed of 256kbps or higher. The ACMA report defines broadband as "high-speed internet access that is always on and faster than traditional dial-up access".
Fact Check asked the NBN for the source of Mr Ryan's claim that "Australia has a total fixed-broadband penetration rate of around 90 per cent".
The NBN's corporate media executive manager, Tony Brown, said the data was sourced from the Informa World Broadband Information Service. The data is not publicly available.
"This does not include mobile broadband and includes only fixed and fixed-wireless connections," Mr Brown said.
The term "fixed-broadband penetration" refers to the proportion of premises with "active" broadband services.
Access in Kenya
In 2013, the Kenyan government released a National Broadband Strategy, aiming to "transform Kenya to a knowledge-based society driven by a high-capacity nationwide broadband network".
According the latest quarterly statistics report published by Communications Authority of Kenya, at the end of the June quarter 2017 there were 15.4 million broadband subscriptions, up from 13.7 million the previous quarter.
That represents a broadband penetration level of 34.2 per cent, up from 30.4 per cent the previous quarter. The growth of broadband in Kenya is shown in the following graphs.
Dr Henry Lancaster, an expert on African telecommunications at research and consulting business BuddeComm, wrote in a recent report that the broadband market in Kenya had been evolving rapidly, with most broadband subscribers in Kenya remaining on mobile networks, with a range of services including video streaming, evolving quickly.
"Most broadband subscribers remain on mobile networks. These are being migrated from the dominant 3G segment to LTE [4G] as this technology is built out by operators. A range of services including video streaming, e-commerce, e-learning and e-government are evolving rapidly on the back of this improved infrastructure," Dr Lancaster said.
Dr Lancaster told Fact Check it was true that fixed-line broadband penetration was very low in Kenya, and that broadband speeds comparable to those in Australia were "limited to a few providers concentrating their efforts in certain urban areas".
"In common with most countries on the continent, the overwhelming majority of all internet connections are via the mobile networks," Dr Lancaster said.
"3G still dominates in Kenya, though the MNOs (mobile network operators) are trialling LTE [4G]."
The NBN's Mr Brown said the figure of 1.75 per cent used in relation to Kenya was also taken from the Informa World Broadband Information Service database. He confirmed that this only related to fixed broadband services.
"Kenya has around 185,000 premises subscribing to fixed-broadband — from a total household count of around 10.5 million households — with the bulk of these on either cable or FTTP (Fibre to the Premises)," Mr Brown said.
"As with all African markets, the vast majority of end users are served by mobile broadband rather than fixed broadband services."
"Kenya has mobile ... penetration of 80 per cent. This is based on population rather than households, with some 38 million mobile subscriptions, dwarfing the rather meagre 180,000 premises connected to fixed-broadband.
"However, 28 million of those are on 2G connection [not broadband], with nearly 10 million subscriptions on either 3G or 4G connections, which are broadband."
What the experts say
Mark Gregory, an Associate Professor at RMIT's School of Engineering, said technically Mr Turnbull was wrong.
"If he hasn't qualified it by saying 'fixed' then technically he is wrong," Dr Gregory said.
"No excuses there, he keeps on claiming he is a tech PM."
Associate Professor Gregory said the NBN's Peter Ryan had been much more careful with his language by referring to "fixed" broadband, although his failure to mention Kenya's reliance on mobile broadband was misleading.
"To get a true position of the comparison between the two countries, you need to talk about mobile," Dr Gregory said.
He added that the Kenya versus Australia debate was not helpful: "The countries that we need to be looking at, which are our competitors … we are falling behind them."
However, Rod Tucker, a Laureate Professor at the University of Melbourne and a leading expert on broadband, said he was prepared to give Mr Turnbull some latitude for failing to use the word "fixed" in his comments.
"Sure, with 3G and 4G mobile, it is possible to access the Internet," Professor Tucker said.
"But mobile services are not necessarily 'broadband'. 'Broadband' is a word that is normally applied to fixed services. I'd be willing to give the PM some latitude on this one."
Professor Tucker said Mr Turnbull's comment regarding Kenya was "basically correct".
"The Akamai data that Neil Mitchell was referring to applies only to fixed-line broadband connections and it is true that only about 2 per cent of customers in Kenya have fixed-line connections. The remainder use mobile. What the data says is that those two per cent of customers in Kenya who have fixed-line broadband connections have slightly better average download speeds than what 90 per cent of customers in Australia have. So it is not really an apples-to-apples comparison."
- Malcolm Turnbull, radio interview, 3AW, October 6, 2017
- Angus Whitley, Australia Has Slower Internet Than Kenya, Russia and Hungary, Bloomberg Technology, October 4, 2017
- Akamai, State of the Internet, Q1, 2017
- David Belson, Akamai State of the Internet Metrics: What Do They Mean?, Akamai blog, February 5, 2015
- Peter Ryan, Does Kenya really have better internet than Australia?, NBN Co blog, October 5, 2017
- Government of the Republic of Kenya, National Broadband Strategy, 2013
- Communications Authority of Kenya, Fourth Quarter Sector Statistics Report for the Financial Year 2016/2017
- Australian Bureau of Statistics, Internet Activity, June 2017, catalogue 8153.0
- Australian Bureau of Statistics, Household Use of Information Technology, 2014-15, catalogue 8146.0
- Australian Bureau of Statistics, 2016 Census Datapacks, Australia, G37 - Dwelling internet connection by dwelling structure
- Australian Communications and Media Authority, Communications report 2015-16
- Henry Lancaster, Kenya - Telecoms, Mobile, Broadband and Digital Media, BuddeComm, April 2017
- Productivity Commission, Telecommunications Universal Service Obligation, April 2017