Think the NBN is slow? Well, it isn’t says NBN chief, the data just isn’t measured right

If you’ve ever been gaslit by someone you know it’s a pretty awful experience, being made to think that it’s you that’s crazy, and not some other external explanation for what’s going on. Well, that’s exactly what the NBN would have you believe this last week, coming out and saying no, Australia’s broadband speeds aren’t slow, you’re just measuring them wrong.

How so? Well, NBN Co chief executive Stephen Rue argued that Australia’s poor standings in world speed test rankings wasn’t because our broadband speeds are generally slower, but because the measurements aren’t done in a representative way.

Ausdroid is Advertising Supported

A variety of fixed broadband speed test results show that Australia’s ranking isn’t high, and it’s slipping. Tests from firms such as Ookla, M-Lab and Akamai show how each country fares for broadband, with Australia lagging behind.

According to the latest Ookla report for September, Australia ranks at 61 in the world for fixed broadband.

These results are – for obvious reasons – seized upon by those who would point out that the Multi-Technology-Mix policy of the coalition government has been an abject failure, but the NBN Co will have none of it.

NBN Co commissioned a research firm to analyse the data behind speed tests internationally, and to come up with a ranking methodology that allowed a more like-for-like comparison with other countries. This report was launched at last week’s Broadband World Forum in the Netherlands.

Broadly, the difference between all other reporting bodies and NBN’s research is that companies such as Ookla and M-Lab rely on users actually submitting speed tests. In other words, users are self selecting and may not be representative of the actual speeds available in the real world.

It’s easy to see how such a thing could skew results – if you know you’ve got a reliable 100 / 40 Mbps connection, you might be less inclined to speed test as regularly as someone who’s trying to diagnose issues with a crappy FTTN connection that never gets up to speed.

With this in mind, and using more representative data, NBN Co’s report contends that Australia’s ranking is closer to 17 out of 37 countries. Australia would still fall behind Singapore, South Korea, Japan, New Zealand and the US … but this really isn’t surprising. Fixed broadband in those countries is significantly faster than it is in Australia, and no fudging the numbers will change that.

Speaking of that fudging, it involves measuring contracted speed rather than speed actually delivered, so FTTN connections will skew those results a lot. Those results are also weighted against availability of broadband – which pegs down the countries where broadband isn’t so widely available.

What’s all this mean? Well, Australia’s broadband speeds are – on average – well short of where they should be and, if you take a particular view on the NBN MTM policy, well short of where we would’ve been had the government not decided to use a mixture of copper, crappy technology and just a little bit of nice fibre.

Chris Rowland: @ozcjr Chris has been at the forefront of smartphone reporting in Australia since smartphones were a thing, and has used mobile phones since they came with giant lead-acid batteries that were "transportable" and were carried in a shoulder bag. Today, Chris publishes one of Australia's most popular technology websites, Ausdroid. His interests include mobile (of course), as well as connected technology and how it can make all our lives easier.

View Comments (27)

  • Wait, so they're arguing that it's not slow, it's just that more people have slow/problem connections?

    • Amongst other things, they're arguing that we should assess contracted speeds, not speeds that are able to be delivered (which seems ridiculous) and that broadband results from other countries should be weighted by availability (i.e. if it's not widely available, the country should be ranked lower, etc). Those two things seem a bit arbitrary, but the one argument they've put forward is somewhat sensible.

      Those doing speedtests upon which most reports are based "self select" - i.e. they're not randomly chosen broadband connections, but those of people who want to have their connections measured. I think there's an argument that this could skew results significantly - it's not hard to see that there could be two distinct groups of people conducting speed tests: 1) those who have the fastest connections available and want to measure them, and 2) those who have problems and/or slow connections and are doing speed tests to try and diagnose issues.

      Regardless of group, if there's a lot of tests from users in either, the results are not going to be representative of anything useful.

      • There is a way to do the NBN speed testing, and make it accurate.
        Have every ISP test the speed of 5000 of their users, randomly selected each day. So every ISP gets a random 5000 of its users speed tested every day.
        With that scale of testing on a daily basis, it would show accurately how the NBN is overall, failing to perform to its advertised standard. With it being a random 5000 it would remove the self select bias you described.

  • Will the Labor blaiming ever stop!

    Prior to of the coalition's deal with the devil we would have all had fibre to the premises.
    But Rupert told Malcolm I don't want fibre to the premises, Foxtel comes before Australia's structure needs.

    • We all would have had FTTP (except for those places just too remote to connect), and while it might've taken longer, and been more expensive, imagine how future-proofed we'd be now.

    • When we stay in Barcelona for MWC each year, typically fibre is everywhere and their even their dirt cheap fibre plans provide 50/50 Mbps speeds for about 25-30 EUR a month. Ridiculous.

  • Omg - stop being a mouthpiece for NBNCO, do some research and do not give these idiots the platform to spread such utter garbage.

    • In case you didn't actually read our story, I think we've made it pretty clear that we don't agree with what the NBN Co has said. At all. They're manipulating the data to (ultimately) make themselves look less bad. Australia's broadband speeds are woeful by international standards, and no amount of manipulating data is going to make it look any other way. We don't have consumer grade broadband faster than 100 Mbps, and so many other countries do now. It's shameful.

      It's 2019, and broadband speeds in Australia have gone nowhere in the last 20 years.

  • Hold up, weren't Labor going to run FTTC? Which would have blown up costs, and made the tiered plans a necessity. Typical Liberal, take away the service and keep the cost high.

    I refuse to pay more than $60 a month for home Wi-Fi, so I've got a 12mbps plan. I just plug my phone into my laptop via usb c and get 20-25mbps all day. Then once the 60gb is used up, switch back to the slow $50B nbn. Funiest thing is, Telstra paid for the 25mbps phone reception, now they're rolling out 5g for us too...

    • Labor was building FTTP (premise). That's a future proof investment. It sits there and does whatever we ask of it for 50 years or more. Then the Liberals came along and said "Hey we can do that cheaper". Only problem is, it wasn't cheaper and it will have to be scrapped and replaced with a fully fibre network. The expression is "why pay twice?".

  • > What’s all this mean? Well, Australia’s broadband speeds are – on average – well short of where they should be

    Agreed. Facts are that less than 40% have FTTN (68Mbps average), HFC & FTTC should support 100Mbps and FTTP (~30%) should have access to 1Gbps.

    > if you take a particular view on the NBN MTM policy, well short of where we would’ve been had the government not decided to use a mixture of copper, crappy technology and just a little bit of nice fibre.

    Except that view would be wrong.
    - Under Labor's FTTP only policy ~85% were connecting at 25Mbps or slower.
    - LNP cut CVC pricing from Labor's $20/Mbps to $8/Mbps and bundled CVC with 50Mbps AVC, resulting in RSPs moving customers to 50Mbps plans.

    Replacing all the copper with fibre wouldn't change the speed issues caused by Labor's speed tiers.
    If NBNCo removed speed tiers from FTTN it would have the fastest average speed of all connections.

    • The difference is that the CVC pricing can be changed at the stroke of a pen - it's a simple commercial decision. But remediating all those stuck on underperforming FTTN connections will cost millions and take years. And I would love to know the rationale behind your assertion that removing the speed tiers would make FTTN have the fastest average speed - given that a significant percentage of FTTN connections wouldn't support faster speeds even if they were more affordable?

  • Chow chew by. Tulan. In sg I get 1 tgbps. Here getting 8mbps when I pay for 49mbps evening speed. Co is Tangerine

  • Wouldn't it be better to get ISPs to get users to opt-in to a daily speed test that occurs at different times of the day and it happens without user input. If it was fully automated and could be scheduled to collect consistent data.. that would be much better representative