If you’d googled “the Netflix effect” a few days ago all you’d have found was articles about binge viewing, or the sudden stardom of previously unknown actors.
That changed on Monday, when in this story the ABC tried to convince us that there is another “Netflix effect”, and according to them it’s the reason why the NBN is dudding up to one in twenty customers. Here’s the full article headline:
It’s rubbish, in my opinion.
This brand-new version of “the Netflix effect” as an explanation for NBN problems is an invention of the NBN, swallowed whole without question by the ABC, and now happily amplified by them in this story without any critical scrutiny whatsoever.
Even the catch-phrase “the Netflix effect” appears to be something invented by the author, or someone in NBN PR.
Netflix, the company, is called out eight times by name in the story, and its corporate logo is featured in the front page thumbnail, yet the author – an ABC journalist – hasn’t even approached them for information, comment, or an alternative explanation. In the story the Netflix service is being held up as the root cause of a serious problem in our national network but they’re not even approached for a response. Frankly it’s embarrassing.
I think the headline and lede should have been quite different:
NBN: over-promised and under-delivered again
How poor, short-sighted design and a rushed implementation have dudded the one in twenty NBN customers forced to use fixed-wireless technology
It’s a bit different, isn’t it? In this headline there’s no mention of Netflix whatsoever. No “dream killed” or “everything changed”. Just bad engineering, and politics over-riding even that, because I believe that’s the simple truth.
Here’s an even simpler, more brutal version:
Our world-class broadband network would perform brilliantly if people didn’t use it
Implicit in both of the ABC headlines is the idea that the network would be fine, even “super fast”, were it not for Netflix “killing the dream”. As with many other assumptions in the story, the ABC doesn’t actually check this, so we have to do it for them.
Is the NBN fast, let alone super fast?
Let’s check. Economically, Australia is a proud member of the G20, the globally ranked twenty largest economies. In fact, according to the IMF we are currently ranked 14 – well done us!
So where does Australia rank for broadband? Well, there are a number of measures, but they broadly agree. In 2009, the year the NBN was formed, Ookla ranked us 39th – already well outside the G20.
In 2016, seven years into the NBN, according to Akamai and the ABC we had dropped to 60th… Way to go, NBN.
It’s worth remembering too that 2016 was when the NBN was to be finished, according to the Government, although current projections have 2016 happening by 2020 at the latest.
In 2019, after ten years of NBN development, where has our “super fast” national broadband placed us? According to Ookla and Business Insider, we are now 62nd.
Number 14 in the G20, but number 62 in world broadband – ranked 50% lower than when the NBN started. So not super fast. Not even fast, and embarrassingly most of those other countries have Netflix as well, but it doesn’t seem to have killed their broadband dreams.
Maybe Netflix isn’t really the problem with the NBN fixed-wireless, despite what the ABC and NBN say.
So, what’s the full story? Well, this article has been hard to write; not because the facts are difficult, and not because the arguments are complex, but because I believe there are so many layers of wrongness in the quoted ABC story.
That story deserves review, but it turns out it’s impossible to separate the article from its subject matter – the NBN. So this article has to be a critique of both the ABC story about the NBN, and the NBN itself.
Here’s a really short summary of the things that I believe are wrong, in case you’re a tl;dr person. If you want reasons and detail, read further.
- The front page headline pictured at the head of this article is deliberately and highly misleading, and contains four separate errors alone
- The short-form story synopsis on the front page contains further misleading errors
- The full article headline is misleading and unbalanced, and also contains errors
- The actual article, while apparently criticising NBNCO, uncritically accepts and repeats specious arguments advanced by NBNCO regarding the cause of fixed-wireless bandwidth problems, and totally fails to investigate them or to put forward any other expert or independent opinion.
- The timing of the ABC story is, I believe, highly suspicious. It prominently and repeatedly names “Netflix” and “the Netflix effect” just six days after NBNCO floated the idea of a ‘Netflix tax’ to its Retail Service Providers – an obvious violation of Net Neutrality.
So, where to begin…
Let’s start in 2006, thirteen years ago and three years before the NBN started. Since then Telstra has been offering the majority of its customers broadband access of 8Mbps or better (Mbps is Megabits per second). The minimum speed tier on NBN fixed-wireless now is 12Mbps, and everywhere else it’s 24Mbps.
By contrast, the speed required for internet HD TV – the same high definition as broadcast HD TV – is only 5Mbps.
Likewise, since 2006 the download packages offered by ISPs have been getting larger and larger, and nowadays some ISPs only offer “unlimited” packages. The smallest package available nowadays is 200GB (Gigabytes) per month.
So assuming you were on an extremely modest 12Mbps / 200GB plan you could watch HD Netflix for three hours a day, always using less than half your maximum speed, and not exceeding your data allowance.
If you were on a more common 500GB plan you could watch for seven and a half hours per day using a fraction of the speed you’ve paid for and only the data allowance you’ve paid for, and of course on an unlimited plan you could watch as much as you liked because that’s what the NBN and your ISP offered and what you paid for.
So what’s the problem?
We didn’t see that coming
The problem is that although both the speeds and data volumes are modest according to available plans, when the “experts” at the NBN designed and built the fixed-wireless network they decided that nobody was going to actually use what was promised and is now being offered and paid for.
They didn’t design the fixed-wireless network to accommodate the bandwidth to support exactly the plans that are offered and that people buy, and the result is that the network is sometimes unacceptably slow and congested.
How could they make such a mistake?
Well, the explanation from the NBN – the people who made the mistake – is that such use could never have been predicted, has come as a total surprise, and they could never have anticipated it, and at least in this article they blame Netflix.
It’s easy to show that this is just rubbish, as a quick check on history will reveal. Internet video has been around since 1995. The international standard for streaming video was agreed in 1998. Skype – a platform based on real-time internet video, was founded in 2003, and Netflix, the company that’s very prominently blamed for the NBN’s problems, started streaming video in 2010, just one year after the NBN started, and before the coalition radically changed the technology mix.
Yet the NBN, full of highly skilled network engineers, is trying to tell us that they were designing a world-class broadband network. A network based on the latest technology that would provide opportunities for as-yet-unimagined applications and uses. A network to take us into the future – certainly the next few years.
Even back in 2009 ideas like tele-health were already envisaged, and streaming video was already in use, and one thing any system or network engineer will tell you is that demand will always outstrip supply. You don’t need a PhD in network traffic to understand the clear message from this graph. In 2009 the pattern was clear, and it hasn’t changed since then.
Even the mythical person in the pub can understand that if you are building new infrastructure, whether it’s a road or a network, you don’t build it for today’s traffic, but for tomorrow’s, and that was the whole point of the exercise – to put us ahead of the game, not to watch us drop steadily down the global rankings year after year. To now offer the excuse “oh, we didn’t see that coming”, where “that” is an increase in network use, exactly the kind of thing that was predicted when the network was commissioned, is just breathtaking:
Netflix may have been introduced here in 2015, four years ago, but it existed overseas in 2010 and the basic technology has been in commercial use since 2003, unless the network engineers weren’t paying attention. But apparently the ABC hasn’t considered this either, and has uncritically accepted the NBN’s rubbish explanation and is now repeating it as gospel:
As I’ve already suggested the correct way to say this is:
But then people started using the network more, and we absolutely didn’t see that coming.
The harm of the “Netflix” lie
The ABC repeating this “Netflix” fable does serious harm in a number of ways.
Firstly, telling the story this way absolves the NBN of responsibility for a problem of their own making – without the author seeking any other expert analysis or commentary, or subjecting their explanation to even basic scrutiny.
Second, in unquestioningly accepting the NBN’s explanation it conveniently shifts the “blame” onto services like Netflix and its customers, who are only legitimately using the network that they were promised and have paid for. They’re saying, in essence, “this isn’t poor network design, it’s Netflix”.
Third, that shift in “blame” in turn sets the stage and provides support for exactly a scenario the NBN suggested six days before the ABC story – that if users of services like Netflix are the cause of the problem then perhaps they should pay some kind of premium (or tax, or tariff) for that particular kind of traffic. Here’s a quote from the proposal:
This whole story about “the Netflix effect” ties in neatly with the NBN’s pricing proposal, and is superficially plausible as long as you don’t challenge the initial assumption – that the problem is caused by Netflix, not poor design – or examine the fundamental inequity of the NBN’s floated charging solution.
What a coincidence
Given the NBN pricing proposal it’s hard not to wonder at the timing and content of the article.
Although the ABC has published a number of stories about the NBN in the last twelve months it has been a year since the ABC made any mention of Netflix and the NBN in the same story. Then suddenly, six days after the NBN proposes some kind of different charging for “Netflix” bytes, this article appears talking repeatedly about “the Netflix effect” and explaining how “Netflix” is the cause of all the fixed-wireless woes. A “Netflix effect” that doesn’t exist anywhere else except in this article.
It’s a catchy term, and much more flattering to the NBN than “the poor design effect”.
The first time it’s mentioned in the article the author calls it “the so-called ‘Netflix effect'”, however he conveniently neglects to say who coined that phrase or where, or to provide any actual explanation as to what it really means. Thereafter in the article it’s referred to simply as “the Netflix effect” – no-longer “so-called” – as though its existence is now established and confirmed, although in fact it has no credible, verifiable source and no definition.
In fact the author goes so far as to put the phrase “the Netflix effect” into the mouth of the NBN’s CEO twelve months ago in evidence to Senate Estimates:
The only problem with this is that Morrow never said that. The story’s author, Mr. Thompson, has artfully put his favourite phrase “the Netflix effect” into the CEO’s mouth during evidence to the Senate a year ago. He has carefully not made this a direct quotation, since that would have been totally improper, but he’s done the next best thing and made Morrow use his catch-phrase in paraphrase, despite the fact that Morrow never said it.
How convenient this story is for the NBN, and in the most generous analysis how credulous and uncritical of the ABC. A less generous interpretation is that the ABC is doing unpaid public relations for the NBN, rather than fulfilling its funded, statutory role as a public broadcaster and member of the fourth estate.
Of the three harms I mentioned from the article it’s the last, setting the stage for a “Netflix levy”, that I believe is the most pernicious. If you uncritically accept the premise that Netflix has caused the current problem, you might easily accept a suggestion that a Netflix levy would be a sensible solution.
What that is basically saying is that although you’ve already paid for a certain speed and for a certain volume of data, your data is now going to be examined, and even if you’re not using more speed or more volume than you paid for you’ll be charged extra for particular kinds of bits and bytes.
Not only is this based on the incorrect assumption that the cause of the problem is Netflix users, it’s also a violation of Net Neutrality and clearly inequitable.
It’s like a parcel service charging you for next-day delivery of a one kilogram parcel, and then after you’ve paid, opening the parcel and saying that because it contains a red one kilogram book rather than a green one kilogram book you have to pay extra!
The principle of Net Neutrality, which is enshrined in law in many countries but not, unfortunately, in Australia, says precisely that – that network carriers should move your data from source to destination independently of what the data represents, because it’s not the business or responsibility of the carrier to know or care what the bits mean. You’re paying for the speed and the volume, and after that it’s none of their business.
In fact, paradoxically, in most businesses the more you use the cheaper the unit price. Here it’s being suggested that if you use more capacity – that you’ve already paid for – you’ll get charged extra!
Over the last week there has been such a backlash from all quarters after the NBN’s proposal became public that they quickly published what they disingenuously call a “Fact Check“, in which they say:
Isn’t that reassuring: “… an area that requires attention …”? I’ll let you judge whether that reasonably describes “… a price response whereby charging of streaming video could be differentiated …”, or whether differentiated charging sounds more like a tax. If it walks like a duck, and quacks like a duck…
To summarise, I think it’s fair to say that if you replaced “Netflix” or “the Netflix effect” everywhere in the ABC story with “poor network design” you’d have something far closer to the truth, and something that I believe shows this ABC story for the shoddy journalism it is. I believe Netflix is being used as a whipping-boy for the NBN’s self-inflicted problems, and this article, while appearing to be critical of the NBN in places, really works very hard to suggest that it’s all due to Netflix while never giving Netflix or any other network experts an opportunity to reply.
Fixed-wireless is only 5% of the NBN
The last point I’ll make before deconstructing the story is that all of this is ostensibly about the fixed-wireless network only.
That is, about 5% of the total NBN customer base!
That’s not to say that their problems aren’t important, or their access less worthy. In fact, because they’re not spread equally across Australia’s demographics but are almost entirely located in regional and remote areas it becomes a critical question about the basic structural equity of the whole NBN. After all this was supposed to be a national network.
However if you read just the headline, as many do, “How ‘the Netflix effect’ killed the NBN’s dream” you could reasonably be excused for assuming this was the whole NBN, not five percent of it. Even the main article headline leaves “… on fixed wireless” as an afterthought, and unless you’re a technical aficionado you could still be excused for not realising that this was about only a small section of the network.
This deliberate vagueness, and the frequent repetition of “Netflix” and “the Netflix effect” serve to cement in the reader’s mind the idea that whatever the NBN problem is, it’s caused by Netflix, and from there it’s a short step to the NBN’s “differentiated charging” as a solution.
Deconstructing the ABC story
In the beginning of this article I promised I’d provide details on my assertions about false or misleading content in the ABC story. Some of that is included in the discussion above, but for completeness here is a quick analysis of the major points.
The headline on the main ABC News page
As already discussed, this headline, which is often all that anyone reads, is highly misleading because it implies the problem affects the whole NBN, whereas it affects 5% of it. I believe this, to begin with, is highly irresponsible.
I understand that sub-editors or producers are usually responsible for headlines, not the story author, however that just shifts the specific responsibility for this somewhere else in the ABC. As we’ll see, this isn’t the only problem.
Analysing the headline itself, I believe almost every word is wrong or misleading, something of an achievement! First of all, as discussed above, there is no such thing as “the Netflix effect”, and the very deliberate naming of Netflix in the thumbnail is amplified by the deliberate use of their corporate logo as the picture tag. This isn’t just irresponsible, it’s arguably malicious, and the mischief is compounded by the fact that Netflix wasn’t approached for comment.
Second, “killed”. A highly emotive word, suggesting finality, completeness and, implicitly, violence. It also implies a killer and a victim, in this case Netflix and the NBN’s “dream”. This is nonsense. Apart from the fact that “the Netflix effect” doesn’t exist, nothing has been killed. Poor design has been exposed, certainly, and expectations dashed, but nothing has been killed, but now Netflix has been associated with a major “crime”.
“The NBN’s dream.” This is laughable. The dream was the ALP’s; the NBN is an engineering and sales organisation created to design, implement and realise it. The NBN is ten years old and was supposed to have finished realising the dream three years ago. The dream was turned into cold, hard engineering reality long ago, and we’ve already spent tens of billions of taxpayer dollars on real infrastructure, not dreams. If the NBN is dreaming it’s time it came down to earth and addressed some real problems. In any case, it’s not the whole dream we’re actually talking about here, but rather a small portion of the realised design.
The real headline I believe should be something like “Regional NBN users denied promised access by poor design“.
The summary paragraph shown in the thumbnail is an almost word for word repetition from the body of the article, so that isn’t the work of a sub-editor but the author. As such, the final sentence “But then along came Netflix and everything changed” can’t be blamed on anyone but Mr. Thompson.
As I’ve discussed, there was no “along came Netflix”. Netflix has existed for almost as long as the NBN, and it’s by no means the first or only user of streaming bandwidth, an entirely predictable growth path for the network. If it hadn’t been Netflix it would have been something else, and not their fault either.
The reality of the situation is that the designed capacity was the problem, and it was simply waiting for the volume of use to reach a critical point to cause the congestion that’s being seen. So instead of “along came Netflix” the real phrase should be “Then usage increased“. When the camel’s back breaks it’s not the straw’s responsibility, it’s the owner of the camel, or in this case the designer and owner. The NBN.
When use increased past the network’s capacity what happened was entirely predictable, as anyone who drives in peak-hour traffic knows. Congestion, and low speed. This is not “everything changed”, but rather, “and a totally predictable and gradual degradation of service occurred“.
So instead of “then along came Netflix and everything changed” I believe the story should read “then usage increased and a totally predictable and gradual degradation of service occurred.”
I’ve already covered most aspects of this slightly longer headline. This time the headline writer has been honest enough to include “on fixed wireless”, however for the average reader the distinction is probably lost, leaving them, again, with the understanding that Netflix has killed the NBN.
As you’ll probably understand by now there is no “streaming revolution”, unless every developing use of the network is a “revolution”. Again, it’s a fact of life in computing, networking and technology in general that increasing demand is as certain as night and day. You can’t always predict what will cause it, but it’s not a revolution when it happens, rather it’s an entirely predictable evolution.
“Super fast broadband” also deserves a comment. It’s a measure of the NBN’s failure that in some fixed-wireless regions it cannot sustain even 3Mbps during peak usage. At last year’s Senate Estimates the CEO confirmed that the NBN “killed” (his word) the 100Mbps speed from the fixed-wireless roadmap – in other words, fast broadband. In most places in the world “super fast broadband” is Gigabit broadband, not Megabit, and certainly not 100 Megabit, so the NBN’s “dream” was a long, long way short of “super fast” to begin with, and our world ranking confirms it. For the author to use the phrase “super fast broadband” in the face of these easily discoverable facts again doesn’t speak well of the integrity of this article, at least in my opinion.
Even more telling is the fact that the CEO, Morrow, confirmed that the reason 100Mbps was being “killed” was that the cost to provision it would be prohibitive. Now cost, for a given speed or volume, is a variable that only ever comes down for telecoms hardware, so if it’s prohibitive now it was certainly prohibitive a year or more ago, and it gives the lie to the idea that it was ever really on the table. Indeed, if the author’s assertion is correct, if “everything changed” when Netflix “came along”, then that was 2015, and in 2015 the engineers would have had five years of overseas experience with Netflix to draw on. They would have known four years ago that 100Mbps fixed-wireless was unattainable, yet it has only recently been removed, demonstrating yet again that this is about design, not some recent realisation about Netflix. But don’t tell the ABC.
So “Netflix” didn’t “kill” fast fixed-wireless broadband, the NBN did, for cost reasons. So this headline is as mendacious and mischievous as the previous one, and should again read “Regional NBN users denied promised access by poor design“
The problems come thick and fast here, and if you’ve read this far you wouldn’t thank me for dissecting every one. I’ll just hit the low spots.
The story starts off with “About 10 years ago … streaming TV was not a thing.”
Arguably true, but disingenuous.
Here are some other things that were “not a thing” around then: Uber, Chrome, Airbnb, iPad, Spotify, Oculus VR, Instagram, Kickstarter, GPS on phones, Lyft, and Pinterest.
I hope you get the point. It’s not possible to predict what’s going to be “a thing”, and for networking projections it’s not necessary – just refer to the graph earlier in this article. What you can confidently predict is that demand for bandwidth will increase, and it will increase more than linearly.
Guess what else wasn’t “a thing” in 2009? 4G, the technology on which fixed-wireless is now based.
So saying that something wasn’t “a thing” is disingenuous, because it sets up the consequent “if it wasn’t a thing, then we couldn’t have been expected to predict it”. But as we can see it’s up to network engineers to anticipate growth in demand, not predict the specific applications that will generate it. At the time the NBN was launched even the technology currently used in fixed-wireless wasn’t “a thing”, so this line of argument is obviously deliberately misleading. What was “a thing” was decades of experience in internet growth, and more than a hundred years’ experience in networking and network growth.
Next, “But then along came Netflix and everything changed.” Here is the pivot point of the article, the sleight of hand that makes Netflix the villain of the piece and removes the NBN from the picture. As I’ve hammered this to death earlier I won’t say any more, except to say that this sentence really sums up what I think is the fundamental dishonesty of this story.
It’s followed immediately by a quote from the NBN, again about the “introduction of Netflix”, effectively confirming the author’s assertion that everything changed, and it was when Netflix arrived.
Note that we’re only a few paragraphs into the story and Netflix has been mentioned three times and their corporate logo featured in the front page thumbnail, but the company isn’t even given the courtesy of a response.
Next we see the first use of the author’s pet catch-phrase, “the Netflix effect”. As discussed earlier, the first time it’s used he refers to it as “so-called ‘Netflix effect'”, but he fails to say who calls it that, or when they did, or in fact what it specifically means. If this reminds you of Donald Trump saying “some people say…” it’s not a coincidence. It’s a cheap trick, to pretend something you made up yourself actually has general currency and is widely accepted.
The next time “the Netflix effect” makes an appearance, it’s apparently straight from the mouth of the NBN CEO, Bill Morrow. As discussed previously, I believe this is another cheap, even shabby, journalistic trick, as Morrow never said that, but the impression the reader gets is that this “effect” was a real thing known to the CEO over a year ago when he supposedly used the term.
It makes yet another appearance later in the story “… because of the Netflix effect the NBN now considers a 6mbps (sic) download speed to be acceptable …”. This time the author is claiming that the NBN’s decision is attributable to something it appears he just made up, because it’s certainly not something that ever appeared in the Senate Estimates discussions from which the rest of the sentence is drawn. However he repeats the catch-phrase as though it’s a real thing, and again blames “the Netflix effect” as being what’s responsible for the NBN’s woeful fixed-line performance, not the NBN’s design and implementation decisions.
The remainder of the story focusses largely on the NBN’s decisions and shortcomings in deploying fixed-wireless technology, and is unremarkable. Although it’s clear that some customers have been forced onto fixed-wireless for what appear purely political, project scheduling reasons, and it’s clear that fixed-wireless will not meet its original bold promises, the fundamental thrust of the article remains: everything was going well until Netflix turned up, and it’s all because of them.
While it’s true that some of the article is critical of the NBN, the headlines, introduction and fundamental thrust of the article remain firmly fixed on this explanation that ultimately it’s Netflix and not the NBN that’s the root cause of all the problems.
This is appalling, lazy journalism in my opinion, and the ABC should retract the story and apologise to Netflix and the public. In particular they should never use the term “Netflix effect” again, and next time they decide to write about networking technology they should invite independent credentialed professionals and affected industry bodies to participate.
[My credentials? I’ve worked in IT for over forty years in a variety of roles that could very broadly be described as software development. I’ve served on the executive of a national user body, and participated in the development of international standards. I was working professionally in a Computer Science department and involved in both LAN and WAN networking in the 1980s before any permanent Australian connection to the US was made, and was personally involved when that first permanent Australian connection to the Internet (AARNET) was put in place. I’ve followed, used and participated in the growth of computer and internet networking ever since, from 300 baud acoustic couplers upwards. I am not a network engineer, but I believe I have enough specific knowledge in the area to review the ABC story and write this article.]