Stilgherrian writes now that speed issues with filtering are out of the way (see ARN’s article about ISPs involved in the trial giving ‘filters the technical greenlight’), ‘come September the debate will be about whether internet filtering “works” or not’. He’s right and therein lies the problem: Senator Conroy hasn’t defined what works.
Here’s Stil:
Now anecdotes are not evidence. The absence of evidence of problems is not evidence of absence — particularly when some ISPs are trialling the technology with as few as 15 users. But this is politics, not science. It puts the kibosh on GetUp!’s original anti-filter campaign, which focused on the potential for filters to slow our internet connections.
Now as Crikey reported in April, before we can say, “It works”, we need to define what “It” is. We have to define the target, i.e. what we intend to block. And we have to define the accuracy, i.e. how much incorrect blocking or failure to block we, as a society, will accept.
[...]
On the question of accuracy, Senator Conroy steadfastly refuses to provide a definition. Witness this exchange with Triple J’s Kate O’Toole from 7 April:
O’TOOLE: So if the trial does fail, and you’ve got to be open to that in a trial…
CONROY: We’ve said we’d be evidence-based.
O’TOOLE: …you’re going to abandon the…?
CONROY: We’ve said if the trial shows that this cannot be done, then we won’t do it.
O’TOOLE: And what’s the definition of “cannot be done”? What would be the acceptable amount to slow the internet down?
CONROY: Well now you’re asking me to pre-empt the outcome of the trial.
O’TOOLE: No no no, I’m not. You’ve got to have a … before you start the trial you’ve got to have an understanding of what is, what’s a pass and a fail, like you’ve got to be able to measure … you can’t sort of wait until the trial finishes and then look back and decide how you’re going to measure the outcome.
CONROY: Well actually that’s how you conduct a trial. You wait to see what the result is and then you make a decision based on the result. If the trial shows it cannot be done without slowing the internet down, then we will not do it. But if the trial comes back and says it can be done, then we will go down the path of blocking the RC, and we will look at how it’s possible for parents to be given a choice, a menu of options, that they can block their children from accessing.
O’TOOLE: So do you have a rate of over-blocking in mind that would be unacceptable?
CONROY: Well as I’ve said, let’s wait to see what the trial shows us.
O’TOOLE: And then you’ll put the goal posts up afterwards?
CONROY: As I said, you want to pre-empt the trial, then we’re happy to wait to see what the trial comes back to us with. Perfectly happy.
You may recall that a Freedom of Information request also confirmed that no success criteria has been set for the filtering pilot.
Whether the trial is accurate has never been important to the government. There is also little comfort to be had when one of the ISPs participating in trial, Webshield, already filters customers Internet access and a slowdown may not be noticed.
Others ISPs participating have said they may offer filtering as a service after the trial, regardless of whether the government makes it mandatory. It’s no stretch to imagine the equipment they use will be the equipment they used in the trial and badmouthing it now wouldn’t exactly make good business sense, would it?
Associate professor at the University of Sydney and no stranger to this debate, Bjorn Landfeldt, told ARN today that ‘while large telcos had the staff and capital required to manage a compulsory blacklist, smaller ISPs probably would not’.
One of my first posts on this blog pointed to a Howard-era report which echoed the same thing.
Landfeldt to ARN:
“Australia has got a very heavy, long tail of very small ISPs. For all those ISPs, every dollar they have to spend and every minute they have to spend on something other than their core business is potentially very damaging for their business,” he said.
“I would be surprised and interested to find out whether there are any solutions available that could be integrated and managed and run by the small players without any major impact.”
Landfeldt also that the small number of customers that took part in some of the trials negated the usefulness of any data gathered during the trial.
“If you only have 15 people that opted into the trial, I don’t know what you’d get out of that,” he said. “The burden of running the filter would be absolutely negligible.”
[...]
“I believe the trial is fairly futile because we know that it can be done,” he said.
That’s debatable; whether it works isn’t. With countless legal websites blocked and the certainity of more to come, the protection of children online untouched and the financial risk to small businesses if their websites are unintentionally blocked, we all know the answer is no, it doesn’t work.
14 comments
Andrew Garrett (werdna) 's status on Monday, 27-Jul-09 14:47:45 UTC - Identi.ca says:
Jul 28, 2009
[...] http://www.somebodythinkofthechildren.com/filtering-what-works-what-kills-small-isps/ [...]
Icaria says:
Jul 28, 2009
It’s been pretty clear for a while that dire network performance impact could be side-stepped. It’s a shame many of us opposed to the filter proved as ignorant of the technology as the filter’s proponents.
We’ve also known this has been coming since it came to light that the, “live trails”, had no real criteria for success and constituted little more than a disingenuous effort by the gov’t to appear objective. While people have certainly de-emphasised the performance argument of late (as opposed to it utterly dominating the debate earlier), it’s still unfortunate that this is what is going to define opposition to the filter; this is how we’ll be pigeon-holed, as we start to argue what should have been the focus from the beginning.
Marcus says:
Jul 28, 2009
There is a relevant article from the Coding Horror blog, describing how different programs can (in terms of speed) scale in different fashions with larger datasets. He actually describes how inexperienced programmers test with small datasets and are then surprised when their program falls over with larger datasets.
http://www.codinghorror.com/blog/archives/000957.html
I agree with Icaria though, the technological failings of this filter are minor compared to the other issues such as the whole RC vs illegality issue.
And those comments from Conroy spoken at O’Toole (it wasn’t really a conversation was it?) are quite incredible, I hope they get a lot of publicity!
Stilgherrian says:
Jul 28, 2009
The government has been in control of framing this debate from the very beginning. The very labelling of the plan as an Internet “filter” and a “clean feed” means it becomes akin to “filtered water” — all the dirty bits removed, something that’s actually quite healthy.
This hasn’t been helped by some anti-censorship writers going on about their “right” to view pornography. That sort of thing won’t sway the battlers in marginal electorates who fear the Internet’s effect on their vulnerable children.
Opponents of compulsory Internet censorship seriously need to read George Lakoff’s work and start to frame the debate in their own terms.
Some possible starting-points are: actually helping children face the real risk, which is bullying; pointing out that every dollar spent (ineffectively) fighting imaginary risks is a dollar that isn’t spent on reality.
If Senator Conroy was here now, he’d probably say, “Well, there are parents who are genuinely worried about this stuff, so what are you offering them?”
Dan says:
Jul 28, 2009
I have been conducting my own tests to see if Internet Censorship causes a slowdown. The test is quite simple, I configure my firewall to block certain domains. I then attempt to access the blocked domains.
From my tests I can conclude that Internet Censorship causes virtually 100% slowdown. I plan to conduct further trials with my router unplugged and see if there is any difference between the results of the two tests.
Mick says:
Jul 28, 2009
Whether or not filtering is feasible is irrelevant. There just shouldn’t be that type of government control on what people access from within their own home. Filtering connections in parliament, schools and libraries for all I care, but stay the fuck away from MY CONNECTION!
Once again Conroy totally misses the point.
voracity says:
Jul 28, 2009
Fascinating. Eagerly awaiting the results of those next tests, Dan.
Again, I agree with Icaria (and not just in hindsight — I argued this on Whirlpool long ago). What bothered me was the claims of ‘Up to 87% slow down!’ when that result was due to one particularly poorly performing filter that no ISP was ever going to choose.
We need to make our arguments not against the worst performing filter, nor even the best. We need to make our argument against the best filter imaginable, because the filtering industry will increasingly create filters that approach that standard. We might be able to bat off ridiculous censorship with technical arguments today, but we do a disservice to the generations living 10, 30 or 50 years from now by not settling the argument for good.
Nick Andrew says:
Jul 28, 2009
Reading Conroy’s exchange with O’Toole reminds me, by way of contrast, with the James Randi Educational Foundation’s million-dollar prize to “any person or persons who can demonstrate any psychic, supernatural or paranormal ability of any kind under mutually agreed upon scientific conditions.”
The key words here are “mutually agreed-upon scientific conditions”. While James Randi and Co are usually seen as debunking psychics and other fraudsters, those people are actually debunking themselves. The million-dollar prize stands, through no lack of challengers, because the JREF agrees upon a test or a series of tests with the person claiming paranormal ability, and this specifically includes the success criteria. This allows an objective assessment of whether the subject passed or failed the test. To date, all subjects have failed the test, despite having previously agreed that the test was fair and the success criteria were fair.
So I see quite a contrast with Senator Conroy who has run a “trial” but refused to publish or perhaps even consider success criteria for this trial, instead handwaving away the question with a weak “You wait to see what the result is and then you make a decision based on the result.” How much better for everybody the outcome may have been if Senator Conroy had taken a leaf out of Randi’s book and had determined success criteria (or even negotiated it) before beginning this trial.
james says:
Jul 29, 2009
I’m reminded of a joke about a child resting near a barn. On the wall, targets have been painted, arrows dead in the centre of every single one.
A man rides past and hails: “That’s some pretty solid shooting, son.”
“That?” He says. “Ain’t nuthin’ to it. You just fire the arrows against the wall and paint the target around it.”
Sam D says:
Jul 30, 2009
@voracity: I totally agree. Whether or not you can do something is much less important that whether or not you should do it!
I would be against any censorship of the internet, even if it didn’t slow it down at all. I’d be against it even if it made my broadband go faster baked me cookies and gave me foot massages.
The trouble is that many people in this debate (especially on Whirlpool) are experts on the technical side of things, but know bugger all about ethics.
Daniel says:
Jul 30, 2009
Of course the filter will slow down the internet. That’s what any filter does. You can’t filtrate without the flow itself slowing down. I’ve seen it in China, agonizingly slow internet due to top down packet filtering and it will happen here too. It’s as certain as gravity.
But then, already the internet’s sometimes very slow in any case. It’s going to be extremely difficult for people to go “Hey! My internetnet’s slow today! Must be because of the filter!” and go and prove that. It would be almost impossible to pin slow internet speeds down to packet filtering. Once again, to use China as an example, some of the internet’s really very fast and some if it just crawls along at snail speeds. What crawled yesterday might speed along today and what sped along yesterday might crawl or just be blocked tomorrow. But you never know if that’s because Beijing is filtering your packets or because your ISP is jammed up today. Slow internet – as well as blocked web sites like Wikipedia – is just a fact of life there and you deal with it and move on.
I reiterate my belief that Conroy will have his packet filtering no matter what anyone says or how much people protest. Because our leaders cannot sit back and just let such an important source of information like the internet continue to go unregulated and uncensored. Historically, it’s just not how we do things in this country. We censor and regulate everything, including the pet cat, in Australia. And once it’s in place, whenever people blame problems on the filter, the government will just shift the blame and say “but you don’t really know if that’s the filter. It’s most likely just a problem with your connection.”
Basicaly: when it comes to ISP level filtering, we are rooted.
Sam D says:
Jul 31, 2009
@ Daniel:
Yes, the Filter will probably slow the Internet. The point is that if Conroy has ‘proof’ that it won’t, then your arguments about speed instantly loose traction with anyone who doesn’t understand the technical issues.
Dan says:
Jul 31, 2009
Voracity, is right. Our argument’s should not be biased on technology.
We need to take the moral stance, not just for this generation but also for all future generations of Australian’s who should have the right to free-speech and not live in a society where other people can oppress and silence them into submission.
Ghostdoc says:
Aug 17, 2009
we won’t be rooted, we’ll be routed (ba-dum-cha, thanks, I’ll be here all week)
As I understand it, there are 5 reasons to oppose the filter:
1. it’ll slow down the internet for users
2. it enables politicians to control what the public can see on the internet (political censorship)
3. the list of blocked sites is secret, and so therefore there is no appeal process for a blocked site
4. the list of blocked sites is secret, so therefore there is no method of determining if it is being used inappropriately by the politicians
5. it introduces a burden on the ISP’s, both procedural and legal,that will discourage competition and raise prices.
The so-called trial proved that, maybe, reason number 1 is a bit dodgy for small numbers of users on a small number of ISP’s. There’s still plenty of other reasons that this is a really really bad idea.
The singe ‘best’ reason for implementing the filter is to protect children from abusers. However, all the studies have shown that the single largest threat to children (outside of family members which no-one can protect them from) is authority figures in their community. Yet no-one is talking about making sure that children are safe from these people. The recent Irish investigation found that the Catholic church has abused thousands of children over decades of abuse. The largest single source of child abuse in the world is the Catholic Church, yet no-one is even talking about monitoring Catholic priests or protecting children from them. Instead they suggest that the internet needs monitoring?
If we oppose the filter, we need to get the message across that the filter is there to save politicians, not children.