Speech-gating and privacy tradeoffs

A friend asked about my zealotry on age-gating social media and why internet anonymity/pseudonymity should be protected. 

I sent this through by email, figured I'd share it here too, very lightly edited. 

I agree that people are more likely to be jerks when anonymous. 

Potential for anonymity is simultaneously important protection when society or governments are too censorious, or could become so, or when someone in one country worries about consequences for family who live in a censorious country. 

If Twitter had a magic tech that could undo VPNs and put a note under each tweet specifying the real country of origin of any tweet, that wouldn’t bother me any. Or if another social network decided to only allow real name accounts, and people wanted to chat there, that’s perfectly fine with me too. 

Government requirements on these platforms seem like a different matter entirely.

The UK’s rules require that every site whose content could possibly be viewed as harmful to children now require adults to prove that they are not children. One very predictable consequence: Discord was hacked, and 70,000 had their government-issued IDs stolen. 

But even without hacks, requirements to provide your ID if you want to look at sensitive content can chill speech and access to information. “Oh, you want to read that article about the STD you think you might have? Just show us your ID. And trust that we won’t tell anybody what you were looking at.” Or, “Oh, I see you’re interested in a very niche form of pornography. Please give me your ID.” “Oh, I see you’re interested in academic journal articles about the heritability of different traits. Papers please.” 

Freedom of speech includes the freedom to read. I would consider it a violation of freedom of speech if a librarian came up and demanded to take a record of my ID if I were minding my own business and peacefully reading subversive books. 

Did any of the proponents of the UK legislation think that ‘just protecting children’ would mean adults would have to give Wikipedia their ID to read some of their articles, or to every dodgy porn site? Probably not. They’re not good at thinking about tech or consequences or how companies protect themselves against substantial potential liability. 

But that’s where the UK now is. As consequence, VPN downloads have massively increased in the UK so people can pretend they’re not in the UK. And the UK government is investigating whether it’s possible to ban VPNs. 

And as consequence, I’ve gotten a multi-year subscription to a good VPN, because the obvious way for governments to ban VPNs is to ban domestically issued credit cards from processing payments to them. And New Zealand seems determined to be every bit as stupid as the UK. I understand that Select Committee has asked for advice about the feasibility of banning VPNs as part of their inquiry into ‘just protecting the children’. Great stuff. 

Australia’s version will require social media platforms to keep under-16s from having accounts. There are triaging efforts they can take, but a pile of users who are over the age limit will wind up having to prove that they are over the age limit. 

The system can then either err on the side of letting youths on accidentally (and suffer penalties for doing so), impose substantial burdens on adult users to avoid letting youths on accidentally (frequent age re-verification to make sure that a youth hasn’t faked being over the age limit with the help of a confederate), or end the potential for anonymous/pseudonymous accounts by requiring real ID. 

If I were a Chinese student studying in Australia and had set up a pseudonymous Twitter account for fear that my family in China would be punished for what I might say in Australia – and to maintain the option to go back to China if I wanted to – I would end my account rather than provide ID. Because IDs can be hacked, and the Chinese government is good at that. Sure, they could probably figure out who owned an account if they tried really hard absent that. But having a giant database that just hands over all the IDs you need to check? Risky. 

There are ways of doing it that are privacy preserving. But it would be very easy for a fifteen year old, with the cooperation of a seventeen year old friend, to get an account under those methods. And when that becomes obvious, those for whom potential harms to children trump everything else in the world will call it a ‘loophole’, demand the loophole be closed, and the potential for pseudonymity disappears. 

The same problem applies to what I’d put up as potential least-bad solution in my submission on this stuff. Government could require that platforms age-restrict at the app-store level such that parental authorisation is required through Family Link or the Apple equivalent. 

It would be far from foolproof: not everyone has an age in their phone account (Google or Apple); people could buy cheap phones and not disclose true age. And it would have some harms too: the State would be enforcing the right of a Gloriavale mum to block her son from important aspects of contact with the outside world. Which is a bit odd when a 15 year old can get an abortion without parental notification or consent; doctors are forbidden from telling the parents if the child so-requests. We have very odd and inconsistent rules about what things kids can and cannot do without parental consent. 

My solution would also have obvious holes. So I urged that, if parliament picked up that option, it do so very deliberately with recognition of that lots of kids would work around the rule and that that was the least bad alternative – that nobody frame it as being other than very second-best, and that nobody pretend afterwards that kids getting around it was a reason to revise the rules. 

It still sucks though and is far more risky than I’m comfortable with. 

I guess one bottom line: shouldn't we let the rest of the world have a few attempts at this before jumping into anything we might regret? 

Other bits of context:

• Justice Harvey on continued government efforts to exert control over internet content
• My submission on this stuff
• Eric Goldman's piece on the "Segregate-and-Suppress" approach to regulating child safety online. He puts up a very good argument against legislating parental signoff for kids installing the apps. He's right. I just think it sucks less than anything else Parliament might do. 
• The Harmful Digital Communications Act has itself been used to abusive purpose. 

The Ring: discount rates edition

Locked at the bottom of a dark well is a beautiful-looking idea. 

"Set discount rates to very low levels when thinking about intergenerational issues."

It is a beautiful looking idea. We don't discount utilities. Future people matter. Conventional discount rates mean that far-off benefits or costs count for next to nil. Don't we care about future people?

Let the beautiful idea out of the well. 

Surely it will be used only for very long-term lovely environmental projects with century-long returns. 

Who locked it in the well anyway? It had to have been bad people for bad reasons.

Let the beautiful idea out of the well. Free it!

A couple of decades ago, Treasury was the place that warned about public choice considerations. About how this sort of thing might be a bad idea, and the reasons it would be a bad idea, and the safeguards that might be needed if you wanted to let the thing out of the well. 

Lately, Treasury's been drinking from the well instead. 

I was pretty sceptical of the initiative, figuring that the proposed dual discount regime would lead to shenanigans. 

An earlier version of this stuff had a 5% discount rate for social investment projects. Tauranga's business case for the Te Manawataki o Te Papa civic redevelopment consequently claimed that the thing was a social investment project that could get the 5% discount rate. Of course this stuff was going to encourage shenanigans. 

The final version sets a standard discount rate for 'commercial' projects at 8%, and that only 'non-commercial' projects would get the much lower discount rate: 2% in years 1-30, 1.5% in years 31-100, and 1% in years 101+. Along with a 'mandatory' sensitivity test at 8%.

And again without and decent definitions or guardrails - just hopeful suggestions about considerations that might affect whether something is commercial or non-commercial. Or any explanation of what enforces mandatory sensitivity tests in a world where zero other Ministries or Agencies respect Treasury as enforcer on anything any longer. 

But boy this sure sounds hopeful and full of rainbows and unicorns.

Updating the discounting regime to use SRTP as the default for proposals with mainly non-commercial costs and benefits makes the CBA settings more technically robust, especially in a policy environment where there is increasing focus on long-term impacts of policy. 

They let it out of the well. 

And it's going about as well as expected.

NZTA in January decided that "most transport activities undertaken by the public sector" are non-commercial. And, in fairness, there's no commercial case for most stuff NZTA now does. So it's 'non-commercial' in that sense. 

And so earlier this week a pile of new Roads of National Significance got announced, many with very low benefit-cost ratios, with the things assessed at a 2% discount rate. And no sensitivity testing at 8% that I can find. 

It's terrible of course. 

But there's a poetry to it. 

Folks wanted to let her out of the well to justify environmental projects. 

And instead we get more motorways.

Dieter warned you about some of this. And you didn't listen. 

Treasury is a fallen place that has forgotten the faces of its fathers. 

Gating speech

Being able to speak online freely, under a pseudonym, is an important backstop for freedom of speech.

If government turns repressive, or bans some kinds of speech, or locks you out of some kinds of employment for having used the wrong words, at least you can online under a pseudonym. 

Well, unless you're only allowed online after logging into a government app. Then it's game over. 

Nevertheless, some in Wellington are itching to follow Australia’s lead. 

While legislation is in its early stages in Parliament with Catherine Wedd’s Member’s Bill, Education Minister Erica Stanford is leading cross-agency work to explore viable legislative options and practical implementation.  

The Prime Minister has publicly supported moving ahead.  

Paul James, the Government chief digital officer, told me on The Business of Tech podcast last week that the digital identity verification system underpinning the upcoming all-of-government app and digital wallet could offer the age verification needed to support a ban – if the Government seeks to enforce one.

I've suggested before that a trilemma applies here, or something like one. Online age gating requires adults to prove that they're not kids. 

And that means at least one of three problems.  

A system could be easily worked around by kids.

It could be very cumbersome for those over the age limit.

Or it could be the end of online pseudonymity and privacy.

Suppose that the government-app runs a zero knowledge proof solution. You log into the app. It generates a key verifying that the person logged into the app is over some age. You use that key with your favourite social media app. It verifies the key's authenticity anonymously, so the app doesn't know which platform is checking. And you're set - privacy maintained. 

But that's easily worked around if a sixteen year old logs into the government app on his fifteen-year-old friend's phone. 

And enough kids doing that means the Before 16 lobby group will characterise it as a loophole that must be closed. And then the verification is no longer done through a ZKP, ending online privacy, or we have to do daily re-verification to impose a differential cost on under-16s, making things cumbersome for other users. 

I do not like where this is heading. 

Failure can be overdetermined

Last week, on The Platform, Auckland Uni's Prof of Macroeconomics blamed my shop for the government's not adopting his proposal to completely overhaul the health, welfare, and retirement system.

Robert MacCulloch debates Oliver Hartwich from the New Zealand Initiative following the Nicola Willis interview.

Watch the full video at https://t.co/cXDdwFy9gF pic.twitter.com/HjFK0G77W1

— The Platform NZ (@theplatform_nz) September 22, 2025

About a decade ago, Prof MacCulloch and Sir Roger pitched their proposal. 

It looked like it would take considerable effort to see whether the numbers added up. 

It also seemed like something that would be difficult to convince anyone to implement. 

We declined to weigh in. We've neither endorsed the policy nor recommended against it. 

No conspiracy is needed to explain successive governments' failure to pick up the Douglas-MacCulloch proposal. Just the usual inertia. 

The default path for a good policy proposal isn't implementation. The default path is failure. 

Most proposals fail. 

And especially for large changes that are complicated to work out and that the civil service isn't likely to support and that politicians are likely to see as risky. 

I don't think that proposal could have succeeded without enormous amounts of work being put into demonstrably ensuring the numbers stacked up, public comms on explaining it to voters, work with officials so that they'd understand what the thing involved and in hope that they wouldn't wreck it, and work with MPs so they'd see the merits. 

It probably would have required hiring an actual lobbying shop to help. 

Anyway. Bottom line here: we didn't do anything to help or hinder Robert's proposal. 

And it is weird to think that the thing could only have failed due to nefarious influence. 

That isn't how anything works. 

In the video, Robert notes having had a meeting in Bill English's office in which Matt Burgess argued against Robert's compulsory savings scheme - and suggests it's part of the Initiative's push to kill his policy. 

Matt didn't work with us until after his time working in Minister English's office. I was sad when Luxon's office later stole him from us.

Success happens through weird mixtures of luck and timing and work and skill. Failure is the norm, and is generally overdetermined.

We've been successful with a couple of our policy ideas. But we work on and pitch lots of policy ideas, and have for a long time. Failure is the norm, particularly in the short term. 

Think about the number of papers that turn up in NZ Economic Papers that have a policy recommendation and that aren't obviously crazy. What fraction of those turn into actual policy change? Half a percent? Less?

And who knows. Maybe MacCulloch and Douglas's proposal will get a re-airing with long-term superannuation costs again being salient.

But failure is the norm, even for meritorious proposals. No conspiracy is needed to explain failure. It's just what happens. 

The case for optimism

Kerry Howley has a superb piece in New York Magazine on the kids in San Francisco building the future.

Houses and motels turned into dorms for kids in their late teens through mid-20s, building everything from brain scanners through to an AI VC that evaluates funding pitches - with the AI having hired a person to be its real-world presence.

We head downstairs to a dark basement attached to a garage. A slight, long-haired man, a paid test subject solicited through Craigslist, sits before a computer screen, wearing a white cap that looks like a medieval linen coif threaded with wires. The screen flashes images — basil, a blazer, Parmesan cheese. With unsettling clarity, the computer will be able to resurrect the image from electrical signals in the subject’s brain. A subject considers a picture of jelly beans. AI offers a picture of similarly colored beads. A subject looks at a red station wagon; AI presents a red sedan. Until very recently, most people thought the data produced by EEGs, an 80-year-old technology, was noisy garbage. “They just didn’t understand the power of large language models,” Jonathan says. He is 24 years old.

Reading minds is what AI engineers mean when they talk about hard problems. Eventually, the tech will advance to interpret “evoked states.” “So we start with, you know, discrete smaller tasks like emotion, like positive, negative, maybe now ten, 20 emotions. And then we add more dimensionality so that eventually we can go into full sort of inner monologue,” a world of superior self-knowledge wherein we sift through our own memory banks rather than selectively recall events through a haze of misperception. “And,” he says, “we do it all in-house.”

Doing it all in-house looks like this: a server rack with LED-lit fans in the garage next to some exercise equipment and some bicycle helmets. Jonathan and his housemates built the rig themselves. “Just asking ChatGPT basically. You know, you can just ask and then order the parts you need and you learn and you debug.”

Kids partnering with AI to build things, no need to ask anyone's permission. 

Howley's kicker here... so very good. 

How do you make contact with the intelligence rising up from the machines around you? Do you build it a body? Do you offer it yours? It is perhaps tedious to point out that we are always operating under the shadow of destruction, deploying tools that might end us, convincing ourselves, not without reason, that if we don’t build the bomb, someone with worse intentions will. Not a single one of the AI kids had attempted to lecture me about a theory or suggested I read a paper; it was not me they were trying to program. Somewhere along the way, drawn into their swell, I had begun to think of large questions about the nature of AI as New York questions, millennial questions, distant from the center of things. Where it mattered, humans were not debating AI; they were merging with it. You order the parts you need, you learn, you debug. The kids carry on with the crisp clarity of engineers, integrating what is immediately useful, discarding or rewriting what is not. No one will ask your permission to build a world you do not understand.

It's like they're living in an 80s William Gibson novel. Working in spaces where policy and regulators can't really stop them. They don't care about policy papers. They're just building. 

Those spaces matter. Policy is so impossibly stupid. 

Here, Parliament's trying to figure out the best way of breaking the internet, with National wanting to copy Australia's insane social media age-gating bill, and Labour wanting to copy the UK's even worse version. News outlets cheer them on, having always figured that social media platforms are the enemy anyway. Both parties see it as crack cocaine for attracting pivotal female votes in the 35-55 age bracket. 

And I understand that Parliament's select committee looking into this is wondering whether it's possible to regulate VPNs. Because that's been the obvious workaround in the UK for adults who don't want to have to show ID to visit every darned website. So long as at least one country remains free, VPN to it and pretend to be from there. 

I'm cheering for the cyberpunks. They may be our best hope. 

Infrastructure roulette

In some respects it's reasonable to think about city council as being a kind of club.

Everyone who owns property in Wellington is a member of the Wellington Council club. The club levies itself to provide things that the club members want, and to cover off the cost of stuff that central government wants the club to provide that club members may or may not want. For some reason, renters were added as voting members of the club. But the debt that the club issues is ultimately backed by each of the club properties. We'll leave that messiness to one side for now.

The club finds that part of its infrastructure is in terrible shape - partially because of decisions of past club executives; partially because of a recent earthquake. 

The club can choose to rapidly replace all of that infrastructure. That would be very expensive. But it would sharply reduce the chances of very bad outcomes where infrastructure blows out.

Or it could choose to pace itself in that infrastructure replacement. That will be much less costly. So much less costly that you could, at least in principle, compensate anyone who suffers from those infrastructure blowouts if it's really the infrastructure that's to blame. 

The latter could be a very reasonable deal. Behind the veil, none of us know which of our properties is sitting on top of a water network pipe that will collapse catastrophically and destroy our home. But we'd all be bankrupted if we tried replacing all the pipes in a giant hurry - it's just impossible. So we're each better off if we all agree to take a more cost-effective path on the infrastructure refresh while compensating any club member who draws the short straw.

The alternative, with no compensation, is more like the club members agreeing to play a giant game of Russian Roulette. We don't know which of us will draw the short straw, but we hope to heck it won't be us because whoever it is will be ruined. 

It could be that the numbers don't actually work out this way. But it seems a reasonable stylised example. 

And in any particular case of blowout, you'd want to be sure that the club member hadn't contributed to the failure through their own negligence. 

But if it were clear-cut, it shouldn't be a legal battle. It should just be compensation. 

I mean, if a private company accidentally drove a bulldozer through your house and wrecked it, nobody would think it reasonable to force you to go to court to get them to compensate you. Everyone would pillory the company. There would be boycotts. Some Vic Uni quasi-academic might call for the company's chief executive to be hanged. 

Here's The Post.

Wellington City Council is refusing to pay for repairs after one of its own burst water pipes triggered two landslides that has left two families facing bills of up to $600,000.

A council-commissioned geotechnical report found the Wadestown slips were “most probably” destabilised by a failed drinking water main, owned by the Wellington City Council and maintained by council-controlled organisation Wellington Water.

The 50mm pipe ruptured on August 4, saturating the embankment, and the slips forced both households to evacuate. Residents are still living in temporary accommodation.

I Am Not A Lawyer. 

But it seems likely that the homeowners could sue council for nuisance, and win, and have costs awarded against council, but the costs won't likely be anything like what would be needed to make them whole as compared to council just providing compensation. 

Council is not a good club. 

Cementing allocations

Recall that New Zealand issues industrial allocations under the ETS to avoid inefficient carbon leakage. 

Basically, if emissions are charged here but aren't charged abroad, and production shifts from here to there because of our charges, net emissions can increase rather than decrease.

That is obviously counterproductive. So industrial emitters facing competition from places with unpriced carbon get allocations of NZU. Done right, it maintains the incentive to reduce your emissions because you can sell off your surplus NZU. But it has to be done right.

The obvious way of doing it right would be to scale industrial allocations not by the NZ plant's emissions but by emission intensity overseas. If a tonne of cement abroad has x tonnes of associated CO2 emissions, then allocate x NZU per tonne produced here. Basically. Then, if emissions intensity abroad reduces, the plant here gets fewer NZU for its own production. It maintains an incentive to reduce your own emissions intensity, and avoids getting into spots where it would actually be carbon-efficient for production to shift to plants abroad that have lower emissions than plants here.

It gets messier if the import mix has stuff from places that are cleaner than here and stuff from places that are dirtier from here. If you scale to the average emissions intensity of the import mix (weighted by proportion of imports), there's still a potential problem. Suppose average intensity overseas drops and so allocations here drop. The NZ producer reduces production. But if that hole is filled by product from the dirtier plants overseas rather than the cleaner ones, you've wound up having inefficient leakage again. 

Carbon border adjustments are an alternative. But it gets messy with trade agreements. And you have to find a way of scaling the adjustment to the emissions intensity of the product. 

New Zealand's industrial allocations seem to have a bit of a problem in cement. 

Here's Energy News:

In a presentation to investors in June, Fletcher Building said a carbon border adjustment mechanism would level the playing field.

Currently, only goods produced in New Zealand face liability under the emissions trading scheme.

As an energy-intensive, trade-exposed emitter, Golden Bay Cement is eligible for an annual allocation of free carbon credits.

Data released last week shows that for last year's production, the company received 488,575 New Zealand Units, worth almost $27 million at the current spot price of about $55.

But the company says that a 2023 law change means that as the country's only cement manufacturer, it is now effectively being "rebaselined" every five years against its own emissions - which means that every time it cuts emissions it reduces the rate at which its free allocation is calculated.

"Significant investment in decarbonising local manufacturing is not viable without certainty a carbon border adjustment mechanism will be in place in the medium-term," it said in the presentation.

"Given regulatory settings, we have reviewed our capital plans for Golden Bay.

"The current investment plan retains flexibility to remain a domestic manufacturer or transition to an import model."

That just doesn't make sense.  

The government is reviewing the settings. Scaling to international emissions intensity would seem obvious. A carbon border-adjustment could also work but I have no clue whether it can be squared with trade agreements.

The annual allocation of free carbon credits to trade-exposed, energy-intensive emitters like Golden Bay Cement was last adjusted in 2023.

The company says that, in the absence of a CBAM or an equivalent mechanism, it would likely need to consider transitioning to an import model by the early 2030s.

That could result in a non-cash impairment and write-down of assets of up to about $165 million, as well as potential make-good and cash redundancy costs of up to $180 million.

For last year's production, Fletcher received 488,575 New Zealand Units, worth almost $27 million at the current spot price of about $55.

The company says it is engaging "productively" with the Government on the issue.

If cement produced abroad is more carbon intensive than cement produced here, then shifting to imports is the kind of carbon leakage that we ought to be avoiding.