There's a line in a report from last year that I keep thinking about. It said the problem isn't that governments are saying no. It's that they're saying nothing at all.
Which is worse, in a way. A "no" you can fight.
Right. And that's roughly what Daniel is circling in this prompt. He's asking about freedom of information legislation — the statutory right to request records from public authorities — and specifically the gap between what these laws promise and what they actually deliver. He wants the whole arc, the history from Sweden in 1766 through the big waves of adoption, how many countries have some version of this now, and then the comparative meat. Who gets to request, which bodies are covered, what exemptions governments can reach for, whether you have to justify your interest, what it costs, how you appeal a refusal. Israel's 1998 law sits at the center of that comparison. And then the question that actually matters: is there a gold standard, or is legal strength a poor predictor of whether you ever get the document.
That last framing is the right one, because it inverts what everybody assumes. The whole transparency movement has been built on the premise that passing the law is the hard part. And the evidence is pretty clear now that it isn't.
So let's take it in order.
Freedom of information legislation, at its core, is a statutory right for the public to request records held by public authorities. The principle underneath it is almost embarrassingly simple. Information that government holds, it holds on behalf of the governed. So the default should be access, and any exception has to be narrowly defined and justified. That's the theory. The translation from that principle into a statute, and then from a statute into somebody actually answering your request, is where the whole story lives.
And the comparative interest is really in that second and third step.
Any country can write a beautiful law. The interesting question is what happens when you file.
Start with 1766, because this is much older than people assume.
Sweden's Freedom of the Press Act of 1766 is the world's first freedom of information law. It comes out of Enlightenment thought, and there's a specific figure behind it — Peter Forsskål, who'd been arguing for press freedom and access to official documents. The framing matters, because the right to know predates the modern democratic state by a long way. It's not a twentieth-century invention. It's an eighteenth-century one that then went quiet.
How quiet?
Finland is next in 1951. Almost two centuries later. That's the scale of the pause. The United States passes its Freedom of Information Act in 1966, effective the following year. Australia, New Zealand, and Canada come in around 1982 and 1983. So up to that point you're looking at a small club of mostly Western democracies, and the whole thing has been running for more than two hundred years with maybe a handful of members.
Which tells you adoption isn't driven by the idea being obvious. The idea was obvious in 1766. Something else has to happen.
And what happens is 1990. A dozen states had FOI laws in 1990. By 2011 it's more than eighty-five. By 2014 you're at roughly a hundred, with Paraguay generally counted as the hundredth. The World Bank's own framing is growth from nineteen mostly Western democracies in 1995 to some hundred today. The Centre for Law and Democracy now counts about a hundred and forty countries with statutory right to information laws. That's roughly four-fifths of the world's population living under one.
A hundred and forty countries in about thirty-five years.
The drivers are pretty legible. The end of the Cold War. Democratization across Eastern Europe and Latin America. Civil society organizations pushing it as a governance marker — the idea being that a state that lets you look at its paperwork is signaling it's a legitimate state. There's a norm cascade quality to it. Countries adopt because their neighbors adopted.
And the holdouts?
Egypt, Myanmar, the Democratic Republic of Congo. It's not universal. A hundred and forty out of roughly two hundred countries. Which is worth sitting with for a second, because the number sounds overwhelming until you notice that about half the world's countries still don't have one.
So there's the spread. But as you said at the top, the headline number tells us almost nothing about what any of these laws actually do.
This is where the variation becomes the story. Take the dimension of who can request. Some jurisdictions restrict it to citizens or residents. Israel is one of those — the law grants the right to every Israeli citizen or resident, and that's a real limitation. Other countries open it to anyone, including foreign nationals, on the theory that the information belongs to the public rather than to a defined political community. That single choice changes who can hold a government accountable.
Coverage next. Which bodies?
Israel's definition is actually expansive on paper. It covers government ministries, the President's Office, the Knesset, the State Comptroller, the courts, local authorities, government-owned companies, and then a catch-all — any other agency fulfilling a public function.
Exemptions are where the laws really diverge.
And Israel is a useful illustration because it splits them into two categories. Section 9A is mandatory non-disclosure. If the material touches state security, foreign relations, public safety, privacy, or is barred by another law, the authority must refuse. Section 9B is discretionary. Internal discussions, policy still being formulated, negotiations in progress, commercial secrets, law-enforcement methods, internal management. Those are cases where the authority may refuse.
That distinction between must and may is where a lot of the practical weight sits. A discretionary exemption is a judgment call, and judgment calls are where culture enters the building.
Then there's section 14, which is the interesting one for comparison, because it's an outright carve-out. The intelligence agencies, Shin Bet, Mossad, the Atomic Energy Commission, and police intelligence are just outside the regime. No balancing test, no request-and-appeal. Those bodies are exempt as institutions.
Some countries do that. Others put those functions inside the regime but give them a strong exemption. Different structures, similar practical result, but the second one at least leaves a paper trail of refusals.
Standing is the next dimension, and here Israel actually scores well. Section 7A explicitly says you don't have to state a reason for your request. You don't need to demonstrate a legitimate interest. That's a meaningful provision, because requiring someone to justify why they want to know is one of the quietest ways to choke a transparency regime.
Costs and process?
Fees are set by regulation under section 18. Response is thirty days, extendable by another thirty, and in complex cases up to sixty more on top of that under section 7. Refusals go to the Administrative Court within thirty days under section 17, and the courts can order disclosure where the public interest outweighs the exemptions. On paper, that's a functional architecture. Request, response window, appeal route, judicial balancing test.
Which is why the RTI Rating is such a useful instrument for this comparison, and also such a trap.
The Global Right to Information Rating scores laws out of a hundred and fifty, across sixty-one indicators in seven categories. The top of that table is surprising. Afghanistan, at a hundred and thirty-nine, assessed in 2014. Mexico at a hundred and thirty-six. Serbia at a hundred and thirty-five. Then Sri Lanka at a hundred and thirty-one, Gambia at a hundred and twenty-eight, Slovenia, Albania, Croatia, India, all clustered in the hundred-and-twenty range.
And the established democracies?
Sweden, the country that invented the concept, scores a hundred and one. The UK is ninety-two. The United States, the law that half the world copied, is eighty-two. Israel is eighty. And then the bottom of the table looks nothing like the top. Germany sits at fifty-five, Saudi Arabia at fifty-three, Qatar at forty-three, Belarus at thirty-eight, Liechtenstein at thirty-seven, Palau at thirty-three.
Germany at fifty-five is the one that stops you cold.
It's a federal system, so the constitutional allocation of competence complicates a national right to information, but the score reflects a weaker statutory framework than Afghanistan's. That's the inversion. The strongest laws on paper are overwhelmingly in newer democracies, and the established ones sit in the middle.
Because newer democracies are writing these from scratch after a rupture, and established ones are amending arrangements that predate the whole concept.
That's my read. A country drafting a constitution in 2003 can bake in the international best practice. A country with a working administrative tradition has to negotiate with it. And then there's the caveat that the rating itself is explicit about — it measures the legal framework only. Not implementation.
Which is the sentence that should be printed on top of that whole table. Because a score of a hundred and thirty-nine tells you the law is excellent. It tells you nothing about whether anyone answers the phone.
So what happens when you actually file? That's what the Centre for Law and Democracy and UNESCO tested. They launched this on the International Day for Universal Access to Information, in September 2025. The design is what makes it credible. They filed the same two requests in as many of the roughly hundred and forty countries with right to information laws as they could. One request on COVID vaccine purchases, one on the most costly environmental incidents.
Same two questions, every country.
Same two questions. They got results from a hundred and forty-six requests across seventy-six countries. And the findings are the most useful thing in this entire field.
Give me the shape of it.
Thirty-eight percent were what they call mute refusals. No substantive reply at all. Not a denial, not an exemption claim, not a partial release. Nothing. Forty-two percent produced full disclosure. Fifty-four percent produced some or all of the information, so there's overlap in the categories. Receipts were issued in only fifty-seven percent of cases. Average response time was twenty-two days, with just over a quarter arriving after thirty days. And then the number that reframes everything: only two percent were explicit denials.
So the exemptions you spent five minutes cataloguing are almost never the thing that stops you.
The report's own line is that the problem isn't exceptions, it's not answering.
What's the worst case they found?
One authority extended its own deadline by a hundred and fifty days. And the fee barrier that dominates so much conversation about these laws turned out to be almost absent in practice — fees were charged in only three countries.
Here's what I keep circling on, though. The mute refusal isn't just a slow version of a denial. It's a different kind of act.
Say more.
Every accountability architecture you described — the thirty-day clock, the appeal to the Administrative Court, the judicial balancing test — assumes a decision exists. You appeal a refusal. You challenge an exemption claim. You argue the public interest outweighs the harm. If the answer is silence, there's nothing to appeal. The citizen has no standing to go anywhere, because the authority hasn't done anything you can point at. It bypasses the entire structure without ever having to justify itself.
Which is why it's the rational strategy if you want to avoid disclosure. Denying a request creates a record, invites a legal challenge, and forces you to articulate a reason that a court can test. Doing nothing produces none of that.
The law's remedy is a decision, and the evasion is the absence of one.
That's the mechanism exactly. And it's cheap. It requires no legal reasoning, no supervisor's sign-off, no exposure to a balancing test. It just requires an office that doesn't prioritize the request, and a system where nobody's career depends on the outcome.
And that gets us to Israel, because that's a case study in exactly this.
Israel is the right case, because on paper it should be near the top. The Freedom of Information Law passes on the nineteenth of May, 1998, takes effect in May 1999, and officials at the time framed it as a transparency revolution. And the statute has real strengths. No reason required, section 7A. A thirty-day clock. Broad coverage. A court you can go to. If you were scoring purely the text, you'd expect it to perform.
And the Israel Democracy Institute's study from 2008 is where the picture fractures.
Their findings were minimal invocation of the act, a general lack of awareness, conflicting approaches to its implementation, and a lack of appropriate resources for enforcement. And then the line that I think is the most important sentence anybody's written about this law. Officials, they found, assume that government bodies own the information and that the citizen must justify acquiring it from them.
That's the default reversed, isn't it.
That's the lot. The statute says the information is public and the citizen needs no reason. The culture says the information is the ministry's property and the citizen is a supplicant. And when those two things are in tension, the culture wins, because the culture is the person reading your request.
There was an earlier study too, 2005, on what happens to applications.
About twenty percent were never dealt with at all because the applicant didn't pay the fee. Another twenty percent got negative responses. And then fewer than ten percent of negative replies ever led to court. So you have a law with an appeal mechanism that almost nobody uses, and a fee barrier that quietly kills a fifth of requests before they're even considered.
A fee that functions as a filter, then.
The IDI study called the application fee a significant impediment. Which is worth noting against the global finding that fees were almost never charged in the stress test. Same instrument, wildly different effect depending on how it's set.
Now the audit.
The State Comptroller's July 2024 audit is the most damning document in this whole story, and it's damning in a very specific way. The law applies to two thousand and fourteen public authorities. The Government Freedom of Information Unit — the body that's supposed to regulate this — has powers covering ninety-one of them. Thirty-one ministries and sixty auxiliary units. That's five percent.
So ninety-five percent of covered authorities have no regulator.
One thousand nine hundred and twenty-three authorities, including two hundred and fifty-seven local authorities, with nobody whose job it is to make sure the law gets implemented. The Comptroller's own line is that despite twenty-five years since the law was enacted and eleven years since the unit was established, most public authorities lack a regulatory body to ensure effective implementation.
What does that look like in the numbers on requests?
In 2022, Israeli local authorities received eight thousand eight hundred and forty-seven requests, against thirteen thousand and twenty-eight to central government. So local government is taking about forty percent of the load. And then the petitions filed tell you what happens after — two hundred and fifty-eight petitions against local authorities, two hundred and three against ministries. Locals are getting sued disproportionately relative to their request volume.
And winning?
In 2021, the courts accepted a hundred and thirty-eight petitions out of a hundred and fifty-nine filed against local authorities, and imposed something like three hundred and seventy-eight thousand shekels in obligations. So when citizens do push it to court against a municipality, they mostly win.
Which tells you the refusals weren't legally defensible in the first place. They were just the default behavior of an office with nobody watching.
And the staffing tells you why. Information officers in local authorities often hold the role as an unpaid additional post with no fixed term. It's somebody's side duty. And some authorities reportedly require a court order even for routine photographic data.
A court order for photographs. What would those be, for planning or for municipal records?
Routine imagery that should be within the scope of the law. And there's another pattern in the audit worth naming. About forty percent of Arab-sector authorities in Israel — thirty-four out of eighty-five — reported no FOI requests at all across 2019 to 2022.
None in four years.
Which could mean nobody asked, and it could mean the process isn't functioning in a way that produces records of requests. Those are very different problems with the same number attached.
The Meida op-ed from 2011 put it in one line, I think. It said the country's well-known culture of secrecy and bureaucratic exhaustion can be seen at full power.
Which is a better description of the mechanism than any of the legal analysis, honestly. The law is modern. The statute is fine. What's missing is a functioning body that treats a request as a task rather than an imposition.
So let's catalogue the failure modes properly, because Israel is running maybe four of them at once.
Expansive exemptions is one, though as we said, it's less common than people assume. Administrative delay another. Non-response, which is the dominant one globally. Excessive redactions. Poor recordkeeping — you can't release a document that was never properly filed. Under-resourced information officers. And weak independent oversight, which is the one Israel fails hardest on.
And there's an inversion in the paper-versus-practice data that's worth stating flatly.
The countries scoring highest on legal strength — Afghanistan, Mexico, Serbia — are not the countries that perform best when you actually file a request. And the countries with functional, high-compliance transparency — the ones where the culture supports the law — are often sitting mid-table on legal scoring. Legal strength predicts remarkably little about responsiveness.
No country in that stress test achieved full compliance across both requests.
None. A hundred and forty-six requests in seventy-six countries, and every single jurisdiction failed at least one.
So the gold-standard question Daniel asks — is there one?
Not in the sense he's asking. There's no jurisdiction that's universally recognized as combining the strongest law with the strongest practice. There are countries that do one extremely well. Afghanistan wrote arguably the best statute in the world and then the state collapsed. Mexico built a serious constitutional right to information and then faced the question of whether enforcement would survive political change. Serbia scored a hundred and thirty-five and nobody points at Serbia as the model.
So what actually distinguishes a meaningful right to know?
Independent enforcement with real teeth is the first thing. An information commissioner with the power to order disclosure, not just advise. Then proactive publication regimes — the default should be that routine categories of information get published rather than staying dark until somebody requests them. An adequately resourced information officer in every covered body, professionally independent, whose job is the law and not whatever else lands on their desk. And underneath all of it, a culture where officials treat information as public unless there's a specific reason to hold it back.
The Austria reform is useful here, because it shows even the settled cases can move.
Austria's new constitutional freedom of information provision and its Freedom of Information Act took effect on the first of September, 2025. And what it replaced was the constitutional principle of official secrecy.
The default flipped.
The default flipped, at the constitutional level. Which is the point. For a hundred and forty countries the interesting question isn't whether to have the right, it's what the default is when nobody's looking. And Austria, which is not a young democracy, spent years deciding that official secrecy should stop being the constitutional baseline.
That's the whole episode in one reform, isn't it.
The word's wrong. Everyone in this building says transparency. It's a bad word.
Go on then.
Transparency is a property of glass. You want glass, you buy glass. What you're describing is a filing clerk. There's a difference between a window and a man in an office who answers the phone. That's what you've been circling for twenty minutes.
I'll take that.
I filed a request with the Jerusalem municipality. February. Form seven, the paper one, not the online, because the online one doesn't give you a stamped copy. You want the stamp.
You wanted the stamp.
You want the stamp. Otherwise it's your word against theirs on the date. Anyway. I asked for the maintenance log on the lift in my building. Twelve years of it. Two questions: how often it was serviced, and who signed off. That's it.
What came back?
A letter. Nine weeks later. Saying the request was very broad and could I narrow it. So I wrote back and said, fine, only the entries where the inspector noted a fault. I sent it the same way, with the stamp. Nothing for two months. Then a second letter saying the file had been archived and they'd need to retrieve it, and there might be a retrieval cost.
Did they say how much?
Eighty shekels. I paid it at the post office and sent them the receipt. That was in June. Then in September I got a phone call from a woman at the municipality who said — and I want you to hear this properly — she said the request was being handled by the previous tenant of my flat.
The previous tenant.
Apparently there's a man who has handled lift-log requests for that building since before I moved in. He lives in Beit Shemesh now. She gave me his mobile number and asked me to call him.
And did you?
I called him. He was very nice. He said he had the logs but they were in a storage unit in Ramle and his brother had the key, and the brother was in Cyprus for the season, and he'd sort it when he got back.
So at this point there's a man in Beit Shemesh, a storage unit in Ramle, and a brother in Cyprus.
And a lift that's still running. I ride it every day. It makes a noise on the fourth floor.
What did you actually want to know?
Whether anyone signed off on it, or whether they just kept running it.
And the woman at the municipality, the one who called — what's her name?
Irit. She was helpful. That's the thing, nobody's rude about it. Everyone's pleasant. It just doesn't happen.
So the document exists, it's in a storage unit, and the person holding the key is on an island.
The brother's name is Shuki. He comes back in November. I'm not chasing it.
You paid eighty shekels for that.
I paid eighty shekels to find out who handles the paperwork. That's worth eighty shekels.
My aunt did the same thing years ago. Different office, different form, same outcome. It isn't the law that stops you. It's that the man who has the file doesn't work there anymore.
Which is the finding, actually. The strong law meets the storage unit in Ramle. The statute gives you a right to a document. It can't give you the man with the key.
And note what happened in your story — there was no denial at any point. Nobody invoked an exemption. Irit didn't refuse you. You got two letters and a phone call and a suggestion to ring a man in Beit Shemesh.
Silent, but politely.
The thirty-eight percent in miniature.
Which is the thread to pull on, because if the dominant way it goes wrong is silence rather than refusal, then the next round of reform probably isn't about the text of the exemptions.
The exemptions are already narrow in most of the hundred and forty countries. The stress test showed only two percent of responses were explicit denials. The statutes are mostly fine. It's the answering that's broken.
So what would you actually change?
The enforcement body is the first lever. Israel's case makes it vivid — five percent of covered authorities have a regulator. If you extend the unit's powers to the other ninety-five percent, you've changed the incentives for information officers overnight. Suddenly there's somebody who can be embarrassed, and an information officer who doesn't answer becomes a problem for their authority rather than a non-event.
Proactive publication is the second leaver. If the routine categories are already published, the request volume drops and the remaining requests are the ones that actually need a judgment call.
And the third is the culture, which is the hardest, because you can't legislate it. But you can build it, slowly, by making the default visible. The IDI finding is that officials assume the state owns the information. That assumption doesn't change because a statute says otherwise. It changes when the assumption starts costing somebody something.
And there's the structural point that the stress test can't see, which is that a mute refusal is a decision not to decide. It's an act of omission that functions as an act. If you want to fix the system, you have to make the omission visible. A registry of requests, published deadlines, an automatic escalation when the clock runs out.
A decision after thirty days, by default, in favour of disclosure unless the authority can show why not. That would flip the incentive completely and it would probably be the single most effective reform available.
The one place that's actually done something like this?
Austria is the recent example, though it's a constitutional default flip rather than a procedural one. Mexico built a independent enforcement mechanism and it's the reason it scores near the top. But nobody has a system where non-response is automatically treated as a decision, and that's the gap.
Because the answer is silence and the remedy is a decision.
Which is why the gold-standard question doesn't have an answer yet. Not because nobody's tried hard enough. Because everyone's still designing around the assumption that the authority will respond.
The other thing you're both missing. Nobody asks for the thing you want directly. You ask for the file, they give you the file, and the file is not the answer. I got the lift logs eventually — Shuki came back, I have them now. Eleven years of service entries. Every single one signed. The lift was serviced exactly as often as it should have been.
And the noise on the fourth floor.
Nobody logged the noise.
