How do you block Freenet? Seriously, how do you block it and not other services?

If Freenet is banned, the government can collect the address of every "opennet" Freenet node in a matter of hours. Then it's a question of finding the "darknet" nodes. A simple heuristic will probably catch most of them: recursively look for any address that has at least three long-lived, encrypted, two-way UDP streams to known or suspected Freenet nodes. The standard of proof at this stage is probable cause (or the French equivalent), rather than overwhelming evidence, so a heuristic approach is good enough. Wholesale traffic interception isn't needed: it's sufficient to monitor known or suspected nodes.

Now the government raids the owners of all the French nodes, confiscates their hard drives and decrypts their Freenet caches. There's bound to be some nasty stuff cached there on behalf of other nodes, even if the owners never uploaded or downloaded anything bad. The government charges the owners with "running a Freenet node" (so it's not necessary to prove what they uploaded or downlaoded) and makes a highly public announcement that it busted an extensive child porn / terrorist / neo-Nazi network thanks to the new anti-Freenet law. Then it waits for the handful of node operators it didn't catch to shut down their nodes and never say the word "Freenet" again.

Part of the problem here is that Freenet's design requires all nodes to belong to a single network, so if you have a heuristic for identifying Freenet traffic you can start from any node and 'unravel' the whole network. But to be fair to the Freenet designers, the alternative - lots of small, isolated darknets - isn't very appealing to users, because the only people you end up communicating with belong to the small intersection of "people I trust" and "privacy nuts". I'm a privacy nut who trusts his friends, and even for me that intersection isn't large enough to make for much of a conversation.

I've always wondered why people in this situation didn't build private networks based on protocols other than IP. A quick glance at /etc/protocols shows dozens of different protocols that can be carried by ethernet --- there must be something there that's sufficiently flexible to build a useful network out of but can't be carried by the Internet without protocol conversion.

It's even easier than that - just patch every host (and every router, unfortunately - but hey, Cisco, here's where you get your billion dollar contract) to set the version field of IP packets to something that's invalid on the internet - let's say 3 - and to reject all other versions. That's got to be, what, a ten line patch? After that you can use off-the-shelf software for all the higher protocol layers, but if someone accidentally connects the private network to the internet, no packets will pass between the two networks.

If the above happens, no one will want to invest in research, because they'd lose money, even if they "invented" the next IPod.

But Apple didn't invent the portable MP3 player. "Research, invent, commoditise, sell" is a plausible-sounding business plan, and I'm sure it sometimes works out that way, but much more commonly, companies learn from each other's mistakes and release competing products with small improvements. Apple realised people wanted an MP3 player that was slick rather than geeky-looking, so they repackaged it. That was their innovation. And it's a good thing - I'm not knocking that kind of incremental innovation. Patents harm that way of innovating, though, because the only companies that can play the game are those with big enough patent portfolios to deter their competitors from suing.

The portfolio problem applies to blue-sky innovation too. Imagine you invented the portable MP3 player from scratch in your garage and patented it. A year later, Apple releases the iPod. What are you going to do? If you sue, they can just pull some ridiculously broad patent out of their portfolio and counter-sue until you lose everything. The best you can do is to sell your patent to one of their competitors for use in their portfolio, and good luck getting a decent price when the buyer has all the lawyers.

There is one area where patents work more or less as expected, though, and that's drug development. Drug companies have a pretty good track record of throwing money at a problem until they get a usable drug (often usable for a different problem, admittedly), patenting the drug, and recouping their investment within the lifetime of the patent. Everything would be wonderful except for two catches: the money available to pay for a drug doesn't always match its social importance (the malaria problem), and the price of the drug while it's under patent may be too high for many of the people who need it (the HIV problem).

We've tried to patch the malaria problem through charitable funding of drug development, and the HIV problem through charitable subsidisation of drug prices, but both patches exacerbate the underlying problem by putting yet more patents and yet more money into the hands of the incumbent drug developers, meaning that next time we run into similar problems they'll be even more expensive to solve. The only solution I can think of is to create a public interest exception for patent licenses, coupled with public funding of socially important research, because the private money will move to areas that aren't covered by the public interest exception. But that sounds too much like dirty commie talk for a lot of people's liking. ;-)

The idea of libertarian (small 'l') thought is simplicity itself.

That alone should make you skeptical.

Consenting adults should be free to do whatever they please with their property and their own body and should be free to believe whatever they want. They should be able to exercise those freedoms whether or not someone else doesn't like it; anyone who doesn't like their actions is free to provide a counter-example in the form of how they deal with their own body, property, and beliefs.

The selfish asshats are the ones who would use the force of law to tell you what you may not do with your own body or your own property.

Not all attempts to restrict what people do with their property are motivated by selfishness. Example: pollution.

They typically do this out of some kind of Puritannical desire to enforce their morality on others.

That's a very limited and, if I may say so, typically American view of the world. Not everyone's a libertarian or a Puritan. Some parts of the world still remember feudalism (some are still enduring it), and are consequently skeptical of the idea that property rights alone are sufficient to ensure a free and just society.

You like simplicity, so here's a simple model of how libertarianism devolves into something less pretty. Assume we have a population of individuals who all start out with equal wealth, and who are free to invest that wealth in enterprises, some of which succeed, leading to a multiplicative increase in the invested wealth, and some of which fail, leading to a multiplicative decrease. Now, if there's an arbitrarily small random factor involved in the success or failure of enterprises, the distribution of wealth will over time approach the lognormal distribution. In short, a few people will become very rich, and most people will become relatively poor, even if they're all equally skilled investors. (I say "relatively poor" in anticipation of some hand-waving argument about how removing the shackles of taxation will lead to a jump in productivity. My argument doesn't depend on whether or not that's true.)

Now we have two problems. (1) The very rich people may be able to use their wealth to distort the perfect libertarian free market, entrenching their advantage. A shocking idea, I know, but please remember that this is just a thought experiment. (2) The poor people may be forced by short-term needs such as food and shelter to enter contracts that are not in their long-term interest. As long as any employer exists who offers them only the means of survival in return for their labour, the market will drive any more generous employer out of business. Having earned nothing but the means of survival, the poor will invest nothing and leave nothing to their children. So the underclass, once created, will persist.

We are now quite far from a libertarian utopia in which everyone frolics freely in his private meadow and shits in the collective river - we have an entrenched ruling class, a political process corrupted by money and an exploited underclass trapped in a hand-to-mouth existence. Sound familiar? Yet we've been brought here by nothing but property rights and free markets.

The people who want to be left alone by them so long as they don't violate anyone else's freedoms are not selfish in the slightest. They are reasonable.

It's possible to be both.

I think this can be answered by looking at how the question is framed. The question doesn't ask why politically radical people are likely to be engineers. It asks why that subset of politically radical people who decide that the best solution to political problems is through the direct application of technology are likely to be engineers. Well guess what? That subset of any group that tries to solve every problem by applying technology probably contains a lot of engineers.

It's unfortunate for the world that most problems can't be solved that way. But that doesn't stop a lot of technically adept people from trying.

It simply isn't practical to keep a bunch of pigeons for every destination you would want to go.

Sure, but nor is it practical to run a wire to every destination - that's why we have routers. Same principle here: each pigeon travels one hop, packets are removed from the pigeon and assigned to outgoing pigeons for the next hop, meanwhile the pigeon goes back the other way carrying data in the opposite direction. The 'router' is a wooden box with a solar-powered computer that wirelessly updates the pigeons' memory cards and gives them a treat every round-trip and an extra one at Christmas.

While the situation described in TFA is clearly a stunt, there are actually serious proposals to use vehicle- or animal-based networks for extremely remote areas. NICTA in Australia is working with IIT to develop a delay-tolerant network for rural India, and there's work out there that uses devices attached to animals to collect data from remote environmental sensors. The only limits I can think of on the use of pigeons for such purposes would be the durability of large containers of pigeon treats and the lifetime of the pigeons. Judging by the pigeons here in London, you could addict them to nicotine to solve the former problem, though that might exacerbate the latter...

Thanks, that's a useful distinction and I'll bear it in mind. However, I wasn't arguing with either a theist or a deist - I was arguing with an atheist. The argument that there's some conception of God that doesn't require any more assumptions than atheism and that isn't contradicted by the evidence is still valid for refuting atheism, even if that conception of God doesn't happen to be the one that's popular among religious people.

My point is that there's no logical reason to prefer either model. The model with a God in it requires a single assumption (assume there's a God, who has no cause, and who causes everything else), as does the model without a God in it (assume there's a Big Bang, which has no cause, and which causes everything else).

To play the devil's advocate (or in this case, God's), the problem with your argument is that qualifiers such as "likely" and "unlikely" only make sense for events that are in principle repeatable, even if only in thought experiments. It's legitimate to argue that a china tea set orbiting the sun is "unlikely" because we have a good understanding of how china tea sets come to exist, so we can conduct a thought experiment in which the tea set construction process is run a million times, and no tea sets are produced in deep space. So we call such a tea set "unlikely".

When it comes to God, on the other hand, we have no such generative model. If you assume the Big Bang emerging from nothing then you have a model from that point on, but equally if you assume God creating the Big Bang then you have a model from that point on, and both models produce identical results from different assumptions. There's no basis on which to say "likely" or "unlikely" - all we can say is "under this assumption" or "under that assumption".

On behalf of the British technology industry, it's my privilege to issue this response: Mr Sharkey, fuck yourself in the eye. After thirty years of smear campaigns and righteous hysteria you've finally realised that you can't make money without us, and now you want to be friends? Sorry old man, but it's too little, too late. Everybody knows your house is on fire and we're not going to help you put it out. All we wanted was a share of the groupies and the coke, Feargal. Was that too much to ask? But supplies are drying up, standards are dropping and in the meantime we've invented Craigslist. What do you have left to offer us? Box sets? Get the fuck off my doorstep, Sharkey.

Semi?

Apple users consider the platform semi-open because you can look through the glass at the front and see your apps.

Potable water, good infrastructure, lots of various industry, a very good, middle-class standard of living, and less-than-average corruption in their police force.

This makes me wonder whether the scanners should be installed in Ciudad Juarez instead - because if you want to know whether technology like this would be dangerous in the hands of a corrupt government, that's the place to find out.

From the Haystack FAQ we can surmise that you run some small client program and point your browser at that.

That makes sense, but if the local proxy's going to encode your request for whyweprotest.net in a stream of requests for weather.com, there needs to be a proxy on the other side of the firewall that intercepts those requests and extracts the request for whyweprotest.net.

So here's the problem: how does your local proxy get the address of the remote proxy, without the Iranian secret police being able to run their own copy of Haystack, get the address of the remote proxy, and block it (or, worse, use the firewall to record all the addresses that connect to it)?

In the literature, this goes by the imaginative name of the "proxy discovery problem". Solutions include privately sharing proxy addresses with trusted friends, distributing addresses by email, requiring clients to solve computational puzzles, requiring users to solve captchas, and using the structure of social networks to limit the number of proxies an attacker can discover.

Which method does Haystack use? We don't know, because Austin Heap hasn't published any technical details of the design, or submitted it for review by a trusted party like the EFF, despite calls for him to do so.

The gold standard here is Tor: all the code is open source, there are detailed design documents, they submit their designs for peer review by the security community, and they have an excellent track record of fixing the weaknesses that are found. Austin Heap needs to learn a lesson from them, because just saying "It's ok, we encipher everything" doesn't cut it in 2010.