Nuclear weapons are hard to engineer [1]. Assembly requires a lot of expensive engineering resources and access to raw materials. By the time you've got the infrastructure together you're no longer an average terrorist, but more like a small state or maybe a corporation, and vulnerable to espionage or military disruption by other players. Terrorists are interested in using terror to achive

BUT the US government ran a study in the early sixties, called the "Nth Country Experiment" [2], that took a few recent physics PHDs and paid them to see if they could develop a design for a nuclear weapon. They surprised everyone by designing a credible implosion weapon. So it can be done. And the amount of available information and technology is better than fifty years ago.

[1] http://fas.org/nuke/intro/nuke/design.htm

[2] http://en.wikipedia.org/wiki/Nth_Country_Experiment

They are hard to build because once the chain raction starts in the nuclear material, the resulting reaction tends to break up and disperse the nuclear material before the chain reaction progresses very far. It's called a fizzle. Fission bombs are designed with a wrap-around concave shaped charge of conventional explosive around the nuclear material. This is detonated as the critical mass of nuclear material is slammed together, and the compression wave from the conventional explosion holds the fissioning mass of nuclear material together long enough for the chain reaction explosion to complete. The engineering for this and the precise timing and controll of the detonation sequence is extremely hard.

My understanding is that the engineering for this can be done by someone who has studied the stuff, and uses publicly available materials. He or she does need to be competent but does not need to be Einstein. See John Aristotle Phillips. [0]

The main challenge is simply the industrial process required to collect the fissionable material. That is why Iranians have centrifuges and why Stuxnet attacked those centrifuges. [1]

[0] http://en.wikipedia.org/wiki/John_Aristotle_Phillips

[1] http://en.wikipedia.org/wiki/Stuxnet

For a fission bomb yes, not s fusion (thermonuclear/hydrogen) bomb.

Fire bombing can (and has done) more damage to a city than a fission bomb.

But a terrorist can't fire bomb a city, but they can potentially build a fusion bomb.

A terrorist can't really build a fusion bomb, because you need a government to support the infrastructure. But the fear is not entirely unjustified because a government would be able to give the bomb to a terrorist who then uses it. Or perhaps a government employee might steal it (for bribes) and give to a terrorist. So I can imagine terrorist having a small fission bomb. But I cannot imagine an arrangement where a terrorist would acquire a fleet of B-29's that could be used to fire-bomb a city.

To build it from scratch yes, it's possible to build a primitive bomb from stolen refined plutonium. After the collapse of the soviet union, a lot of it went missing and there a dozen cases of people getting caught trying to sell it on the black market (and those are just the people that've been caught.)

Mass arson is possible, though not on the scale of firebombing. The US in WWII developed the idea of strapping incendiary bombs on bats and releasing them in a city, where they would roost in random buildings and start hundreds of small fires.

The ones in 1945 were prototypes with everything that implies. And fundamentally we don't build or use a lot of them, so they're slow to improve (think space rockets, not microchips).

That said they're not that hard, and they get easier every day. At an anti-proliferation conference a few years ago Iran argued that any world-class university physics department probably has the capacity to build a nuclear weapon, and commentators agreed. Eventually it'll reach the point where a high school AP student could do it.

Getting the fissionables to make a bomb requires major industrial effort. And an "average terrorist" would only be able to build a gun assembly bomb like Little Boy, which requires a lot of uranium 235.

If we think fission. If we think fusion - you could build something nasty with tritium and deuterium with some depleted uranium as neutron deflectors ... but the ignition will be bitch ...

Edit: replaced neutrinos with neutrons

I think you mean neutrons - neutrino deflectors would be quite interesting...

They are quite easy to build. Finding pure enough fissile material is the hard part.

I've always wondered about the Chicago Pile: https://en.wikipedia.org/wiki/Chicago_Pile-1

It's a very primitive device compared to a nuclear weapon, but it could have created an runaway nuclear reaction and do a ton of damage.