Hacker News new | ask | show | jobs
by benjaminsuch 1617 days ago
As a german, can't agree more. From what I get talking to people is, that they have an irrational fear of a nuclear holocaust and nuclear waste.
3 comments

Why «irrational»? It just to add color to your message, or you truly believe that radiation contamination is harmless?
People are bad at statistics.

The probability of a catastrophic event in a modern nuclear plant is vanishingly low. Even the overworked ancient plants from the 60's and 70's we are currently running are exceedingly safe.

People will rather take constant death from coal power than risk a low-probability event.

Yeah, but we have multiple catastrophes already, despite the vanishingly small number of nuclear plants. If we want to increase the number of nuclear stations by 10x, we need to make them 10x safer just to keep the number of catastrophes at the same 1/30y level. Do you have an idea how to make a nuclear plant 10x safer comparing to today?
It looks like the all time death toll of nuclear power so far is <20,000 total. Meanwhile coal plants kill 30,000 per year in Europe.

https://en.m.wikipedia.org/wiki/List_of_nuclear_and_radiatio...

https://www.independent.co.uk/climate-change/news/air-pollut...

It's because we ran away from contaminated sites.
> Do you have an idea how to make a nuclear plant 10x safer comparing to today?

Yes. All the reasons for the catastrophes are well known.

There are 1) known reasons for known catastrophes, so we can protect against them, 2) unknown reasons for known catastrophes, so we need to make a guess, and 3) unknown reasons for unknown catastrophes, which are not happened yet, including state sponsored attack on a nuclear plant. We cannot be prepared for unknown unknowns, so we need to plan for the worst case scenario. The worst-case scenario for nuclear plant is continent scale Red Forest (about 1M Chornobyl's).

Do you know how to reduce continent scale threat to just the size of a nation or a town?

Nuclear fusion or LENR can do that, because of small amount of radioactive materials and no positive coefficient by design, but how you can do that for massive fission?

We know how to build safer nuclear now than we did 60 years ago when the current gen were mostly built.

Current generation nuclear plants default to off, they need active operation to stay on. If something goes wrong, they automatically, without any intervention by anyone, go offline.

They are also designed to withstand collisions from aircraft and multiple other attack vectors: https://www-pub.iaea.org/MTCD/Publications/PDF/Pub1159_web.p...

Not sure I would describe a fear of nuclear holocaust as "irrational" - had the Cold War gone hot Germany would have been very heavily targetted.

Mind you, I'm not saying that justifies the current fear of nuclear power but then again it doesn't really surprise me. Maybe it's just a lingering fear of anything with "nuclear" in the name.

We live at the dawn of cyberterrorism, I'm glad we are stepping out of nuclear now.
Not only cyber. A nuclear power plant makes an excellent target for terrorism or war.

Even if the attack is completely unsuccessful it would cause panic.

Furthermore, highly centralized electricity generation makes the distribution network very vulnerable as well.

It's not impossible to purpose-build hardened, incompatible, read only systems that can submit telemetry to the outside world while only providing actual control on-site (or via restricted channels). Stuxnet wouldn't have happened (or would have been a very rare event) if they built their system this way.
Stuxnet happened despite being air gaped. Regardless, I am confident you can place physical safeguards that could not lead to nuclear emissions even in the event of loss of control over the computer systems.
It happened despite being air-gapped, because they used general purpose hardware and software. If their systems were built on purposely incompatible hardware and software (as I proposed) and could mainly communicate using a serial console, the attack surface would be much, much lower, and the attacks would be much, much harder.
Having worked on for a short stint with some power plant control systems, I can say that, at least the systems I worked with, were quite niche. The actual control was happening on these racks that ran a VxWorks OS on some Motorola, I think they were, MCU's. Despite this, the systems were interfaced with some Windows machines that did supervision. When they were operating, they had redundancies, and were quite locked down. Of course, at that time, I was a noob and did not understand _everything_ that was going on in there.

Actually, now that I think of it, the WDPF system it was derivated off was used in some nuclear power plants as well.

Regardless, what I wanted to say was... being obscure, while it makes things mildly harder for skiddies is not a big deal for state actors or more resourceful attackers. The Stuxnet was highly targeted and they got access to specific vulnerabilities in the Siemens DCS systems that were running there. Just having exotic systems is no guarantee. I agree, obscurity is a layer of defense in depth, but no guarantee. Surely you don't suggest they use a new purpose built HW for each control system design. Also, control systems DO need to have their SW updated as well. It's obvious you can't make it hard read only. You do have physical lockout mechanisms for this though.

Here's a radical idea: Maybe your potentially dangerous industrial machinery does not need to be directly connected to the internet?
Hopefully the few computers needed do not use generalist OSes and aren't connected to the Internet either ?
IIRC, some nuclear readout panels were accessible over the public internet using badly secured VNC, so I think that's not the case.
I assume you never worked in security.
Should we be concerned then about nuclear weapons to be hacked too ?
Not to the point of detonating them - they have a lot of physical safeguards.
And why you need any electric central connected to internet ?