Thanks for that link, that is the best source of facts without unnecessary blabber that I've seen so far.
The IAEA site also confirms the radiation level that I heard in various places: As reported earlier, a 400 millisieverts (mSv) per hour radiation dose observed at Fukushima Daiichi occurred between Units 3 and 4.
Just so that we're clear: 400mSv/h is a serious amount of radiation. A short-term exposure of 1Sv is a ballpark value for acute radiation sickness, and you'd get that in 2.5h.
Like the IAEA says, that's a point value right in the plant, but if the workers were exposed to that sort of radiation level, saying that "no one will die from radiation here" is not justifiable.
I find it extremely interesting (and by that I mean suspicious) that the Siemens astroturf post "Why I am not worried about Japan's Nuclear Reactors" has been migrated from a quote-unquote "personal" WordPress blog onto a quasi-official MIT website that was suspiciously purchased Sunday evening. Not only that, the text of the post has been so dramatically altered that it no longer says anything about not being worried, or how the concrete containment will prevent anything from possibly going wrong.
This entire thing smells like a huge PR stunt from a company (Siemens) that recently donated money to MIT and has a huge economic interest in preventing any bad publicity about nuclear power from appearing in the media.
I actually find it shameful that HN can be so easily duped by this obviously doctored information.
If you have reason to believe that the information is doctored, it would be a better idea to explain why. Attacking the messenger is fine for scoring points in a debate, but it isn't going to bring anything valuable to the discussion. We should be thinking about the people who are confused and scared right now, not Siemens and their bottom line.
Because the guy made an incorrect assumption before he had the proper data, or something else? Most people did that, it's obviously not an unfixable problem. Also, can't we just ask some guy who was working at the plant if there is a third stage? How does this guy know that it hasn't been upgraded?
I think the primary problems here are insufficient concrete information, and also masses of confirmation bias in both directions (leading people to essentially find whatever they want to find in a mass of conflicting information.) At least the information should become increasingly clear over the next month or two, which should cut down somewhat on the second problem.
This is a good site. And our prayers are with the quake survivors in Japan and the responders who are working to contain the problem during this horrible disaster.
Any model of nuclear power reliability would have a high degree of uncertainty (due to intractability) and would need to account for arbitrary faults in design, siting, construction, operation, and monitoring. In CS there's a concept of "byzantine fault tolerance" where every processor in a network can fail in arbitrary ways.
The minimum reliability rate predicted by the safety model would need to be on the order of 1-10^-8 to have assurance that major catastrophic events would not be likely to occur within some reasonable time span.
How much confidence can you have that a model is consistent with reality to within 8 or 9 decimal places in order to overcome the minimum acceptable reliability level? We cannot come close to this in a complex, realistic system analysis. This is not like other problems we normally encounter, where the minimum acceptable reliability to prevent an immediate global catastrophe is orders of magnitude lower.
Due to high model uncertainty and low acceptable failure rate, it is difficult in principle to come up with a convincing pro-nuclear argument whose lower interval didn't have a high expected death rate.
In this case, the heroism of the emergency responders was the difference between an immediate catastrophe and the current situation. Logistical problems and equipment failures impeded the response. Had the delays been exacerbated or had there been other complications (such as aftershocks that weakened the system), it's easy to see the delays could've been extended to the point where intervention wouldn't have been as effective.
Solar thermal power generation and wind are safe alternatives that can provide all 474 exajoules of earth's global energy demand. It's cheaper than the alternatives when externalities are properly accounted for and the initial investment would be a reasonable fraction of global GDP (certainly much less than the global financial crisis cost and it would actually stimulate the economy). Technology is ready. However, there are primarily sociological hurdles now (how do we get powerful polluters to change course?)
Why should anyone have to die for electricity? Renewable energy is safe. Emergency responders and residents are all at serious risk.
Depends - if it's our citizens following an accident obviously never. If it's some civilians in the middle east that happen to be living on top of our oil, or some poor people in the 3rd world that are being poisoned by air pollution - well tough, can't make an omelet without breaking eggs etc.
Look at how many accidents there have been since 1961. That's in the world, total. Only a few of those actually resulted in radiation being released to the environment. Nuclear power is unbelievably safe. Think about what it would take to replace the amount of nuclear energy being produced in the world now. Just a guess of course, but I think the butterfly effects of getting rid of nuclear power would be even worse than if there were some sort of catastrophic failure in Japan right now. It's not as simple as "nuclear energy in an accident would be catastrophic, let's replace it with clean alternatives". There are many many other factors at play.
Historical analysis doesn't have enough comparable data points and there are significant failures on record. How many times has a reactor been hit with an 8.9 magnitude earthquake and a tsunami? There's only 1 data point.
Historical and simulation models both have large negative expected value (unless you can rule out a black swan event with a miniscule probability at around 10^-8 - unattainable certainty with current analysis).
