The question How would you develop software for a nuclear plant? got some interesting answers, despite being closed at least once (I think). Yet when I asked the OP what he wanted from the question, he said:

Actually I just wanted to hear some war stories but apparently there is no such thing as "software that controls nuclear plant". – Pavel Chuchuva

My question is what to do with this question for posterity. It turns out that there was no actual answer, as there is no software that actually controls a nuclear power plant. Should we:

  1. Close and eventually delete it?
  2. Ask the OP to edit it to be more general, and to better fit the answers, so that one answer might eventually be chosen?
  3. Do the editing ourselves?
  4. Something else?

Ideas, thoughts, meta-discussion welcome.

  • 2
    Actually there is software that controls ... or at least monitors nuclear plants. Could be SCADA or PLC. I recall some controvosy about the testing of the Sizewell B monitoring software making it to the mainstream news. – Richard Jul 3 '09 at 10:26
  • 1
    Where is everyone getting this "no software in nuclear plants" idea? – Richard JP Le Guen Oct 10 '12 at 23:13
  • I believe it's "no software that controls a nuclear plant". – John Saunders Oct 10 '12 at 23:33
  • @JohnSaunders - Ok, where are people getting that idea then? – Richard JP Le Guen Oct 10 '12 at 23:39
  • 1
    This question is from July 2009. It's possible they are no longer getting this idea at all. – John Saunders Oct 10 '12 at 23:50
  • @JohnSaunders - Fair point! – Richard JP Le Guen Oct 10 '12 at 23:53
  • As per today's standards, shouldn't that question be closed again as not constructive? – bfavaretto Oct 11 '12 at 2:24
  • @bfavaretto: that's what I think, which is why the Close vote is mine. – John Saunders Oct 11 '12 at 2:37
  • @JohnSaunders I can only see one close vote, which is mine. How many do you see? – bfavaretto Oct 11 '12 at 2:49
  • I see one and it won't let me cast another. It may have been closed (which is when i voted) then reopend (which is when you voted). – John Saunders Oct 11 '12 at 2:50
  • @bfavaretto: good question, and I've asked meta.stackexchange.com/questions/150035/…. – John Saunders Oct 11 '12 at 2:55
  • Wow - I hope no one has deployed faulty or untested software for a nuclear plant. :/ – Anderson Green Feb 25 '13 at 3:24

I think the question has merit. It probably could have been worded better, but I get the gist. I think he was just curious about the software development procedures that are used in such a safety critical environment. I'm curious too.

| improve this answer | |
  • How would people feel if the question was edited to be about safety-critical systems in general? – John Saunders Jul 3 '09 at 0:36
  • Just fine. It doesn't have to be about nuclear power. What about the software used in medical life support devices, the air traffic control system, the New York stock exchange? – raven Jul 3 '09 at 1:36
  • Fully agree. "Mission Critical" is a better term. Even if the question was meant as a joke, it seems to have generated serious answers and deserves to stay open. – Euro Micelli Jul 6 '09 at 2:39
  • The question was deleted - I'm not sure why. :( – Anderson Green Feb 25 '13 at 3:25
  • You don't want to know about software used for the NYSE...trust me. – Michael Brown Sep 25 '14 at 21:32

Since there is no software that controls nuclear plants, he should accept that answer. Since it's the correct one.

If he is really asking a question that is not really about nuclear software :P Then he should edit the question since his original premise doesn't work (only as an exercise). If the edit is good, then the question could garner some good answers. (not like the joke ones currently at the top)

| improve this answer | |
  • You're making assumptions that correctness was the criteria for accepting the answer; is that necessarily true? – Paul Sonier Jul 2 '09 at 22:05
  • I've amended to my answer. – Ólafur Waage Jul 2 '09 at 22:07

I agree that my question is vague and really calls for opinion or speculation, not for an answer. raven is right - I was just curious. However the question got some good answers and at least one of them is from person in the field.

So go ahead, edit the question to match good answers.

| improve this answer | |

I think questions about nonexistent software are to some extent speculative questions; I think there's a place for them if they're asked as speculative questions, such as "how would one write software for high reliability such as for a nuclear power plant", but questions about software that are only assumed to exist are a little bit TOO speculative (and assumptive).

| improve this answer | |
  • 1
    Ok, but how would you feel about it if he edited the question to be more like your suggestion? – John Saunders Jul 2 '09 at 22:20
  • I think it would be a reasonable question, and I think there's several examples of such on SO, actually.... – Paul Sonier Jul 3 '09 at 2:06

I voted to close it because it was very vague to start with. This is much like asking how databases work--it's not a question that can be answered without throwing the book at somebody. Nuclear power plants are complicated collections of mechanical devices, embedded code, and software. Thinking that you're going to get an answer other than "very carefully" even if there was software that controlled a nuclear power plant is naive, at best.

Plenty of more focused questions have been closed due to them being too broad. I fail to see how this is any different. This is something that belongs on howstuffworks.com, not on stackoverflow.com.

| improve this answer | |
  • @Eric, I agree in terms of the question. I'm mostly interested in seeing if there's something we can do to salvage it for the answers. They're interesting answers to some question, after all! – John Saunders Jul 2 '09 at 22:21
  • 1
    @John: They're absolutely interesting, but I'd still close it, since they question that they're answers to doesn't really belong at SO. – Eric Jul 2 '09 at 22:27

I honestly didn't expect to get up-voted for my answer. In fact, I thought I would get down-voted. The question was so vague that I started to site some mil-specs and some Common Criteria stuff. But, then I remembered a fun fact I knew about the java license agreement and decided to respond with that because at least it was interesting and correctly responsive.

As for what to do? I don't know. The problem in my mind is that it is a slippery slope if you just delete it. I would edit the question to be about developing highly secure software. For example, it might be interesting to discuss how to architect develop Common Criteria software.

| improve this answer | |

It turns out that there was no actual answer, as there is no software that actually controls a nuclear power plant.

  1. Close and eventually delete it?

Is this any different from a question like.. "how to read and write into file using javascript" - to which the answer is "No"...?

The problem is, it's not really a question, but rather a topic for discussion, something SO isn't really suited for..

That said, quite a few of the answers are interesting, it's programming related and I don't think it's really harming the site..

| improve this answer | |

Many nuclear reactors are controlled by software. This has not generally been necessary in the United States where enriched fuel permits a physically small but high power reactor. Any reactor that is physically large with a longer neutron lifetime is generally software controlled. The CANDU reactor is one example of this.

This is nothing new - computer control of reactors goes back to the early 1960s and I seem to recall that the PDP-5 was designed with this purpose specifically in mind. The IBM 1800, Varian 72 and PDP 11 are all currently used as reactor control computers. Typically the actual control is done from assembly language code without an intervening operating system.

All safety critical software undergoes a rigorous categorization and qualification process. There are both national and international standards than must be adhered to (can't recall the standard numbers offhand). The software involved is typically not huge - maybe a couple hundred thousand lines. Due to the qualification process any safety-critical code is extraordinarily expensive.

| improve this answer | |
  • 1
    I am very confused as to why you answered this question here and not at SO. This is the question about the question. Meta :) – Eric Jul 6 '09 at 2:59
  • Me too. Also, when you answer over there, please explain how your answer fits with the others saying software isn't used to control nuclear power plants. – John Saunders Jul 6 '09 at 7:28
  • The answer on stackoverflow is wrong and the question is closed. Should the question be reposted? – Michelle Jul 6 '09 at 21:25

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .