Yes, that is true. However, assuming all flu epidemiological data suffers from the same systematic error, the H5N1 strain still has a much higher mortality rate than the other strains--ie: 58% may really be 35%, but then the 20% for H1N1 is actually 12%, assuming the same sampling error. So H5N1 is still a dangerous strain to take very seriously.
The minimal nature of our approach will likely lend itself to further elaboration, as we envision incorporating this system into a fully synthetic cell. We are also exploring practical applications of triazole membrane assembly, for instance in packaging and delivering therapeutics, improving transfection efficiencies, reconstituting functional membrane proteins, and performing confined biochemical reactions.
No. In fact, they mention in the paper that is fairly difficult to do. Knowing something about the enzyme megacomplexes that catalyze these reactions, I believe them. Hence why they developed this system, and why it made it into JACS (click chemistry by itself is nothing new). Overall, the triazole would not be expected to drastically affect the properties of the membrane, which they affirm by measuring a handful of bulk properties. So I think it's a pretty good model.
Well, they say it is biomimetic. It's about as close as anybody ever gets. The important part is that the properties of the artificial membranes (at least the ones they measured) are the same as for the natural membranes they were trying to mimic. It's a JACS communication, so there isn't a lot of detail, but it looks like a pretty good model. There are a lot of potential uses, not the least of which is they can more easily study how additional components in the membrane (ex: proteins) affect its properties.
The problem is the plants. Processes we take for granted when working with our model organisms in research labs, like recombination, don't exist in plants. So the precise introduction of genes is a very difficult thing to do.
We do this today with mixtures such as gut flora samples or cancer xenografts with varying degrees of success.
Oh yes, this is true. However, it is still fairly complicated and I'm not sure if it is a better (faster/easier) way to diagnose. Take, for example, strep throat. What does that entail, some Streptococcus variety (likely pyogenes) growing in your throat and threatening to invade your lungs. So you can cough into the machine and it can sequence everything, and you can use a fancy statistical model that somehow lets you differentiate between the Streptococcus growing in your throat and the Streptococcus that is a normal part of your oral flora within some degree of confidence. Alternatively, you can do a throat swab, and there are plenty of assays that you can do in minutes to hours (ELISA, PCR) to determine whether or not it is Streptococcus and which variant. Antibiotic susceptibility will still take a culture, but it seems to me that this is a solved problem, and whole genome sequencing provides more information than you really need, potentially making the diagnosis much more complicated.
The "genes" for antibiotic resistance often turn out to be plasmids.
Or modifications to the cell wall or translation machinery, such as in the development of vancomycin resistance in some strains. I don't mean to knock whole genome sequencing. I think it's an awesome tool for research purposes. It just isn't a panacea. Microarrays weren't either, for that matter, or pretty much any new development. It's novel and new, so it's cool, it will enable us to do things we weren't able to do before, but we aren't going to transform medicine overnight.
Indeed. I think the problem here is twofold: that there is such an overwhelmingly large number of organisms, and that we don't know how to culture the vast majority of them. So until deep sequencing was fully developed, it just wasn't possible to know what was really out there, aside from hints from a handful of 16s rDNA sequencing projects. Now that we have a much better idea, we still don't know how to do these studies. But there are people who are desperately trying. The human microbiome people are trying to link Lactobacillus to everything, from diabetes to cancer, by looking at the gut microbiome samples of different patient populations. It's pretty unsatisfying overall, but when you think about it the problem is fairly enormous. There are some limits in methodology that have to be overcome before we can really start to address these questions in a more constructive way. Until then, I think we will continue to see this "let's just sequence everything" approach.
I guess I just don't like the idea of unequivocally claiming that we know exactly how something works. It's just backfired so many times in the history of science. There was a rather famous exchange during the 1970s between Bob Abeles (the guy who worked out the mechanisms of vitamin B12 mediated reactions) and some Harvard chemists who said the chemistry he was proposing was "impossible." Well, long story short, Abeles was right and the Harvard consortium was wrong.
Well, if anything I think the reductionist models have been much more useful than the exhaustive ones to date. Understanding that there are control points at key steps in metabolism has been much more useful (for understanding and prediction), in my opinion, than previous efforts to model the contribution of every enzyme in the pathway.
Bioinformatics is great. Many advances have been driven by developments in that field. But it is no more a silver bullet than x-ray crystallography was. Right now, efforts to understand native microbiomes is almost entirely driven by bioinformatics, because there is currently no other way to study these systems. But so far, the predictive power has been useless. Knowing which organisms (and in some cases which genes) are present has helped us appreciate the diversity out there, but hasn't contributed much to the clarity of how these micorbial communities develop and change in response to environmental stimuli.
No, I don't think so. Simulation of complex biological systems, especially when you include molecular details like protein folds and conformation changes, takes a tremendous amount of computer time. If you can approximate some things using empirical measurements, you can speed things up a bit, but even so we are pretty far from knowing all of the rules we need to know to simulate even relatively straightforward things like the progression of an infection in a host.
People were saying similar things decades ago when mechanistic enzymology was a hot field. "All we need to know is get binding and catalytic parameters for every enzyme in primary metabolism, and then we will be able to model it." Now, decades later, we still can't even simulate a simple pathway like glycolysis much less all of the interacting pathways.
One huge application is diagnosis of infectious disease. Not only will you know whether what you have is viral or bacterial, if it's bacterial you'll know exactly which antibiotics will work. And you'll probably even be able to correlate you infection with known outbreaks: "There's been 50 other infections with this pathogens in the northeast corner of this city over the last two weeks."
I think you're overselling this a bit. Sequencing a human genome is not going to tell you whether you are infected with a pathogen, and it won't identify the pathogen. To do that, you will need to isolate the organism and sequence it that way. Also, having the genome of an organism does not automatically make it easier to determine things like antibiotic resistance. In some ways it makes it more difficult, because you have too much information. High throughput sequencing is definitely a great advance for the scientific community but, like all great advances, it's a tool that is useful for some things, not a magic answer machine.
Oh and by the way, everything you mention for diagnosis above can already be done with current methods faster and cheaper than with whole genome sequencing.
Yes, that's true. However, I also think it is because of this concept of a "program directory" that contains everything. On Linux there are packages, but no such thing as a program directory. So software has no choice but to write user data and run time variables to/tmp or user home directories. They can't just throw in a temp file with the binaries. On Windows, with everything already conveniently in one place, it's easier to just make your own temp directory in the program directory.
Installing into a different location is just a workaround. No better than just giving write permissions to Program Files. The fundamental problem is that program data is mixed up with user data. Without a clean separation, simple tools like Reset and Refresh mentioned in this topic will break horribly.
Another fellow mentioned each user installing their own copy of the software. This works for stuff like Endnote, but not for the software that runs the LC-MS. It would be nice if instrument makers would write better software, but that just isn't happening.
You obviously don't use much scientific software. Almost all instrumentation software writes config settings and user data to Program Files. The only way to get them to work is to give users write access to that folder. Stupid, yes, but that's the way it is. Also quite a bit of analysis software refuses to acknowledge multiple users, and wants to write in the directory where it was installed. So people make elaborate trees of subdirectories for labs and individual users inside the program directory!
It would, of course, be the same problem on Linux if the programmers were equally lazy/stupid, but for some reason that doesn't seem to happen. Even Linux versions of the same Windows program just behave better.
Because, just maybe, the VAST bulk of today's plant aren't passive cooled? What ifs are great, but the here and now is not passively cooled reactors.
None are passively cooled. What exactly are we talking about? You are not getting any argument from me that light water reactors have safety concerns. Nuclear can be done safely. That's my only argument. The technology exists. It just has to be used.
Sources? My understanding is that breeder reactors are the sort used for proliferation...and we precisely didn't use them for that reason. As for terrorism, that wasn't a serious concern when most of these were being built.
The primary concern is that all fissile and radioactive material is tracked, securely handled, and accounted for. This includes fuel that enters the reactor, waste that leaves the reactor, any material (like sources) used in the reactor, isotopes produced by the reactor during operation (dissolved gases, mostly), and anything else that shows up on a Geiger counter in the reactor building, even if it is a natural source of radiation (like the cement walls). Some of it is to ensure safe operation of the reactor, yes. For example, regular fuel rod inspection and monitoring for cesium isotopes is to ensure the fuel cladding is intact. But the majority is for access control, record keeping, and auditing. Dirty bombs are one concern, but there are others as well.
They cost a lot because it would be B.A.D. if they were allowed to be as shoddy as regular coal plants.
Yes, light water reactors are over-engineered with redundant safety systems. That is a part of the cost. I already said that. The regulations are good, I'm not arguing otherwise. I'm just saying that, because of the regulations, the cost to operate a nuclear plant must be entirely internalized. Coal will always be cheaper because it is allowed the externalize a significant amount of its environmental, accident recovery, and cleanup cost. It is not held to the same standard. If clean coal or cap and trade start being seriously discussed, the cost of coal will increase to meet these new regulations, and then everybody will probably switch to natural gas.
Call it what you want, but the amount that candidates raise (through their PAC or otherwise) during the run up to the primary is often a huge factor in determining whether they will be viable in a race. You obviously can't restrict the free speech of an individual. But PACs are already legally distinguished groups. They can be (and are) regulated. The Supreme Court has just made it more difficult with respect to corporate donations.
Nice concept and I agree. Now back to the reality of our situation today...
Why do you think that is not the reality of today? Like I said, prototype designs exist. Experimental reactors have been built. This is not a fantasy. It can be done. The reality of today is that a new commercial power plant reactor has not been built in the US for over three decades. That is why the designs are from three decades ago.
Ask yourself *why* it is highly regulated? Perhaps because of the risks of it not being 'safe'?
Actually, it is mostly due to fears of proliferation and terrorism, not because they are viewed as unsafe. Monitoring radioactive releases is not unreasonable and overly costly. Coal plants should be held to the same standard, but they aren't.
Nuclear simply always will have massive risks.
There are no conceivable massive risks in a reactor that cannot melt down. The mild to moderate risks only amount to environmental releases that can be monitored and controlled, as they should be with all power plants.
It gave us bombs for the cold war and energy to boot.
A bomb is as much like a reactor as gasoline is like whiskey. In other words, not at all.
Not necessarily. Plenty of restrictions to free speech have been upheld by the judiciary. Considering we are talking about public office here, not private enterprise, I don't see why it would be unacceptable to restrict the actions of candidates seeking it. But perhaps you don't have to be quite so explicit. A modification of campaign finance laws would get you most of the way there. The only reason candidates campaign the way they do is because they can raise hundreds of millions of dollars from undisclosed sources to do it. If they were limited to something more reasonable, say the annual income of an average middle class family, I think the landscape would be quite different.
If nuclear goes boom you're potential failure area pretty quickly becomes fatalities. Simply because it hasn't happened yet doesn't mean it isn't plausible.
It isn't plausible. It is physically impossible. A meltdown is the worst thing that can happen with a reactor of Fukushima's design, and it won't result in massive fatalities.
Coal mine deaths are part of operational deaths. It's a planned activity.
Interesting. I would expect the collapse of a coal mine to be a "failure," not an operational expectation. Perhaps we don't hold coal mine operators to a high enough safety standard.
As far as 'good' solutions go. Fukushima was considered 'safe' prior to this disaster.
For relative degrees of "safe." It has been well-known for a long time what the potential failure points of light water reactors are. Nobody ever claimed it was impossible for Fukushima to melt down. It was acknowledged as a disaster possibility, and massively redundant safety mechanisms were put in place as a result. Yes, even redundant safety systems can fail, as was evidenced in this case. That is why it is better to build a passively safe reactor that cannot melt down. The worst-case disaster scenarios then change completely. Prototype designs have been available for quite some time, but none have been pushed into commercial use yet.
As you say, the 'good' solutions are massively expensive - expressly because of the risks involved in nuclear.
Not sure about that. If that is true, the safer designs should be cheaper, because they are safer and pose less risk. No, the reason it is expensive is because it is complicated technology that is highly regulated (rightly so). And the newer designs also require some R&D costs that an older established design doesn't have. For coal, the principle is relatively simple, has been around for over a century, the fuel is cheaply mined and readily available (ie: no expensive fuel processing is required), and the environmental regulations are few (far fewer than should be).
Coal has very very few failure fatalities, nuclear has fairly high potential failure fatalities.
No, not really. A high potential failure zone, but not fatalities. By your own admission, a coal plant may explode and kill all workers on site at the time (plus coal miner deaths when the mine collapses, btw). So what is that, several hundred? In a nuclear accident, the only likely fatalities will be anybody who is in the reactor containment room at the time, and maybe some people in the adjacent control room (likely, less than a dozen in total). The problem with nuclear is not the fatalities, it is the contamination zone that makes it unsafe to work/live in after an accident.
nuclear obviously the failure mode has no 'good' solutions.
There are actually quite a number of "good solutions." It shouldn't be a surprise to anybody that the first and second generation reactor designs from the 70s have problems, much like the first generation cars and airplanes from last century (somebody was mentioning cars and airplanes above). There are some very good new designs available, but they have to be developed. The problem is nobody is pursuing them because coal and natural gas are far far cheaper options.
Teaching takes intelligence. The smarter you are, the better you do. There is a reason why colleges try to get the smartest, most published people to teach.
Uh, no. Colleges hire the most published because they will be able to bring in grant money and increase the prestige of the college. I agree that teaching takes intelligence, but you have to care about teaching also. I've met many college professors who are brilliant in their fields, but they really hate teaching and do a lousy job as a result.
I know this will sound crazy, but I think the problem is campaigning. We should get rid of it. Think about it. How would you vote for a candidate if they didn't campaign and plaster all media outlets with their garbage 2 years ahead of the actual election? You would be forced to vote for people you personally know and interact with. You would have to know something about their actual character and record of public service, not just what they say about themselves (or their opponents say about them). I think this would go a long way toward improving the electoral process and reducing the influence of large money donors in government. And it would have the side effect of forcing representatives to spend more time working in their districts, listening to public concerns, instead of just doing quick stop tours and catering to campaign donors at fundraiser dinners.
Slashdot Mirror
Yes, that is true. However, assuming all flu epidemiological data suffers from the same systematic error, the H5N1 strain still has a much higher mortality rate than the other strains--ie: 58% may really be 35%, but then the 20% for H1N1 is actually 12%, assuming the same sampling error. So H5N1 is still a dangerous strain to take very seriously.
The .docx support for the Mac version of Office is what gives me the most trouble. I can send some of the problem documents if that is helpful.
From the paper,
The minimal nature of our approach will likely lend itself to further elaboration, as we envision incorporating this system into a fully synthetic cell. We are also exploring practical applications of triazole membrane assembly, for instance in packaging and delivering therapeutics, improving transfection efficiencies, reconstituting functional membrane proteins, and performing confined biochemical reactions.
No. In fact, they mention in the paper that is fairly difficult to do. Knowing something about the enzyme megacomplexes that catalyze these reactions, I believe them. Hence why they developed this system, and why it made it into JACS (click chemistry by itself is nothing new). Overall, the triazole would not be expected to drastically affect the properties of the membrane, which they affirm by measuring a handful of bulk properties. So I think it's a pretty good model.
Well, they say it is biomimetic. It's about as close as anybody ever gets. The important part is that the properties of the artificial membranes (at least the ones they measured) are the same as for the natural membranes they were trying to mimic. It's a JACS communication, so there isn't a lot of detail, but it looks like a pretty good model. There are a lot of potential uses, not the least of which is they can more easily study how additional components in the membrane (ex: proteins) affect its properties.
The problem is the plants. Processes we take for granted when working with our model organisms in research labs, like recombination, don't exist in plants. So the precise introduction of genes is a very difficult thing to do.
We do this today with mixtures such as gut flora samples or cancer xenografts with varying degrees of success.
Oh yes, this is true. However, it is still fairly complicated and I'm not sure if it is a better (faster/easier) way to diagnose. Take, for example, strep throat. What does that entail, some Streptococcus variety (likely pyogenes) growing in your throat and threatening to invade your lungs. So you can cough into the machine and it can sequence everything, and you can use a fancy statistical model that somehow lets you differentiate between the Streptococcus growing in your throat and the Streptococcus that is a normal part of your oral flora within some degree of confidence. Alternatively, you can do a throat swab, and there are plenty of assays that you can do in minutes to hours (ELISA, PCR) to determine whether or not it is Streptococcus and which variant. Antibiotic susceptibility will still take a culture, but it seems to me that this is a solved problem, and whole genome sequencing provides more information than you really need, potentially making the diagnosis much more complicated.
The "genes" for antibiotic resistance often turn out to be plasmids.
Or modifications to the cell wall or translation machinery, such as in the development of vancomycin resistance in some strains. I don't mean to knock whole genome sequencing. I think it's an awesome tool for research purposes. It just isn't a panacea. Microarrays weren't either, for that matter, or pretty much any new development. It's novel and new, so it's cool, it will enable us to do things we weren't able to do before, but we aren't going to transform medicine overnight.
Indeed. I think the problem here is twofold: that there is such an overwhelmingly large number of organisms, and that we don't know how to culture the vast majority of them. So until deep sequencing was fully developed, it just wasn't possible to know what was really out there, aside from hints from a handful of 16s rDNA sequencing projects. Now that we have a much better idea, we still don't know how to do these studies. But there are people who are desperately trying. The human microbiome people are trying to link Lactobacillus to everything, from diabetes to cancer, by looking at the gut microbiome samples of different patient populations. It's pretty unsatisfying overall, but when you think about it the problem is fairly enormous. There are some limits in methodology that have to be overcome before we can really start to address these questions in a more constructive way. Until then, I think we will continue to see this "let's just sequence everything" approach.
Hehe, yes well, it was a good explanation.
I guess I just don't like the idea of unequivocally claiming that we know exactly how something works. It's just backfired so many times in the history of science. There was a rather famous exchange during the 1970s between Bob Abeles (the guy who worked out the mechanisms of vitamin B12 mediated reactions) and some Harvard chemists who said the chemistry he was proposing was "impossible." Well, long story short, Abeles was right and the Harvard consortium was wrong.
Well, if anything I think the reductionist models have been much more useful than the exhaustive ones to date. Understanding that there are control points at key steps in metabolism has been much more useful (for understanding and prediction), in my opinion, than previous efforts to model the contribution of every enzyme in the pathway.
Bioinformatics is great. Many advances have been driven by developments in that field. But it is no more a silver bullet than x-ray crystallography was. Right now, efforts to understand native microbiomes is almost entirely driven by bioinformatics, because there is currently no other way to study these systems. But so far, the predictive power has been useless. Knowing which organisms (and in some cases which genes) are present has helped us appreciate the diversity out there, but hasn't contributed much to the clarity of how these micorbial communities develop and change in response to environmental stimuli.
Well, yeah, and the appendix was also a junk organ until quite recently. Just because we don't know what it does doesn't mean it is non-functional.
No, I don't think so. Simulation of complex biological systems, especially when you include molecular details like protein folds and conformation changes, takes a tremendous amount of computer time. If you can approximate some things using empirical measurements, you can speed things up a bit, but even so we are pretty far from knowing all of the rules we need to know to simulate even relatively straightforward things like the progression of an infection in a host.
People were saying similar things decades ago when mechanistic enzymology was a hot field. "All we need to know is get binding and catalytic parameters for every enzyme in primary metabolism, and then we will be able to model it." Now, decades later, we still can't even simulate a simple pathway like glycolysis much less all of the interacting pathways.
One huge application is diagnosis of infectious disease. Not only will you know whether what you have is viral or bacterial, if it's bacterial you'll know exactly which antibiotics will work. And you'll probably even be able to correlate you infection with known outbreaks: "There's been 50 other infections with this pathogens in the northeast corner of this city over the last two weeks."
I think you're overselling this a bit. Sequencing a human genome is not going to tell you whether you are infected with a pathogen, and it won't identify the pathogen. To do that, you will need to isolate the organism and sequence it that way. Also, having the genome of an organism does not automatically make it easier to determine things like antibiotic resistance. In some ways it makes it more difficult, because you have too much information. High throughput sequencing is definitely a great advance for the scientific community but, like all great advances, it's a tool that is useful for some things, not a magic answer machine.
Oh and by the way, everything you mention for diagnosis above can already be done with current methods faster and cheaper than with whole genome sequencing.
Yes, that's true. However, I also think it is because of this concept of a "program directory" that contains everything. On Linux there are packages, but no such thing as a program directory. So software has no choice but to write user data and run time variables to /tmp or user home directories. They can't just throw in a temp file with the binaries. On Windows, with everything already conveniently in one place, it's easier to just make your own temp directory in the program directory.
Installing into a different location is just a workaround. No better than just giving write permissions to Program Files. The fundamental problem is that program data is mixed up with user data. Without a clean separation, simple tools like Reset and Refresh mentioned in this topic will break horribly.
Another fellow mentioned each user installing their own copy of the software. This works for stuff like Endnote, but not for the software that runs the LC-MS. It would be nice if instrument makers would write better software, but that just isn't happening.
Really? That's your solution for a program used by 100 users to run a mass spectrometer?
You obviously don't use much scientific software. Almost all instrumentation software writes config settings and user data to Program Files. The only way to get them to work is to give users write access to that folder. Stupid, yes, but that's the way it is. Also quite a bit of analysis software refuses to acknowledge multiple users, and wants to write in the directory where it was installed. So people make elaborate trees of subdirectories for labs and individual users inside the program directory!
It would, of course, be the same problem on Linux if the programmers were equally lazy/stupid, but for some reason that doesn't seem to happen. Even Linux versions of the same Windows program just behave better.
Because, just maybe, the VAST bulk of today's plant aren't passive cooled? What ifs are great, but the here and now is not passively cooled reactors.
None are passively cooled. What exactly are we talking about? You are not getting any argument from me that light water reactors have safety concerns. Nuclear can be done safely. That's my only argument. The technology exists. It just has to be used.
Sources? My understanding is that breeder reactors are the sort used for proliferation...and we precisely didn't use them for that reason. As for terrorism, that wasn't a serious concern when most of these were being built.
The primary concern is that all fissile and radioactive material is tracked, securely handled, and accounted for. This includes fuel that enters the reactor, waste that leaves the reactor, any material (like sources) used in the reactor, isotopes produced by the reactor during operation (dissolved gases, mostly), and anything else that shows up on a Geiger counter in the reactor building, even if it is a natural source of radiation (like the cement walls). Some of it is to ensure safe operation of the reactor, yes. For example, regular fuel rod inspection and monitoring for cesium isotopes is to ensure the fuel cladding is intact. But the majority is for access control, record keeping, and auditing. Dirty bombs are one concern, but there are others as well.
They cost a lot because it would be B.A.D. if they were allowed to be as shoddy as regular coal plants.
Yes, light water reactors are over-engineered with redundant safety systems. That is a part of the cost. I already said that. The regulations are good, I'm not arguing otherwise. I'm just saying that, because of the regulations, the cost to operate a nuclear plant must be entirely internalized. Coal will always be cheaper because it is allowed the externalize a significant amount of its environmental, accident recovery, and cleanup cost. It is not held to the same standard. If clean coal or cap and trade start being seriously discussed, the cost of coal will increase to meet these new regulations, and then everybody will probably switch to natural gas.
Call it what you want, but the amount that candidates raise (through their PAC or otherwise) during the run up to the primary is often a huge factor in determining whether they will be viable in a race. You obviously can't restrict the free speech of an individual. But PACs are already legally distinguished groups. They can be (and are) regulated. The Supreme Court has just made it more difficult with respect to corporate donations.
Nice concept and I agree. Now back to the reality of our situation today...
Why do you think that is not the reality of today? Like I said, prototype designs exist. Experimental reactors have been built. This is not a fantasy. It can be done. The reality of today is that a new commercial power plant reactor has not been built in the US for over three decades. That is why the designs are from three decades ago.
Ask yourself *why* it is highly regulated? Perhaps because of the risks of it not being 'safe'?
Actually, it is mostly due to fears of proliferation and terrorism, not because they are viewed as unsafe. Monitoring radioactive releases is not unreasonable and overly costly. Coal plants should be held to the same standard, but they aren't.
Nuclear simply always will have massive risks.
There are no conceivable massive risks in a reactor that cannot melt down. The mild to moderate risks only amount to environmental releases that can be monitored and controlled, as they should be with all power plants.
It gave us bombs for the cold war and energy to boot.
A bomb is as much like a reactor as gasoline is like whiskey. In other words, not at all.
Not necessarily. Plenty of restrictions to free speech have been upheld by the judiciary. Considering we are talking about public office here, not private enterprise, I don't see why it would be unacceptable to restrict the actions of candidates seeking it. But perhaps you don't have to be quite so explicit. A modification of campaign finance laws would get you most of the way there. The only reason candidates campaign the way they do is because they can raise hundreds of millions of dollars from undisclosed sources to do it. If they were limited to something more reasonable, say the annual income of an average middle class family, I think the landscape would be quite different.
If nuclear goes boom you're potential failure area pretty quickly becomes fatalities. Simply because it hasn't happened yet doesn't mean it isn't plausible.
It isn't plausible. It is physically impossible. A meltdown is the worst thing that can happen with a reactor of Fukushima's design, and it won't result in massive fatalities.
Coal mine deaths are part of operational deaths. It's a planned activity.
Interesting. I would expect the collapse of a coal mine to be a "failure," not an operational expectation. Perhaps we don't hold coal mine operators to a high enough safety standard.
As far as 'good' solutions go. Fukushima was considered 'safe' prior to this disaster.
For relative degrees of "safe." It has been well-known for a long time what the potential failure points of light water reactors are. Nobody ever claimed it was impossible for Fukushima to melt down. It was acknowledged as a disaster possibility, and massively redundant safety mechanisms were put in place as a result. Yes, even redundant safety systems can fail, as was evidenced in this case. That is why it is better to build a passively safe reactor that cannot melt down. The worst-case disaster scenarios then change completely. Prototype designs have been available for quite some time, but none have been pushed into commercial use yet.
As you say, the 'good' solutions are massively expensive - expressly because of the risks involved in nuclear.
Not sure about that. If that is true, the safer designs should be cheaper, because they are safer and pose less risk. No, the reason it is expensive is because it is complicated technology that is highly regulated (rightly so). And the newer designs also require some R&D costs that an older established design doesn't have. For coal, the principle is relatively simple, has been around for over a century, the fuel is cheaply mined and readily available (ie: no expensive fuel processing is required), and the environmental regulations are few (far fewer than should be).
Coal has very very few failure fatalities, nuclear has fairly high potential failure fatalities.
No, not really. A high potential failure zone, but not fatalities. By your own admission, a coal plant may explode and kill all workers on site at the time (plus coal miner deaths when the mine collapses, btw). So what is that, several hundred? In a nuclear accident, the only likely fatalities will be anybody who is in the reactor containment room at the time, and maybe some people in the adjacent control room (likely, less than a dozen in total). The problem with nuclear is not the fatalities, it is the contamination zone that makes it unsafe to work/live in after an accident.
nuclear obviously the failure mode has no 'good' solutions.
There are actually quite a number of "good solutions." It shouldn't be a surprise to anybody that the first and second generation reactor designs from the 70s have problems, much like the first generation cars and airplanes from last century (somebody was mentioning cars and airplanes above). There are some very good new designs available, but they have to be developed. The problem is nobody is pursuing them because coal and natural gas are far far cheaper options.
Teaching takes intelligence. The smarter you are, the better you do. There is a reason why colleges try to get the smartest, most published people to teach.
Uh, no. Colleges hire the most published because they will be able to bring in grant money and increase the prestige of the college. I agree that teaching takes intelligence, but you have to care about teaching also. I've met many college professors who are brilliant in their fields, but they really hate teaching and do a lousy job as a result.
I know this will sound crazy, but I think the problem is campaigning. We should get rid of it. Think about it. How would you vote for a candidate if they didn't campaign and plaster all media outlets with their garbage 2 years ahead of the actual election? You would be forced to vote for people you personally know and interact with. You would have to know something about their actual character and record of public service, not just what they say about themselves (or their opponents say about them). I think this would go a long way toward improving the electoral process and reducing the influence of large money donors in government. And it would have the side effect of forcing representatives to spend more time working in their districts, listening to public concerns, instead of just doing quick stop tours and catering to campaign donors at fundraiser dinners.