Netbeans 3.6 is quite nice as well. I agree with the parent that it isn't really that slow. It's fairly slow on my OS X box (which is a G5, so there's no excuse, you hear me Apple), but at work on windows and linux it's snappy enough that I can't tell it isn't native.
I really wish SUN would just bite the bullet and provide a good profiler with netbeans though. They already include a debugger, why not a profiler too? It seems like netbeans is to the level where it really just needs some additional plugins, it already pretty much surpasses visual studio in terms of usefulness as far as I'm concerned.
In addition, be sure to run it on a modern VM, the newer JVMs have pretty substantially improved SWING performance. Apple's offering seems to be the weakest in this area, but it's made the most improvement from the early days, so maybe by 1.5 they'll be competitive.
It has a stronger foothold everywhere the GDP is lower. The more expensive M$ licenses are compared to the cost of a worker, the more enticing opensource is.
Well, in addition, lets talk about STL and pre-processor macros. Pre processor macros and STL wildly increase the volume of code generated for a program. If I had to guess, I'd say that very little of a modern program's code was actually written by humans. There is just no way people could write so much crap that a word processor would have 100 megs of code.
This is (one reason) why higher level languages are so nice. They give you MUCH smaller executables, as everyone has standardized on the same libraries, namely those that came with the environment. There is just much more code sharing in Java or.NET (or pretty much anything else) than there is in C or C++, and it shows.
Good point. I always just figured that the hydrogen storage tanks could be ultra high pressure segmented tanks, but that might be a tough sell for aircraft, etc.... On of the advantages that you have with hydrogen, is that it isn't as easy to destroy things with it. It tends to rise so rapidly that it doesn't easily burn people (who tend to be on the ground). Also, it doesn't contain soot, so it doesn't radiate heat very efficiently, which helps to cut down on burns. Might be useful in military situations, but maybe overshadowed by the high pressure tank problems.
don't be foolish. Any idiot will just tell you to run everything on Hydrogen, which you can make from the electricity. So in a way, everything could be nuclear powered.
Speaking of which.....
I haven't heard much about it yet, but I wouldn't be surprised if there were big projects in the pentagon to consider hydrogen powered tanks and planes and warships wherever possible. If you had a nuclear flagship (aircraft carrier) it could use its reactors to generate all the hydrogen it needs. Then it doesn't have to carry fuel for the fighters, the other escort ships (currently diesel) wouldn't run out of fuel, etc.... You would completely eliminate the fuel costs and weaknesses from the equation. Same thing for tanks. Just park a carrier or other nuclear vessel nearby, make hydrogen from the water it's sitting in, and have all the fuel you need to run your land campaign.
Also, fuel cells would probably get better mileage than the standard parts for tanks and ships at least. Aircraft might be much tougher.
true enough, but it's hardly required to have a civilian nuclear program in order to have weapons. As exhibit A, I give the United States of America. During the 1940s this country has no civilian nuclear program to speak of, and yet still managed to become a nuclear power (using 50 year old technology, and blazing the trail all the way) in about a decade, or less, depending on how you count.
I'm not arguing that civilian fuel can't be used to make weapons, I'm just arguing that having a civilian program doesn't imply that weapons will also be made (or available). Since you took a nuclear physics course, we both know that reprocessing nuclear fuel is difficult for the unprepared, and releases detectable krypton. It's probably much harder to get away with nuclear reactors + reprocessing than it is to get away with Uranium enrichment. Especially because the reactors need enriched Uranium anway, so why not just enrich it a few more times while you're at it and dispense with the reactors entirely.
So though civilian plants are a path to nuclear weapons, they are probably much tougher than just brute forcing it using a good (or even a sucky) Uranium diffusion plant. Now the US did use plutonium producing piles, we also were a Huge country with very little to fear from anyone and plenty of room to hide the program. In addition, technology wasn't as advanced as it now is. I imagine that modern laser diffusion (not sure of the technical name of this) or gaseous diffusion isn't even very difficult anymore, whereas it would have been very tough 50 years ago.
Ok, this is nice, but neither side gives any evidence. Since when does "no it isn't" count as a refutation?
Everything that guy has to say is about nuclear weapons. Well, guess what. WE ALREADY HAVE NUCLEAR WEAPONS. There, accept it. Get over it. There is no danger of additional reactors turning the US, or China, or India, or Western Europe into nuclear armed powers. NONE, because they already are.
It's easy to tear down someone else's proposal when you don't have on of your own and need rely on nothing but juvenile comebacks. Get some actual evidence. And you know what, even if you count the victims of Hiroshima and Nagisaki against nuclear power (but don't count the victims of conventional warfare against fossil fuels) and you throw in Cherenoble, and maybe round everything up by a few hundred thousand just to be sure, Nuclear killed far fewer people per kWh of energy. It is almost impossible to imagine a scenario in which it might be otherwise. Fossil fuels kill tens (hundreds, depending on how you count) of thousands of people each year.
A nuclear disaster would have to kill tens of millions (at least) in order to even the score. Nobody can even conceive of how that could happen with civilian reactors built to even the most incompetent of standards, like Cherenobl. About the only real possibility is if WW-III breaks out and people start tossing around nuclear weapons (which they already have, and don't need civilian reactors for), and that is far MORE likely if we start fighting over oil.
Just once I'd like to hear a well reasoned out anti-nuclear position. Include some numbers (you know, dollars and cents, lives lost, that sort of thing) and keep them accurate. Include an honest asessment of nuclear waste dangers (assuming various means of disposal) and honest asessments of nuclear proliferation. I have never seen any evidence that civilian nuclear power leads to proliferation, but it seems to be a given for the anti-nuke types. Japan and South Korea both have reactors, and neither has nuclear weapons.
The only scenario the anti-nuke types ever argue against is such a complete straw man. They assume we dump all the nuclear waste into the nation's beer supply, give away spent fuel to everyone with a driver's license, and somehow (though nobody can really imagine exactly how this happens) have lots of melt downs in highly populated areas. Seriously. Assume an even marginally competent nuclear program (needn't be perfect) and then try a comparison with our fossil fuel system. See how that treats you.
It's like comparing against an oil economy where it's assumed that 99% of the oil is dumped raw into the ocean, the rest is burned in the foulest, dirtiest machines imaginable, and that somehow access to oil allows every fool who can rub two sticks together to build a jet fighter with which to kill people. Be serious.
It might be cheaper to just get more Uranium, but isn't that the mentality that got us here in the first place? Don't worry about efficiently using what you've go, it's cheaper to just drill for more oil....
The fact is, something needs to be done. We would be very wise to partition our nuclear waste (separate out Cs-137, St-90, and Tc-99, and a few others) so that we can separate the short lived isotopes from the long lived ones. This way the short lived isotopes can be allowed to decay, and the long lived ones can be stored safely as they will not produce much heat.
In addition, we really should separate out the Uranium and Plutonium from waste if for no other reason than that they can cause criticalities in waste storage, and thus have special needs. If you don't want to put them back in a reactor, fine, but store them separately so that if we change our mind at some point we at least have that option.
As for reprocessing producing more waste than it takes in, that is not an inherent property of reprocessing, but implementation specific. Just becaus the brits (no offense) can't do it correctly with a 30 year old plant doesn't mean it can't be done.
Thermal depolymerization is pretty much a blanket solution to our waste management problems. It will reduce any carbon based waste (pretty much) to the equivalent of crude oil. This means all our trash/ag waste/sewage, etc... could be decomposed. It might even negate the need for recycling, as you could just depolymerize everything, then separate out the glass and metal at the end, the plastic would be returned to oil, and all the trash is gone.
New york city was considering a similar process for dealing with its trash. If this process was used for most of our waste, then it's not hard to imagine that it could make up for our oil imports. We import ten million barrels of oil a day, there are three hundred million people. You don't think that 30 people produce a barrel of oil worth of trash (and agricultural waste, probably in higher volume than urban trash)?
This is a huge step in the right direction. A great way to rid ourselves of our wastes, and possibly help get the fuel we need.
This was only a pilot plant. It's not hard to imagine that other plants hundreds of times larger could be built. For instance, if you constructed one to use all of NYC's trash, that would make a significant dent in the US's oil consumption.
What we need is so simple, DNS security. The root servers have private keys which are well known. They hold public keys for the tld servers. When you look up your tld server, get the public key too. Tld servers hold public keys for lower DNS servers, etc... recursive system, etc... This has several advantages.
1) No more public key madness. Everyone's public keys are part of the DNS system if you have a domain name. Simple, easy. Everything can be ssl with the press of a button, no need to setup keys.
2) Now require that the sender of email signs the email with the appropriate (as determined by DNS) key.
Simple, easy, problem solved. No more email spoofing, no more certificate BS. WHen you get a domain name, you register your public key (for free, presumably) and you're done.
So you might ask, why hasn't this been done before? The answer has something to do with the fact that Verisign controls the TLD servers, and makes a killing off of selling Certs. So if this caught on, no need to pay for certs, and that's bad for Verisign, so they'll torpedo any such proposal.
This idea has been around for a long time, perhaps it should finally be implemented this time.
Re:Fission and coal, if we have to
on
Out of Gas
·
· Score: 1
First of all, the US is fantastically lucky on these scores. We are basically the Saudi Aribia of both coal and Uranium, having some huge fraction of the world's supply (50%?) of each.
I've already posted most of this on johnkerry.com under the heading "Nuclear Power?" but here goes the abridged version.
1) Oil isn't the problem, energy is the problem. We're an advanced society, and our technology is sufficient to make fuel in any form we need provided we have the energy to start with. We can make hydrogen from water if it comes to it.
2) Current world energy usage is just a drop in the bucket. As china and india become industrialized the worldwide energy usage will soar, regardless of efficiency measures taken. End of story.
3) Current energy usage is unsustainable through fossil fuels with the load we have right now. It will be completely hopeless in the near future. It remains to be seen whether environmental damage or supply shortages will end the fossil fuel age, but one or the other will become a serious factor very soon (within the decade, if not already).
4) Coal can last longer than oil, but probably only with severe environmental consequences. Even so, it is still insufficient for the long term, and will probably run out within the century if a sizeable fraction of the world gets pulled out of poverty.
5) This leaves very few energy sources capable of providing for worldwide demand (somewhere between 10^20 and 10^21 joules per year, probably). Nuclear is one of them (well over 10^23 joules of nuclear fuel readily available, all remaining fossil fuels are something on the order of 10^22 joules), solar is another (10^27 joules of sunlight per year).
I will not speak on the virtues and dangers of nuclear power (though I think it has more of the former than the latter), but it seems inevitable unless there's a MAJOR breakthrough in fusion immediately.
Now, this being said, many things can help. Thermal depolymerization of pretty much all of our wastes (trash, sewage, agricultural and industrial wastes, etc...) will be able to help immensely, as will intelligent incorporation of solar into out buildings, etc... but I don't think it'll be enough to eliminate the need for centralized generation of cheap power.
Nuclear plants in the US produced electricity for $.04 per kWh last year, far cheaper than anything else, except for coal and hydro.
OK, IAAP (I am a physicist) and I'll tell you flat out that second order effects only become really significant when the radiation is VERY powerful, and can cause nasty showers. Most of the cosmic radiation showers don't come anywhere close to hitting the earth's surface, so this won't be an issue. Now if you were flying at 60,000 feet, who knows, it might almost be relevant, but even then I wouldn't count on it.
You tend to get secondary showers when you have things like proton beams at huge energies hitting targets (or high energy beta radiation, etc....). The secondary showers are worse than the original beam, but only if the shielding is "thin". There is very little reason to believe that our atmosphere (dozens of miles thick) would count as "thin"
The direct/secondary effects logic you propose is almost entirely opposite from what I would tend to say. I'd say that usuall direct effects are the operative force, and only within certain corner cases to secondary effects become prominent, in general.
In any case, the vast majority of cosmic radiation would be stopped by the atmosphere, magnetic field or no.
That may be true, but with somewhat better energy storage the advantage will reverse. In addition, if they allowed you to plug in your car at home you could probably get a few miles out of the batteries before needing the gasoline engine at all, which might cut your fuel use in half, in especially advantageous cases.
It's likely that flywheels will replace batteries soon enough, and that would radically improve the hybrid cars.
Basically, the batteries are the weak link, but even lugging around hundreds of pounds of lead the cars are roughly comparable to what a normal car would be. When the day comes that the lead becomes graphite or Li+ batteries, they will perform much better.
In addition, electric (and hybrid) cars have two huge advantages.
1) They get to use electric motors, which are vastly better than IC motors in essentially every way.
2) They can be partially "refueled" with electricity, which is much more flexible (cheaper and easier to get) than gas. Granted, we need clean sources of electricity (nuclear is the obvious choice*), but at least standardize on one energy source that can be produced from anything, and then worry about how to best produce it. While we use oil we can't really do any better because we can't (easily) make oil out of other energy sources.
* I am a physicist, so I am biased. However, by my back of the envelope calculations, assuming power plants are about 50% efficient, and we get all our energy from nuclear (everything that currently runs on oil or coal now (including industry, cars) uses electricity instead) then we can provide for all our needs with less than 10,000 tons of Uranium a year. This would produce somewhat less than 10,000 tons of fission fragments (some of them would decay within hours or days, others within a few years). This ammount of material would easily fit in a convoy of trucks (about 100 perhaps, not sure how much a Semi can haul) or all but the smallest cargo ship. It seems radically better than releasing billions of tons of gunk into the atmosphere and dynomiting whole regions of the country to get coal.
OK, first of all, comparing Diesel and Gasoline mileage isn't really apples to apples. They do after all use different fuel, and I wouldn't be surprised to find the energy density of diesel to be higher.
Secondly, the primary reason diesels aren't popular is pretty much based on pollution. They just aren't as clean as gasoline based cars, and that's that. Good low sulfur diesel fuel (coming in 2012, a little late methinks) will help that, but for now, there's not much to be done.
OK, this is not insightful. Clearly the system isn't going to be a glorified web page. They said it'd work offline, so I don't see dial up connections being that big of a problem. Cache the software on the client, check for latest versions over the web, etc....
It seems clear that it's going to be a java webstart like program, and thank god too. I have long thought that if we could get all our office software in webstart form there would be very little need for most of the techies who run around updating software and disinfecting machines, which is most of them.
I'm not fond of the subscription model, but my god this almost can't help but be a step up from the $400/seat microsoft system that also bombards you with virii and a slew of software that doesn't work, or is incompatible, etc.....
A work computer should be the following things..... 1) A cpu. 2) Ram 3) graphics card 4) usb/firewire connections 5) nothing else.
It should have a little bit of firmware that lets it netboot off of a central server (copy down the OS image, boot up normally) and all the "drives" shoud just be netshares from the central file servers. No BS about users harddrives crashing, or viruses, or other nastiness. THe machine is pristine on each and every boot up, the data is backed up and protected, the apps are always the correct version, etc... And these things should cost much less than the average workplace workstation.
Of course for the security conscious (should be everyone) the firmware should keep the fingerprint of the server's public key so no spoofing would be possible. The techies could enter it in by hand or at least verify it when the thing first boots up, and then you're secure from there on out. No more business data spread over dozens of computers, etc. You just sit down at any computer, log in, and voila, it's your environment, your apps, etc., and it's all running locally (not citrix nastiness).
It would also be nice to have the phones be a program on the computer, give people headsets. Then there's no need for a phone either, you can hear the phone ring even when wearing headphones, and your phone follows you around. Whenever you sit down and log in, your phone is there. Log out, and you're known to be away. Redirect missed calls to email, life is good.
The network computer never worked because people wanted to offload calculation, even though calculation is cheap and bandwidth is expensive. That is not the correct way to go. Offload storage, let the computer do its own calculation, and you've got a winner.
Needed query plan is two index scans and a merge join, should be obvious. With huge datasets I don't join on any columns that aren't indexed, so it should always index scan and merge join. You can make it do this by turning off seqscan and hash join. This improves the speed by almost a full order of magnitued (5+ times). In addition, a hash join that big pretty much locks up the box because it makes it thrash so badly. No thank you. Not only is the merge join much faster, but the box remains responsive.
The real problem with hash joins is lack of locality, a problem that sort joins don't have. So hash joins generally take a substantial performance hit because you're fetching each page in multiple times. In addition, real men index their columns that they expect to join on. With several billion rows (once we fill it up) there is no other option.
I'd even go so far as to say that a database should add an index to any column that's used in a join if the tables in question are each multi million rows (the size of our test database that this data comes from. Would be much worse on the production one). Any other query plan is simply never going to finish.
OK, hmmm, lets seee. I vacuume all the freaking time. Not that this is really relevant, as with the 5-10 billion rows we'll be looking at, it looks like we won't be able to vacuum more than once a week, if that.
And I did edit the postgres.conf. I had to actually turn off the Hash joins and the sequential scan, in addition to adjusting all the buffer sizes. Both of them are simply not an option for a db of that size. And yes, it does run about 5 times faster now. Don't you think it's a little broken when turning off some optimizations gets almost a full order of magnitude speed improvement. Reading the query plans the problem is obvious. Postgres doesn't keep track of the number of unique keys in a database (or most other things about the database) which is perfectly possible to do with B-trees. As it doesn't know how many unique keys, or how many keys there are within an interval, it cannot accurately guess the nubmer of rows returned by even a simple query. So it selects out of a multi million row table, thinks a hash join is feasible, then gets killed when half the table comes back. Open and shut.
You're out of your league here buddy. Go back under your bridge. Return if you learn a little about databases.
I had a paper sitting around that discussed good escape analysis, can't seem to find a link to it now. Anyway, it suggested that the average performance gain was something like 20-40%, though that was mostly from synchronization elimination. Anyway, I can't imagine that the performance hit for escape analysis matters for a Server VM. When I'm starting up Tomcat it already takes two minutes to start. I don't really care if it takes ten minutes, if it makes it 40% faster. Seems to be true of most things. For instance, I don't restart jedit all that often, or Squirrel, or most of my other java apps.
Anyway, pretty much agree with everything you have to say, though I think the startup time thing is an old saw. People wouldn't really care that much if their apps take a long time to start, so long as they run fast once they're started, I know I don't. This is within reason of course. A minute or two isn't devastating, especially if you can take most of it at boot time, or take it in bits and pieces as the app gets exercised.
Take this from a physics major who decided not to go into grad school. I now work in Comp. Sci. on wall street.
There are several problems, but the primary one is that the sciences and engineering doesn't pay enough. As far as Comp. Sci. goes, I've never seen any indication that grad school in comp sci is anything other than useless. A BA is already highly overqualified for pretty much any job, if you actually love the field, and you'll be underqualified if you don't love the field, even if you have a PhD. I've seen some sucky work coming from comp. sci. PhDs.
Basically I could go to grad school, but it couldn't possibly increase my earning power, and I'd have to pony up more than 100,000 and five years to get through, it's just not worth it. The reason it isn't worth it is because the country has been flooded with foreign students for whome it is worth it, so all the citizens have decided to take another path.
It also doesn't help that the physics research I did was horribly mismanaged. The place I work now is a joy, whereas my physics research was horrifying. Can nobody in the sciences actually manage a project anymore?
This can be done with escape analysis, which java currently doesn't do, but hopefully will in the future. Basically, what happens is that the bytecode compiler analyzes the program and detects all pointers that don't leak, basically the rules are like this...
A pointer in a function is "stackable" if it is...
1) Never assigned to a static variable. 2) Never assigned to a member variable of a non-stackable class. 3) Never passed as a non-stackable argument to a function.
And there might be other conditions as well, some function arguments might be semi-stackable, that is, stackable only if the pointer to the object itself is stackable, and no non-stackable variable is ever assigned from a member of the object. A semi-stackable pointer can be assigned to any member of the "this" pointer.
In this way, for instance, Map.add(object o) would probably have a semi-stackable argument o, Comparable.compareTo(object o) should have a stackable pointer for o....
At this point, all stackable objects can be allocated on the stack, and no garbage collection is needed for them. Just run their finalizers when the stack frame returns and discard them. You probably also have to check that the finalizer can't ressurect the object, but that should be rare.
That should wipe out the advantage of value objects, and reduce the amount of garbage to be collected, probably cut it in half.
Another possibility is to consider (I don't know the technical term for this...) object compounding. Basically any object that is internal to another object should just be inlined inside it. That would reduce the total number of objects by 1, and the pointer to the new object (someObject.blah) is just calculated as (someObject.this + blah_offset). As long as nobody can ever get this pointer outside of someObject, you don't have to track it separately. This could probably greatly reduce the number (but not the total size) of live objects, and thus accelerate garbage collection.
Basically, what I'm getting at is that intelligent algorithms would have very little to gain (and a lot to lose) by listening to human input.
Slashdot Mirror
Netbeans 3.6 is quite nice as well. I agree with the parent that it isn't really that slow. It's fairly slow on my OS X box (which is a G5, so there's no excuse, you hear me Apple), but at work on windows and linux it's snappy enough that I can't tell it isn't native. I really wish SUN would just bite the bullet and provide a good profiler with netbeans though. They already include a debugger, why not a profiler too? It seems like netbeans is to the level where it really just needs some additional plugins, it already pretty much surpasses visual studio in terms of usefulness as far as I'm concerned. In addition, be sure to run it on a modern VM, the newer JVMs have pretty substantially improved SWING performance. Apple's offering seems to be the weakest in this area, but it's made the most improvement from the early days, so maybe by 1.5 they'll be competitive.
It has a stronger foothold everywhere the GDP is lower. The more expensive M$ licenses are compared to the cost of a worker, the more enticing opensource is.
Well, in addition, lets talk about STL and pre-processor macros. Pre processor macros and STL wildly increase the volume of code generated for a program. If I had to guess, I'd say that very little of a modern program's code was actually written by humans. There is just no way people could write so much crap that a word processor would have 100 megs of code.
.NET (or pretty much anything else) than there is in C or C++, and it shows.
This is (one reason) why higher level languages are so nice. They give you MUCH smaller executables, as everyone has standardized on the same libraries, namely those that came with the environment. There is just much more code sharing in Java or
I'm currently reorganiznig about 300 gigs of data (zipping and unzipping), I think that counts.
Good point. I always just figured that the hydrogen storage tanks could be ultra high pressure segmented tanks, but that might be a tough sell for aircraft, etc.... On of the advantages that you have with hydrogen, is that it isn't as easy to destroy things with it. It tends to rise so rapidly that it doesn't easily burn people (who tend to be on the ground). Also, it doesn't contain soot, so it doesn't radiate heat very efficiently, which helps to cut down on burns. Might be useful in military situations, but maybe overshadowed by the high pressure tank problems.
don't be foolish. Any idiot will just tell you to run everything on Hydrogen, which you can make from the electricity. So in a way, everything could be nuclear powered.
Speaking of which.....
I haven't heard much about it yet, but I wouldn't be surprised if there were big projects in the pentagon to consider hydrogen powered tanks and planes and warships wherever possible. If you had a nuclear flagship (aircraft carrier) it could use its reactors to generate all the hydrogen it needs. Then it doesn't have to carry fuel for the fighters, the other escort ships (currently diesel) wouldn't run out of fuel, etc.... You would completely eliminate the fuel costs and weaknesses from the equation. Same thing for tanks. Just park a carrier or other nuclear vessel nearby, make hydrogen from the water it's sitting in, and have all the fuel you need to run your land campaign.
Also, fuel cells would probably get better mileage than the standard parts for tanks and ships at least. Aircraft might be much tougher.
true enough, but it's hardly required to have a civilian nuclear program in order to have weapons. As exhibit A, I give the United States of America. During the 1940s this country has no civilian nuclear program to speak of, and yet still managed to become a nuclear power (using 50 year old technology, and blazing the trail all the way) in about a decade, or less, depending on how you count.
I'm not arguing that civilian fuel can't be used to make weapons, I'm just arguing that having a civilian program doesn't imply that weapons will also be made (or available). Since you took a nuclear physics course, we both know that reprocessing nuclear fuel is difficult for the unprepared, and releases detectable krypton. It's probably much harder to get away with nuclear reactors + reprocessing than it is to get away with Uranium enrichment. Especially because the reactors need enriched Uranium anway, so why not just enrich it a few more times while you're at it and dispense with the reactors entirely.
So though civilian plants are a path to nuclear weapons, they are probably much tougher than just brute forcing it using a good (or even a sucky) Uranium diffusion plant. Now the US did use plutonium producing piles, we also were a Huge country with very little to fear from anyone and plenty of room to hide the program. In addition, technology wasn't as advanced as it now is. I imagine that modern laser diffusion (not sure of the technical name of this) or gaseous diffusion isn't even very difficult anymore, whereas it would have been very tough 50 years ago.
Ok, this is nice, but neither side gives any evidence. Since when does "no it isn't" count as a refutation?
Everything that guy has to say is about nuclear weapons. Well, guess what. WE ALREADY HAVE NUCLEAR WEAPONS. There, accept it. Get over it. There is no danger of additional reactors turning the US, or China, or India, or Western Europe into nuclear armed powers. NONE, because they already are.
It's easy to tear down someone else's proposal when you don't have on of your own and need rely on nothing but juvenile comebacks. Get some actual evidence. And you know what, even if you count the victims of Hiroshima and Nagisaki against nuclear power (but don't count the victims of conventional warfare against fossil fuels) and you throw in Cherenoble, and maybe round everything up by a few hundred thousand just to be sure, Nuclear killed far fewer people per kWh of energy. It is almost impossible to imagine a scenario in which it might be otherwise. Fossil fuels kill tens (hundreds, depending on how you count) of thousands of people each year.
A nuclear disaster would have to kill tens of millions (at least) in order to even the score. Nobody can even conceive of how that could happen with civilian reactors built to even the most incompetent of standards, like Cherenobl. About the only real possibility is if WW-III breaks out and people start tossing around nuclear weapons (which they already have, and don't need civilian reactors for), and that is far MORE likely if we start fighting over oil.
Just once I'd like to hear a well reasoned out anti-nuclear position. Include some numbers (you know, dollars and cents, lives lost, that sort of thing) and keep them accurate. Include an honest asessment of nuclear waste dangers (assuming various means of disposal) and honest asessments of nuclear proliferation. I have never seen any evidence that civilian nuclear power leads to proliferation, but it seems to be a given for the anti-nuke types. Japan and South Korea both have reactors, and neither has nuclear weapons.
The only scenario the anti-nuke types ever argue against is such a complete straw man. They assume we dump all the nuclear waste into the nation's beer supply, give away spent fuel to everyone with a driver's license, and somehow (though nobody can really imagine exactly how this happens) have lots of melt downs in highly populated areas. Seriously. Assume an even marginally competent nuclear program (needn't be perfect) and then try a comparison with our fossil fuel system. See how that treats you.
It's like comparing against an oil economy where it's assumed that 99% of the oil is dumped raw into the ocean, the rest is burned in the foulest, dirtiest machines imaginable, and that somehow access to oil allows every fool who can rub two sticks together to build a jet fighter with which to kill people. Be serious.
The joke was from an old onion article, relating to cloning though.
Apparently everyone skipped that issue. More precisely it would be far better to keep this from the environmentalists than the christians.
Your karma whoring worked better than mine.
Nothing like a "meltdown" can occur in modern fission reactors either, but that doesn't stop fools from pontificating on the subject.
This could be a useful technology, please, please, please, nobody tell the christians about it.
It might be cheaper to just get more Uranium, but isn't that the mentality that got us here in the first place? Don't worry about efficiently using what you've go, it's cheaper to just drill for more oil.... The fact is, something needs to be done. We would be very wise to partition our nuclear waste (separate out Cs-137, St-90, and Tc-99, and a few others) so that we can separate the short lived isotopes from the long lived ones. This way the short lived isotopes can be allowed to decay, and the long lived ones can be stored safely as they will not produce much heat. In addition, we really should separate out the Uranium and Plutonium from waste if for no other reason than that they can cause criticalities in waste storage, and thus have special needs. If you don't want to put them back in a reactor, fine, but store them separately so that if we change our mind at some point we at least have that option. As for reprocessing producing more waste than it takes in, that is not an inherent property of reprocessing, but implementation specific. Just becaus the brits (no offense) can't do it correctly with a 30 year old plant doesn't mean it can't be done.
Sorry, you are wrong.
Thermal depolymerization is pretty much a blanket solution to our waste management problems. It will reduce any carbon based waste (pretty much) to the equivalent of crude oil. This means all our trash/ag waste/sewage, etc... could be decomposed. It might even negate the need for recycling, as you could just depolymerize everything, then separate out the glass and metal at the end, the plastic would be returned to oil, and all the trash is gone.
New york city was considering a similar process for dealing with its trash. If this process was used for most of our waste, then it's not hard to imagine that it could make up for our oil imports. We import ten million barrels of oil a day, there are three hundred million people. You don't think that 30 people produce a barrel of oil worth of trash (and agricultural waste, probably in higher volume than urban trash)?
This is a huge step in the right direction. A great way to rid ourselves of our wastes, and possibly help get the fuel we need.
This was only a pilot plant. It's not hard to imagine that other plants hundreds of times larger could be built. For instance, if you constructed one to use all of NYC's trash, that would make a significant dent in the US's oil consumption.
What we need is so simple, DNS security. The root servers have private keys which are well known. They hold public keys for the tld servers. When you look up your tld server, get the public key too. Tld servers hold public keys for lower DNS servers, etc... recursive system, etc... This has several advantages.
1) No more public key madness. Everyone's public keys are part of the DNS system if you have a domain name. Simple, easy. Everything can be ssl with the press of a button, no need to setup keys.
2) Now require that the sender of email signs the email with the appropriate (as determined by DNS) key.
Simple, easy, problem solved. No more email spoofing, no more certificate BS. WHen you get a domain name, you register your public key (for free, presumably) and you're done.
So you might ask, why hasn't this been done before? The answer has something to do with the fact that Verisign controls the TLD servers, and makes a killing off of selling Certs. So if this caught on, no need to pay for certs, and that's bad for Verisign, so they'll torpedo any such proposal.
This idea has been around for a long time, perhaps it should finally be implemented this time.
First of all, the US is fantastically lucky on these scores. We are basically the Saudi Aribia of both coal and Uranium, having some huge fraction of the world's supply (50%?) of each.
I've already posted most of this on johnkerry.com under the heading "Nuclear Power?" but here goes the abridged version.
1) Oil isn't the problem, energy is the problem. We're an advanced society, and our technology is sufficient to make fuel in any form we need provided we have the energy to start with. We can make hydrogen from water if it comes to it.
2) Current world energy usage is just a drop in the bucket. As china and india become industrialized the worldwide energy usage will soar, regardless of efficiency measures taken. End of story.
3) Current energy usage is unsustainable through fossil fuels with the load we have right now. It will be completely hopeless in the near future. It remains to be seen whether environmental damage or supply shortages will end the fossil fuel age, but one or the other will become a serious factor very soon (within the decade, if not already).
4) Coal can last longer than oil, but probably only with severe environmental consequences. Even so, it is still insufficient for the long term, and will probably run out within the century if a sizeable fraction of the world gets pulled out of poverty.
5) This leaves very few energy sources capable of providing for worldwide demand (somewhere between 10^20 and 10^21 joules per year, probably). Nuclear is one of them (well over 10^23 joules of nuclear fuel readily available, all remaining fossil fuels are something on the order of 10^22 joules), solar is another (10^27 joules of sunlight per year).
I will not speak on the virtues and dangers of nuclear power (though I think it has more of the former than the latter), but it seems inevitable unless there's a MAJOR breakthrough in fusion immediately.
Now, this being said, many things can help. Thermal depolymerization of pretty much all of our wastes (trash, sewage, agricultural and industrial wastes, etc...) will be able to help immensely, as will intelligent incorporation of solar into out buildings, etc... but I don't think it'll be enough to eliminate the need for centralized generation of cheap power.
Nuclear plants in the US produced electricity for $.04 per kWh last year, far cheaper than anything else, except for coal and hydro.
OK, IAAP (I am a physicist) and I'll tell you flat out that second order effects only become really significant when the radiation is VERY powerful, and can cause nasty showers. Most of the cosmic radiation showers don't come anywhere close to hitting the earth's surface, so this won't be an issue. Now if you were flying at 60,000 feet, who knows, it might almost be relevant, but even then I wouldn't count on it.
You tend to get secondary showers when you have things like proton beams at huge energies hitting targets (or high energy beta radiation, etc....). The secondary showers are worse than the original beam, but only if the shielding is "thin". There is very little reason to believe that our atmosphere (dozens of miles thick) would count as "thin"
The direct/secondary effects logic you propose is almost entirely opposite from what I would tend to say. I'd say that usuall direct effects are the operative force, and only within certain corner cases to secondary effects become prominent, in general.
In any case, the vast majority of cosmic radiation would be stopped by the atmosphere, magnetic field or no.
I'm impressed. This actually sounds fairly reasonable, as theology goes.
Being a bit of a dieist myself I don't really see the same theological problems, but it's a good attempt to grapple with them regardless.
That may be true, but with somewhat better energy storage the advantage will reverse. In addition, if they allowed you to plug in your car at home you could probably get a few miles out of the batteries before needing the gasoline engine at all, which might cut your fuel use in half, in especially advantageous cases.
It's likely that flywheels will replace batteries soon enough, and that would radically improve the hybrid cars.
Basically, the batteries are the weak link, but even lugging around hundreds of pounds of lead the cars are roughly comparable to what a normal car would be. When the day comes that the lead becomes graphite or Li+ batteries, they will perform much better.
In addition, electric (and hybrid) cars have two huge advantages.
1) They get to use electric motors, which are vastly better than IC motors in essentially every way.
2) They can be partially "refueled" with electricity, which is much more flexible (cheaper and easier to get) than gas. Granted, we need clean sources of electricity (nuclear is the obvious choice*), but at least standardize on one energy source that can be produced from anything, and then worry about how to best produce it. While we use oil we can't really do any better because we can't (easily) make oil out of other energy sources.
* I am a physicist, so I am biased. However, by my back of the envelope calculations, assuming power plants are about 50% efficient, and we get all our energy from nuclear (everything that currently runs on oil or coal now (including industry, cars) uses electricity instead) then we can provide for all our needs with less than 10,000 tons of Uranium a year. This would produce somewhat less than 10,000 tons of fission fragments (some of them would decay within hours or days, others within a few years). This ammount of material would easily fit in a convoy of trucks (about 100 perhaps, not sure how much a Semi can haul) or all but the smallest cargo ship. It seems radically better than releasing billions of tons of gunk into the atmosphere and dynomiting whole regions of the country to get coal.
OK, first of all, comparing Diesel and Gasoline mileage isn't really apples to apples. They do after all use different fuel, and I wouldn't be surprised to find the energy density of diesel to be higher. Secondly, the primary reason diesels aren't popular is pretty much based on pollution. They just aren't as clean as gasoline based cars, and that's that. Good low sulfur diesel fuel (coming in 2012, a little late methinks) will help that, but for now, there's not much to be done.
OK, this is not insightful. Clearly the system isn't going to be a glorified web page. They said it'd work offline, so I don't see dial up connections being that big of a problem. Cache the software on the client, check for latest versions over the web, etc....
It seems clear that it's going to be a java webstart like program, and thank god too. I have long thought that if we could get all our office software in webstart form there would be very little need for most of the techies who run around updating software and disinfecting machines, which is most of them.
I'm not fond of the subscription model, but my god this almost can't help but be a step up from the $400/seat microsoft system that also bombards you with virii and a slew of software that doesn't work, or is incompatible, etc.....
A work computer should be the following things.....
1) A cpu.
2) Ram
3) graphics card
4) usb/firewire connections
5) nothing else.
It should have a little bit of firmware that lets it netboot off of a central server (copy down the OS image, boot up normally) and all the "drives" shoud just be netshares from the central file servers. No BS about users harddrives crashing, or viruses, or other nastiness. THe machine is pristine on each and every boot up, the data is backed up and protected, the apps are always the correct version, etc... And these things should cost much less than the average workplace workstation.
Of course for the security conscious (should be everyone) the firmware should keep the fingerprint of the server's public key so no spoofing would be possible. The techies could enter it in by hand or at least verify it when the thing first boots up, and then you're secure from there on out. No more business data spread over dozens of computers, etc. You just sit down at any computer, log in, and voila, it's your environment, your apps, etc., and it's all running locally (not citrix nastiness).
It would also be nice to have the phones be a program on the computer, give people headsets. Then there's no need for a phone either, you can hear the phone ring even when wearing headphones, and your phone follows you around. Whenever you sit down and log in, your phone is there. Log out, and you're known to be away. Redirect missed calls to email, life is good.
The network computer never worked because people wanted to offload calculation, even though calculation is cheap and bandwidth is expensive. That is not the correct way to go. Offload storage, let the computer do its own calculation, and you've got a winner.
Needed query plan is two index scans and a merge join, should be obvious. With huge datasets I don't join on any columns that aren't indexed, so it should always index scan and merge join. You can make it do this by turning off seqscan and hash join. This improves the speed by almost a full order of magnitued (5+ times). In addition, a hash join that big pretty much locks up the box because it makes it thrash so badly. No thank you. Not only is the merge join much faster, but the box remains responsive.
The real problem with hash joins is lack of locality, a problem that sort joins don't have. So hash joins generally take a substantial performance hit because you're fetching each page in multiple times. In addition, real men index their columns that they expect to join on. With several billion rows (once we fill it up) there is no other option.
I'd even go so far as to say that a database should add an index to any column that's used in a join if the tables in question are each multi million rows (the size of our test database that this data comes from. Would be much worse on the production one). Any other query plan is simply never going to finish.
OK, hmmm, lets seee. I vacuume all the freaking time. Not that this is really relevant, as with the 5-10 billion rows we'll be looking at, it looks like we won't be able to vacuum more than once a week, if that.
And I did edit the postgres.conf. I had to actually turn off the Hash joins and the sequential scan, in addition to adjusting all the buffer sizes. Both of them are simply not an option for a db of that size. And yes, it does run about 5 times faster now. Don't you think it's a little broken when turning off some optimizations gets almost a full order of magnitude speed improvement. Reading the query plans the problem is obvious. Postgres doesn't keep track of the number of unique keys in a database (or most other things about the database) which is perfectly possible to do with B-trees. As it doesn't know how many unique keys, or how many keys there are within an interval, it cannot accurately guess the nubmer of rows returned by even a simple query. So it selects out of a multi million row table, thinks a hash join is feasible, then gets killed when half the table comes back. Open and shut.
You're out of your league here buddy. Go back under your bridge. Return if you learn a little about databases.
I had a paper sitting around that discussed good escape analysis, can't seem to find a link to it now. Anyway, it suggested that the average performance gain was something like 20-40%, though that was mostly from synchronization elimination. Anyway, I can't imagine that the performance hit for escape analysis matters for a Server VM. When I'm starting up Tomcat it already takes two minutes to start. I don't really care if it takes ten minutes, if it makes it 40% faster. Seems to be true of most things. For instance, I don't restart jedit all that often, or Squirrel, or most of my other java apps.
Anyway, pretty much agree with everything you have to say, though I think the startup time thing is an old saw. People wouldn't really care that much if their apps take a long time to start, so long as they run fast once they're started, I know I don't. This is within reason of course. A minute or two isn't devastating, especially if you can take most of it at boot time, or take it in bits and pieces as the app gets exercised.
Take this from a physics major who decided not to go into grad school. I now work in Comp. Sci. on wall street.
There are several problems, but the primary one is that the sciences and engineering doesn't pay enough. As far as Comp. Sci. goes, I've never seen any indication that grad school in comp sci is anything other than useless. A BA is already highly overqualified for pretty much any job, if you actually love the field, and you'll be underqualified if you don't love the field, even if you have a PhD. I've seen some sucky work coming from comp. sci. PhDs.
Basically I could go to grad school, but it couldn't possibly increase my earning power, and I'd have to pony up more than 100,000 and five years to get through, it's just not worth it. The reason it isn't worth it is because the country has been flooded with foreign students for whome it is worth it, so all the citizens have decided to take another path.
It also doesn't help that the physics research I did was horribly mismanaged. The place I work now is a joy, whereas my physics research was horrifying. Can nobody in the sciences actually manage a project anymore?
This can be done with escape analysis, which java currently doesn't do, but hopefully will in the future. Basically, what happens is that the bytecode compiler analyzes the program and detects all pointers that don't leak, basically the rules are like this...
A pointer in a function is "stackable" if it is...
1) Never assigned to a static variable.
2) Never assigned to a member variable of a non-stackable class.
3) Never passed as a non-stackable argument to a function.
And there might be other conditions as well, some function arguments might be semi-stackable, that is, stackable only if the pointer to the object itself is stackable, and no non-stackable variable is ever assigned from a member of the object. A semi-stackable pointer can be assigned to any member of the "this" pointer.
In this way, for instance, Map.add(object o) would probably have a semi-stackable argument o, Comparable.compareTo(object o) should have a stackable pointer for o....
At this point, all stackable objects can be allocated on the stack, and no garbage collection is needed for them. Just run their finalizers when the stack frame returns and discard them. You probably also have to check that the finalizer can't ressurect the object, but that should be rare.
That should wipe out the advantage of value objects, and reduce the amount of garbage to be collected, probably cut it in half.
Another possibility is to consider (I don't know the technical term for this...) object compounding. Basically any object that is internal to another object should just be inlined inside it. That would reduce the total number of objects by 1, and the pointer to the new object (someObject.blah) is just calculated as (someObject.this + blah_offset). As long as nobody can ever get this pointer outside of someObject, you don't have to track it separately. This could probably greatly reduce the number (but not the total size) of live objects, and thus accelerate garbage collection.
Basically, what I'm getting at is that intelligent algorithms would have very little to gain (and a lot to lose) by listening to human input.
Deuterium-Deuterium fusion doesn't produce neutrons.
However, it is much tougher than Deuterium-Tritium fusion (don't know numbers off hand, requires maybe 1/2 more heat and pressure, just a guess).