There are new table saws that only make a small cut in your finger if you try to cut it off, maybe future designs can make it as safe to use as a plastic butter knife.
Oh yeah, I forgot about that. But I'd bet good money that this saw's manual comes with *more* warnings than a normal saw. (WARNING: Never use your actual finger to test failsafe mechanism! CAUTION: Electrical interference may cause false triggering!, etc.)
Any statements in the documentation that start out "Don't" or marked "Warning" or "Notice" are always present because of flaws -- the right approach is to fix the software (and remove the statement from the documentation).
Let's apply that to other products in the real world. How about a table saw: that's covered with warning stickers and the instruction manual is full of safety notices. These are all flaws, and we'll change the saw's design to remove them one by one. At the end of the process, we'll have a flawless and user friendly cutting tool: a plastic butter knife.
No thanks, I'll take my powerful but "dangerous" software over dumbed down pablum.
For everything else, that is, using a computer, it's back to binary.
It is not. RAM is the only quantity in computers commonly measured in binary. Hard drives have always been in decimal. Floppies have always been in an even more stupid system where "MB" == 1000*1024. Clock speeds have always been decimal.
Going farther, measuring IO or network performance, to cite two trivial examples, or understanding any of those subjects in general, you're binary to binary.
You appear to have been bambooozled yourself by the confusion caused by this issue. I/O speed of buses is always decimal because it derives from MHz and GHz, which are decimal. Network bandwidth is more often measured in decimal megabits, not binary.
You seem to think that just because one user app, Windows Explorer, confusingly shows binary based quantities, then everything else in the computer is or ought to be measured that way as well. You're incorrect.
I don't see why learning powers of two, and then extending that (for the "power users") to base 16, is unreasonable.
If you were advocating that people learn and work in pure hexadecimal, you might have a point. However, these units aren't a consistent radix. They're a strange mishmash of binary and decimal based on the accident that 2**10 is somewhere close to 10**3. They have completely different math for each of KiB, MiB, GiB, etc. You're telling people that they need to work with four or more distinct new number systems, and be prepared to convert between any and all of them, depending on approximately how much data they're working with. That's just stupid.
Everyone other than hard drive manufacturers has been using the SI prefixes to refer to power of two quantities when referring to binary data for 40 years. Attempting to redefine them retroactively just causes confusion.
No, the confusion is cause by using a pseudo-binary based number system in a world where almost everything else is decimal.
Quick question: You have a 2000 MiB video file and a 2470 MiB video file. Will they both fit on a 4.37 GiB DVD? Now you need your calculator.
It's much easier to figure out if a 2097 MB and a 2590 MB file fit on a 4.7 GB disk. You can do that in your head.
I've been burned numerous times by programs ambiguously reporting sizes in KiB and MiB causing me to run out of space on something that I'm trying to fill. All storage sizes should always be reported in decimal numbers. If RAM manufacturers want to keep using powers of two due to the implementation detail of how their chips are constructed, they should *always* use KiB, MiB and GiB.
Well, the airlines were all set to do this. But when they checked the per-minute rates on the back of their Airfone handsets, they realized that keeping a call open for the duration of each flight would bankrupt them.
but if you can interact with it then it's not a problem, because once it interacts with something then it's gone.
With cosmic rays, it's not just "gone". Instead, you get a shower of new energetic particles generated by the collision which compounds the risk of operational errors. The patent specifically mentions alpha particles knocked out of the atoms in the chip by the ray which travel through the circuits causing havoc.
The patent also mentions that the detector may sense side effects of collision (such as voltage spikes) rather than the ray particle itself. Thus, the damage has already been done by the time the detector sees the event.
IMO, there should be a change in the definition of what's patentable that goes along the lines of your argument. Instead of being able to patent anything that's not immediately obvious to a random mediocre engineer, the test should be: "It is unlikely that this idea would have been independently developed by anyone else in the world over the entire term of the patent."
(Maybe patent terms could be variable, and you could apply for a 4 year or whatever year patent if you couldn't make a convincing argument that the idea wouldn't be independently invented in the next 20 years.)
That probably would cut the volume of patents down by 99%, while still rewarding the few people who come up with truly groundbreaking ideas.
This would destroy the GPL and probably all other copyright licenses that support FOSS.
I doubt it would actually matter much, given that most software needs to be continuously updated to remain relevant. Each update would have a fresh copyright. Proprietary "freeloaders" would necessarily be stuck with a rather stale public domain fork, and would have to independently author and maintain any updates for the software. That major hassle would probably deter such proprietary forks in most cases.
that, and I assume NORMALLY there's always several of them in the air at once, continually refueling mid air?
Not since the 1960s. After they accidentally scattered bits of H-bomb around Spain and Greenland, they stopped the continuous flights of nuclear armed bombers.
They can continue to do the same and it will be no more wrong than it is now.
But it is wrong now. The entire US healthcare risk underwriting system is wrong.
By having thousands of individual risk pool managers obsess over saving money by kicking out people who might actually use healthcare services, we ironically end up with a system that costs us almost twice as much overall as any other country, while at the same time not even covering a huge swath of the population.
Meanwhile, needlessly stupid thing like worrying about who gets a hold of medical tests causes stress for millions. Millions more are tied to their corporate jobs like feudal serfs because of fear of losing healthcare benefits.
To stop this insanity, there needs to be one single uniform national risk pool.
We are getting older. And the new deal depends on current workers to pay in to fund benefits. The ponzi scheme is falling apart.
What you say is true, but the fix is simple: make the retirement age variable, and adjust it each year it so that a fixed percentage of the population receives benefits. If you live longer on average, you should work longer on average.
Not sure I like the sound of this. A lot of areas are already having trouble with their water supplies -
The US per capita continuous total energy consumption averages out to about 10.5kW thermal (100e18 J annual national total/300M people). With hydrogen combustion at 286kJ/mol, it would take 62 liters of water per day per person to provide hydrogen for *all* the energy currently used in the US. Residential water consumption is already around 400 liters per person per day, industrial usage is more than that, and agricultural usage is many thousands of liters per day per person. IOW, this won't be a significant increase in overall water use. If it really got to be a problem, just set it up by the ocean and use some extra solar energy to run desalinization plants.
It's true that Niagra chips are probably particularly well suited to Google's workloads. However, the systems that they go into are usually considered to be "big Unix servers", not "mainframes".
Part of the question depends on how much memory bandwidth Google needs. For their search engine, IIRC they try to keep much of the index in main memory. It could be that having dozens of threads simultaneously hammering on one memory bank wouldn't work out. In that case, Google might be better off saving power by using lots of low power laptop x86 chips or ARM CPUs, each with their own memory, than a single highly parallel CPU. OTOH, that speculation could be wrong and the Sun CPUs might be perfect for them.
Assuming, of course, that the US simply don't just hire away all of their best and brightest, like it have been for a long time now.
We've been able to do that because all the money was over here. However, between trade deficits and government borrowing, we've been working really hard on sending that money over to China lately. So before long it may not make much sense for their best and brightest to come over here when they can get paid with US cash right in their own hometowns.
And yes the amount of Co2 being emitted unnaturally by humans is less then.0001% of the total green house gases. And yes, you heard that correctly, less then 1/1000 or 1 percent of the total greenhouse gases in our atmosphere at any given time.
The answer "because humans are emitting lots of CO2" doesn't cut it. Why aren't the natural causes of the past the causes today?
They still are. But you, like so many others, seem to be completely ignorant of the concept of rate of change. Humans are changing the CO2 levels orders of magnitude faster than natural factors have in the past, so those effects get lost in the noise.
It never actually gets around to explaining why these scientists don't think the ice core data throws the link into question.
If you understood the article, it should be pretty obvious that CO2 likely didn't trigger the end of the last few ice ages given that there probably weren't any large releases of CO2 like we're making now. (And before anybody gets any big ideas: Volcanoes aren't the culprit. They release a tiny fraction as much CO2 as humans.) As the article points out, the changes likely were triggered by other factors like changes in the earth's orbit.
If the CO2 didn't trigger the changes, but does participate in a positive feedback loop, then of course it would lag the temperature. But that has nothing to do at all with the question of whether an increase in CO2 levels could also trigger a warming cycle.
Well, the first article is actually a myth busting entry debunking the theory that the lag associated with the past couple of ice ages somehow proves that CO2 does not cause warming.
The second website looks to me like a highly biased collection of cargo cult science put together by people who specialize in fields like economics, not climatology.
Not really. Mainframes do batch processing of predetermined non-interactive workloads best. Google does interactive database searches with a fraction of a second latency, serves up web ads, and is trying to host traditional desktop applications via a web browser.
Mainframes have really puny CPU horsepower relative to their size, cost and power consumption. Their OSes are tuned for batch processing. Almost every compromise in mainframe design is decided in favor of uptime and transactional integrity, things for which Google has almost no use at all. They would be throwing a lot of money at solving issues they don't have if they ran mainframes, and even if they did manage to buy enough mainframes to handle their particular workload, it would probably end up using more power than they're using now.
Slashdot Mirror
Oh yeah, I forgot about that. But I'd bet good money that this saw's manual comes with *more* warnings than a normal saw. (WARNING: Never use your actual finger to test failsafe mechanism! CAUTION: Electrical interference may cause false triggering!, etc.)
Let's apply that to other products in the real world. How about a table saw: that's covered with warning stickers and the instruction manual is full of safety notices. These are all flaws, and we'll change the saw's design to remove them one by one. At the end of the process, we'll have a flawless and user friendly cutting tool: a plastic butter knife.
No thanks, I'll take my powerful but "dangerous" software over dumbed down pablum.
It is not. RAM is the only quantity in computers commonly measured in binary. Hard drives have always been in decimal. Floppies have always been in an even more stupid system where "MB" == 1000*1024. Clock speeds have always been decimal.
Going farther, measuring IO or network performance, to cite two trivial examples, or understanding any of those subjects in general, you're binary to binary.You appear to have been bambooozled yourself by the confusion caused by this issue. I/O speed of buses is always decimal because it derives from MHz and GHz, which are decimal. Network bandwidth is more often measured in decimal megabits, not binary.
You seem to think that just because one user app, Windows Explorer, confusingly shows binary based quantities, then everything else in the computer is or ought to be measured that way as well. You're incorrect.
I don't see why learning powers of two, and then extending that (for the "power users") to base 16, is unreasonable.If you were advocating that people learn and work in pure hexadecimal, you might have a point. However, these units aren't a consistent radix. They're a strange mishmash of binary and decimal based on the accident that 2**10 is somewhere close to 10**3. They have completely different math for each of KiB, MiB, GiB, etc. You're telling people that they need to work with four or more distinct new number systems, and be prepared to convert between any and all of them, depending on approximately how much data they're working with. That's just stupid.
No, the confusion is cause by using a pseudo-binary based number system in a world where almost everything else is decimal.
Quick question: You have a 2000 MiB video file and a 2470 MiB video file. Will they both fit on a 4.37 GiB DVD? Now you need your calculator.
It's much easier to figure out if a 2097 MB and a 2590 MB file fit on a 4.7 GB disk. You can do that in your head.
I've been burned numerous times by programs ambiguously reporting sizes in KiB and MiB causing me to run out of space on something that I'm trying to fill. All storage sizes should always be reported in decimal numbers. If RAM manufacturers want to keep using powers of two due to the implementation detail of how their chips are constructed, they should *always* use KiB, MiB and GiB.
Well, the airlines were all set to do this. But when they checked the per-minute rates on the back of their Airfone handsets, they realized that keeping a call open for the duration of each flight would bankrupt them.
Tritium oxide is radioactive water.
(However, the GP's water is most likely contaminated with run of the mill radon.)
With cosmic rays, it's not just "gone". Instead, you get a shower of new energetic particles generated by the collision which compounds the risk of operational errors. The patent specifically mentions alpha particles knocked out of the atoms in the chip by the ray which travel through the circuits causing havoc.
The patent also mentions that the detector may sense side effects of collision (such as voltage spikes) rather than the ray particle itself. Thus, the damage has already been done by the time the detector sees the event.
That's right. And the most surefire way to convince your boss to let you work 9-5 in the winter and 8-4 in the summer is to institute DST.
(Maybe patent terms could be variable, and you could apply for a 4 year or whatever year patent if you couldn't make a convincing argument that the idea wouldn't be independently invented in the next 20 years.)
That probably would cut the volume of patents down by 99%, while still rewarding the few people who come up with truly groundbreaking ideas.
To qualify for the discount, Microsoft said that it was a student.
I doubt it would actually matter much, given that most software needs to be continuously updated to remain relevant. Each update would have a fresh copyright. Proprietary "freeloaders" would necessarily be stuck with a rather stale public domain fork, and would have to independently author and maintain any updates for the software. That major hassle would probably deter such proprietary forks in most cases.
Not since the 1960s. After they accidentally scattered bits of H-bomb around Spain and Greenland, they stopped the continuous flights of nuclear armed bombers.
But it is wrong now. The entire US healthcare risk underwriting system is wrong.
By having thousands of individual risk pool managers obsess over saving money by kicking out people who might actually use healthcare services, we ironically end up with a system that costs us almost twice as much overall as any other country, while at the same time not even covering a huge swath of the population.
Meanwhile, needlessly stupid thing like worrying about who gets a hold of medical tests causes stress for millions. Millions more are tied to their corporate jobs like feudal serfs because of fear of losing healthcare benefits.
To stop this insanity, there needs to be one single uniform national risk pool.
What you say is true, but the fix is simple: make the retirement age variable, and adjust it each year it so that a fixed percentage of the population receives benefits. If you live longer on average, you should work longer on average.
The US per capita continuous total energy consumption averages out to about 10.5kW thermal (100e18 J annual national total/300M people). With hydrogen combustion at 286kJ/mol, it would take 62 liters of water per day per person to provide hydrogen for *all* the energy currently used in the US. Residential water consumption is already around 400 liters per person per day, industrial usage is more than that, and agricultural usage is many thousands of liters per day per person. IOW, this won't be a significant increase in overall water use. If it really got to be a problem, just set it up by the ocean and use some extra solar energy to run desalinization plants.
Part of the question depends on how much memory bandwidth Google needs. For their search engine, IIRC they try to keep much of the index in main memory. It could be that having dozens of threads simultaneously hammering on one memory bank wouldn't work out. In that case, Google might be better off saving power by using lots of low power laptop x86 chips or ARM CPUs, each with their own memory, than a single highly parallel CPU. OTOH, that speculation could be wrong and the Sun CPUs might be perfect for them.
We've been able to do that because all the money was over here. However, between trade deficits and government borrowing, we've been working really hard on sending that money over to China lately. So before long it may not make much sense for their best and brightest to come over here when they can get paid with US cash right in their own hometowns.
Look, the guy is off by more than 4 orders of magnitude. It would be really interesting to find out where he got this absurd misinformation.
* [Citation Seriously Needed]
They still are. But you, like so many others, seem to be completely ignorant of the concept of rate of change. Humans are changing the CO2 levels orders of magnitude faster than natural factors have in the past, so those effects get lost in the noise.
So "humans are emitting lots of CO2" does cut it.
If you understood the article, it should be pretty obvious that CO2 likely didn't trigger the end of the last few ice ages given that there probably weren't any large releases of CO2 like we're making now. (And before anybody gets any big ideas: Volcanoes aren't the culprit. They release a tiny fraction as much CO2 as humans.) As the article points out, the changes likely were triggered by other factors like changes in the earth's orbit.
If the CO2 didn't trigger the changes, but does participate in a positive feedback loop, then of course it would lag the temperature. But that has nothing to do at all with the question of whether an increase in CO2 levels could also trigger a warming cycle.
The second website looks to me like a highly biased collection of cargo cult science put together by people who specialize in fields like economics, not climatology.
* [Citation Needed]
Not really. Mainframes do batch processing of predetermined non-interactive workloads best. Google does interactive database searches with a fraction of a second latency, serves up web ads, and is trying to host traditional desktop applications via a web browser.
Mainframes have really puny CPU horsepower relative to their size, cost and power consumption. Their OSes are tuned for batch processing. Almost every compromise in mainframe design is decided in favor of uptime and transactional integrity, things for which Google has almost no use at all. They would be throwing a lot of money at solving issues they don't have if they ran mainframes, and even if they did manage to buy enough mainframes to handle their particular workload, it would probably end up using more power than they're using now.
Now I'm afraid that I'm a victim of this scam. It looks like this "Slashdot" site I've been using could actually be nothing more than a bad spoof...