You're not free. You are *more* politically free than the average person in China, but freedom ain't an "on/off" kind of thing, it's a "more/less" kind of thing.
The sad thing is though, that while the average chinese has become steadily more and more free lately, the trend in USA has been the other way, you guys are significantly less *free* now than you where a decade or two ago.
You require government-permission if you want to take pictures of a group of more than 2 people for over 20 minutes in Central Park, using a tripod. You are not allowed to talk about certain kinds of knowledge, like for example even that de-CSS exist. Your government maintains it can legitimately keep people imprisoned indefinitely while giving same neither the rigths of a POV nor the rigths of a criminal. You cannot bring something as trivial as a can of coke with you on a plane. You have to walk trough metal-detectors and accept answering questioning to be allowed to enter public buildings. You're not allowed to take apart objects that you own to figure out how they work. (not generally anyway) and if you *do* figure out how they work, sharing that knowledge with others may be a crime. You've been falling steadily on "freedom-of-press" rankings for the last decade, you used to be near the top, these days you're under average for a western democracy. "Free speech zones" (no comment needed)
USA is still in pretty good shape, certainly miles ahead of countries like china. But you're on the wrong track. You need to wake up.
I get all that. As I said, the misunderstanding was simply that I thougth you where saying something which you wheren't claiming at all. A simple misunderstanding.
That I get. You where just saying something different than I thougth you where saying. In effect you're just saying that using a tele-lens decreases depth-of-field.
It *sounded* as if you where saying that the middle part of a photo on a large sensor has larger DOF than the corners, which was the part I couldn't quite make sense of.
Depends on your jurisdiction. Germany ? "Unterlassene hilfeleistung", punishable by up to 1 year of prison. In Scandinavia similar. In the USA not *generally* forbidden, but with numerous exceptions, for example as a participant in traffic, you have an obligation to assist people hurt in an accident, even if you where not to blame for the accident.
That's the part that I don't get. How exactly does cropping out the middle part of an image improved depth-of-field ?
Are you under the impression that a larger area front-back is focused in the middle of a picture, compared to in the edges ? That isn't generally the case...
If everything between say 2 and 3 meters is in focus, then this is pretty universally the case, from the middle of the picture and to the very edges.
There's two ways of getting a larger focused area with a single camera and a single lens. Both involve getting less ligth, so both will give higher noise, or force you to film at brigther light.
First, like you say, go farther away and use a tele-lens to pull the foreground to the wanted size. This has the side-effect that, as you say, the background becomes bigger and appears closer to the foreground. (because what matters is the *relative* distance, having the actors 5 meters away and the explosion 50 meters away means the actors are 10 times closer. Having the actors 50 meters away and the explosion 100 meters away means the explosion is only twice as far away, so if you compensate by zooming until the actors are same size on screen, the end-result is a explosion that is visually 5 times larger than in the first case)
Second, use a smaller aperture. With an infinitely small aperture, you get everything in focus, with a small aperture you get a very large focused area.
True. But the fact that it may improve someones life isn't enough to put you in the clear. It is perfectly possible to improve someones life, while at the same time acting in a mannger that is morally and ethically bad.
People who have significantly less resources than you are often compelled to do what you want, because as you say, the alternative is worse, but it's not really a free choice because a choice to do X or have your children starve isn't a choice at all. (contracts signed under duress are invalid, the fact that in this sense the threat (hunger) isn't created by you doesn't make the choice any less forced)
Offering to feed peoples children, on the condition that they convert to christianity would be wrong.
Offering a woman who is in the wilderness with a broken leg a ride to the nearest hospital, on the condition that she give you a blowjob would be wrong.
Offering someone whose family is starving $2/day, on the condition that they work as slaves for you is wrong.
Yes, in each of these cases, not doing anything at all could be argued to be even worse. But that ain't enough. By that you've just demonstrated that the action is not the worst-possible-action. But there's a long step from being "not-the-worst" and to being "good".
The second example is particularily interesting; it would actually be a *crime* not to help a helpless person in such a situation.
Somehow though, that responsibility evaporates if it's a nation and not a single woman who is in trouble. And if it's a whole world, rather than a single human being, who choose not to help. (or to demand unreasonable compensation for the help)
What is the problem with buying $50 shoes made by people making $2/hour rather than $100 Nike-shoes made by people making $0.20/hour ? It's not as if the first is more expensive than the last....
Yeah, it's hard to know and avoid sweatshop products generally. But when you *do* know, and you *do* have a reasonable alternative, I don't think there's any question whatsoever what is the best choice.
I didn't say using 100GB was hard. Infact I said spending several hundred GB is easy.
My point was just that your math seemed wrong, that's all.
Audio is reasonably low-bandwith. movies is probably *the* thing that takes space currently. vo-ip is ignorable, it takes less space than listening to radio (typically only 64kbps/mono or less), and few people speak over vo-ip 12 hours a day 30 days a month. Gaming tends to be low-bandwith too, but can be very sensitive to lag. Most games are perfectly playable over 64kbps ISDN, provided the lag is low.
It's not hard to keep the audio-buffer large enough that you've got time to handle an incoming packet and return before the buffer under-runs.
Hell, if you where seriously low on power you could even respond to a network-interrupt by doing "fill-buffer, handle network-packet, return" which only requires you to have a buffer large enough that you've got time to handle 1 packet.
This is a completely moot point though, because a modern dual or quad-core has so insanely much power compared to what is needed to say decode mp3, that it can literally keep the audio-buffer filled by using 1% of one of the cores, which leaves it with 99.5% of its power for other stuff. Ok, so this means a 1% degradation of "other stuff" while listening to mp3s is acceptable.
But what people (including me) see, is a degradation of 1-2 orders of magnitude, which would be justified if the CPU spent 90% - 99% of its time keeping the audio-buffer full. Which is insane.
Must've been bad software, or an abundance of interrupt-generating devices on slow (nonpci) busses. A DX4 100 should have no problem whatsoever realtime-decoding mp3s, not even reasonably high-bitrate ones. Your typical (at the time) 128kbps mp3-file should play perfectly skip-free on a DX4 66, possibly even on 33.
That wasn't the problem. The problem is the numbers are stored with rounding-errors *AND* Excel contains some undocumented method of consistently correcting this and display the number as originally entered.
This method is not documented in the standard. Thus *other* programs that want to read Excel-files have to resort to guesswork to do a very basic thing that Excel does: Display a number that was entered by the user, the way the user entered it.
This means if we both get sent a valid OOXML-document, and you open it with Excel to read out some numbers, I open it with a program implemented by this so-called standard, the end-result is, you get the correct numbers, as entered by the user, I may or may not get the same numbers, depending on luck. This is a problem. It makes everyone else than Excel a second-class citizen.
Happened to me in Germany, Darmstadt. The cabbie managed to mess around for half an hour trough half the town in order to arrive at a major hotel that is located 100m off the highway. I had never been in Darmstadt before, but it was still obvious to me. So, I offered to pay half his meter. Which got him pissed, until I asked if we should together go into the hotel-lobby and ask how long the ride to the airport normally is.
I guess "average decent-quality" means 192kbps, agreed ? (even that is significantly above "average" webradios that are still at 128kbps, a lot of them is even lower, but that can't reasonably be called "decent-quality")
That is 24kB/second, 1.5MB/minute, 90MB/hour, 720MB/workday (8hr). On average, there's about 20 workdays a month, so I get 15GB/month.
Even if you listen 12hr/day, as you say, that upps it to 22GB/month. Which ain't 80GB. Even if you worked 12 hours, 30 days a month, that would be 32GB/month, which is a lot but still not 80GB.
Oh, it's not hard to use several hundred GB/month. But you're unlikely to do so by listening to webradio during your workday.
My point was, the "loss of capital" doesn't generally put their competitors ahead. Remember, the *owners* are the ones who hold the stock, they are the ones who benefit (primarily) when the stock go up. The company itself generally doesn't own much own stock, and isn't really influenced by changes in the stock-price.
It depends what you mean by "dangerous". Dangerous as in you'd be likely to lose a one-on-one figth with one of these monkeys and die, no.
Dangerous as in being attacked by a *group* of these monkeys may well get you severly scratched and bitten, yes. And remember, these are rural regions, you don't know if there's a friendly doctor over the street to patch you up.
The question is, how much risk to self is it reasonable to expect people to accept in order to *avoid* killing or potentially seriously injurying one or more animals that are a) harming your food-supply and b) being physically threathening to the point where people are scared of tending their own fields.
There's also the question about how much resources are available. These are poor farmers. Monkeys are quick. You *don't* want them to learn that humans are harmless, as would likely be the case if you tried something ineffective, like throwing stuff at them. (you'd never hit, other than by pure luck)
If the article is rigth, then the monkeys are threathening the food-supply of the villagers. This ain't a "minor" problem. Taking care of the environment is important, but that doesn't mean self-defence ain't allowed.
Norway, for example, has in general very strict hunting-regulations, mosts predators are completely legally protected. That does, however, not mean that a farmer isn't allowed to defend his animals. By lethal force if nessecary. The moment a bear starts attacking your dog, cow or sheep, you're in your perfect rigth to shoot it dead.
Non-lethal is better offcourse, but in any case the monkeys need to learn that human food-supply and humans themselves are not to be messed with. Philosophical dicussion is unlikely to convince them, they need to have negative experiences. A problem is that the monkeys are large enough to be potentially dangerous, most non-lethal defence-systems would be risky for the women to employ. An electric cattle-prod for example would certainly teach the monkeys a lesson, but you need to be very close, close enough that you won't get away if the monkey goes for an all-out attack. Quite possibly, as from the description it doesn't appear the monkeys have much fear.
The thing is, stock-price doesn't have any direct influence on day-to-day running of a company whatsoever. It matters if you want to raise capital for new investments, because a high market-cap will let you raise more money for a smaller dillution of the current shareholders, but other than that the influence is sligth.
If the companys "bad quarter" is in the red, then it depends on how soon they can get back in the black, and the size of their war-coffers. If they're really likeky to go bankrupt before they manage to return to black, then they *deserve* the low stock-price.
True. But the extra delay from going trough a list of, say, 100 extents should be completely trivial compared to the current cost: seek-time. But you're rigth, the cost of fragmentation ain't zero. It's just a tiny fraction of the *current* cost of fragmentation.
Housing is an investment in the sense that owning a house gives you a very significant low-interest buffer-zone for that rainy day. If you *do* suddenly need a larger amount of cash, owning a house means you can borrow it cheaply.
If you are rigth, by the way, you can yourself get rich by putting your money where your mouth is. If, as you say, people "jump ship" after a bad quarter more than is justified from a rational long-term perspective. Then you should make a point of buying stock 2 days after a company announces a bad quarter. If you are correct, this should net you a nice profit long-term.
I'm 100% in agreement with you that most people are financial idiots. So what ? That's nothing new. Indeed I suspect it was even more so in the 80ies than it is now. And it's generally speaking a *good* thing for those of us who aren't. If people sell things that are really valuable for a too low price because of a bad quarter: buy it ! If people are willing to pay too high a price for something because of a really good quarter, sell it to them !
Similar thinking applies to most other areas. If people are willing to pay more, and do infact pay more, for LCD-tvs than they are worth, buy stock in whoever produces, distributes and retail them. If someone is paying overprice, someone else is making a killing, rigth ?
Stock can even be used as insurance. Are you vulnerable to increases in gas-price ? Buy stock in a oil-company ! If the price of oil goes up, sure you'll pay more for gas. But you'll be able to since your stock will go up. Don't like rain ? Buy stock in a hydro power-plant. It won't keep you dry, but it's more pleasant with rain when you can imagine it's raining cent-coins on you. (I did, AFK.OL tripled its share-price in the last 3 years, I still don't like rain though:-))
It is stored in one of the extra data-areas I mentioned. And yes, this block is itself subject to wear-levelling.
Allthough, notice, that won't happen that often, even if you do a lot of writes to one block. Because after you do 1000 (or 10000) writes to a single block, that block gets physically moved. Which nessecitates *1* write to the block containing the mapping.
Which means that *that* block would get 1000 writes after 1 *million* writes to a single block. address. An in practice much later, nobody has a usage-pattern that consists of *only* writing repeatedly to a single location. In practice, the mapping-block is exceedingly unlikely to need moving at all. But yeah, it too will be moved if that is indeed nessecary. There are several simple tricks that are used for finding it (the simplest being having a double-indirect mapping where the first tells you where the mapping-block is)
That would offcourse mean that eventually, the block that tells you where the mapping-block is needs to be moved. Except in practice that ain't so. The mapping-block is updated at the soonest after 1000 written blocks. (or 10.000), which mean that *that* block may need to be moved after 1 million written blocks (or 100 million if using 10.000), so the *double* indirect block, the one that says where the mapping-block is, may need to be moved after 1000000000 to 1000000000000 blocks are written. But that won't happen, because the flash is worn out long before. (and besides, it'd take hundreds of years to write that much at the possible speed)
Another possibility would be having a "magic-value" that signifies the block that is mappings, and to scan for it on startup. (thereafter storing it's physical adress in a tiny bit of ram) that solution has the advantage that it extends indefinitely. (which is no advantage in practice but for those that are theoretical purists...) on the other hand this means slower startup, which sometimes matter. Your mileage may vary, as they say.
It's claimed in the article that the gap is narrowing. I don't know if this is true. I suspect it isn't, really. As you point out, flash is still a factor of 100, aproximately, more expensive than rotating platters.
What *is* however happening is that flash is getting to be "cheap enough" that is is attractive for other reasons. Silence. Performance. Physical-size. Ruggedness.
It's more like, if you can get a flash-disc of acceptable size for say a laptop for $300, it may be worth it if that compares to a $100 conventional drive with 5 times the capacity. Networks mean you can today store most of your data somewhere else (on magnetic platters) and so a 32GB flash-disk may be sufficient. (rather than the 160GB magnetic one say)
The cache-thing is already done -- by the OS. Unless you spesifically ask the OS to sync the file, it will return success to your application writing data before even initiating disk-io. If you write repeatedly to the same location, quite likely only the last write will ever hit the platters at all.
Internal ram-cache would be a win -- if it could be made dependable enough that you could consider stuff there "written" and not worry that it'd go lost say by powerloss. Perhaps the drive could have a small internal battery sufficient to write the ram to flash in the event of power-failure, I suspect that'd make it costly though.
There are many techniques. If you really want details, get a book or hit wikipedia. But to give you a general idea:
Each block has, infact, a bit more storage than the amount exposed. There are error-correcting checksums and stuff, allowing the drive to detect (and sometimes correct) errors, among these are, typically, a counter saying how many times the block has been written to.
If the drive notices that one block has a lot more writes than the average block, it can swap the contents of those two blocks internally, and then make a note of this swap. (just a simple mapping 0x000FE37 isat: 0x00A32B) The host-OS never even notices this, it keeps asking for block 0x000FE37, and keeps getting the same content that was always there, only that content is now *physically* stored somewhere else.
It's a lot as if your office is worn-down and needs to be redone, and management puts you in a different office, but let you keep your old phone-number. People calling for you won't even notice that *physically* you're now somewhere else, all they know is, they dial that number, they reach you.
Every OS with virtual memory does the same thing to RAM (though for a different reason) the logical adresses that the programs see are related to the *physical* (actual) ram-locations by a lookup-table.
It's *really* not a hard problem to solve, and it's been thoroughly solved for literally decades. Thus you really *CAN* assume that the entire drive is (aproximately) evenly used. Which means those calculations aren't bullshit afterall. Even if you just constantly rewrote a single block, what would happen is after a while (say 1000 writes) that block would, internally in the flash, be replaced with another physical block, if you write another 1000 times that'd happen again and so on.
Yes, there's a sligth overhead: Every time you do 1000 writes, the flash needs to do (aproximately) 1003 writes and 3 reads. That is a small overhead though, and it can be reduced by upping the constant from 1000 to 10000 say. (which would result in wear being sligthly less evenly distributed, but nevermind)
Slashdot Mirror
You're not free. You are *more* politically free than the average person in China, but freedom ain't an "on/off" kind of thing, it's a "more/less" kind of thing.
The sad thing is though, that while the average chinese has become steadily more and more free lately, the trend in USA has been the other way, you guys are significantly less *free* now than you where a decade or two ago.
You require government-permission if you want to take pictures of a group of more than 2 people for over 20 minutes in Central Park, using a tripod. You are not allowed to talk about certain kinds of knowledge, like for example even that de-CSS exist. Your government maintains it can legitimately keep people imprisoned indefinitely while giving same neither the rigths of a POV nor the rigths of a criminal. You cannot bring something as trivial as a can of coke with you on a plane. You have to walk trough metal-detectors and accept answering questioning to be allowed to enter public buildings. You're not allowed to take apart objects that you own to figure out how they work. (not generally anyway) and if you *do* figure out how they work, sharing that knowledge with others may be a crime. You've been falling steadily on "freedom-of-press" rankings for the last decade, you used to be near the top, these days you're under average for a western democracy. "Free speech zones" (no comment needed)
USA is still in pretty good shape, certainly miles ahead of countries like china. But you're on the wrong track. You need to wake up.
I get all that. As I said, the misunderstanding was simply that I thougth you where saying something which you wheren't claiming at all. A simple misunderstanding.
That I get. You where just saying something different than I thougth you where saying. In effect you're just saying that using a tele-lens decreases depth-of-field.
It *sounded* as if you where saying that the middle part of a photo on a large sensor has larger DOF than the corners, which was the part I couldn't quite make sense of.
Depends on your jurisdiction. Germany ? "Unterlassene hilfeleistung", punishable by up to 1 year of prison. In Scandinavia similar. In the USA not *generally* forbidden, but with numerous exceptions, for example as a participant in traffic, you have an obligation to assist people hurt in an accident, even if you where not to blame for the accident.
That's the part that I don't get. How exactly does cropping out the middle part of an image improved depth-of-field ?
Are you under the impression that a larger area front-back is focused in the middle of a picture, compared to in the edges ? That isn't generally the case...
If everything between say 2 and 3 meters is in focus, then this is pretty universally the case, from the middle of the picture and to the very edges.
sounds strange. What is the functional difference between a small sensor on the one hand and the middle part of a large sensor on the other hand ?
One would think the two are functionally equivalent, and thus have the same depth-of-field.
It will, in percent. But it won't in meters.
If, for example, you focus so that the background is barely in-focus, then the nearest thing in focus is closer with wide-angle than tele, sure.
But with a tele-lens, you don't need to have the foreground be actually that close to the lens to *appear* close.
There's two ways of getting a larger focused area with a single camera and a single lens. Both involve getting less ligth, so both will give higher noise, or force you to film at brigther light.
First, like you say, go farther away and use a tele-lens to pull the foreground to the wanted size. This has the side-effect that, as you say, the background becomes bigger and appears closer to the foreground. (because what matters is the *relative* distance, having the actors 5 meters away and the explosion 50 meters away means the actors are 10 times closer. Having the actors 50 meters away and the explosion 100 meters away means the explosion is only twice as far away, so if you compensate by zooming until the actors are same size on screen, the end-result is a explosion that is visually 5 times larger than in the first case)
Second, use a smaller aperture. With an infinitely small aperture, you get everything in focus, with a small aperture you get a very large focused area.
True. But the fact that it may improve someones life isn't enough to put you in the clear. It is perfectly possible to improve someones life, while at the same time acting in a mannger that is morally and ethically bad.
People who have significantly less resources than you are often compelled to do what you want, because as you say, the alternative is worse, but it's not really a free choice because a choice to do X or have your children starve isn't a choice at all. (contracts signed under duress are invalid, the fact that in this sense the threat (hunger) isn't created by you doesn't make the choice any less forced)
Offering to feed peoples children, on the condition that they convert to christianity would be wrong.
Offering a woman who is in the wilderness with a broken leg a ride to the nearest hospital, on the condition that she give you a blowjob would be wrong.
Offering someone whose family is starving $2/day, on the condition that they work as slaves for you is wrong.
Yes, in each of these cases, not doing anything at all could be argued to be even worse. But that ain't enough. By that you've just demonstrated that the action is not the worst-possible-action. But there's a long step from being "not-the-worst" and to being "good".
The second example is particularily interesting; it would actually be a *crime* not to help a helpless person in such a situation.
Somehow though, that responsibility evaporates if it's a nation and not a single woman who is in trouble. And if it's a whole world, rather than a single human being, who choose not to help. (or to demand unreasonable compensation for the help)
What is the problem with buying $50 shoes made by people making $2/hour rather than $100 Nike-shoes made by people making $0.20/hour ? It's not as if the first is more expensive than the last....
Yeah, it's hard to know and avoid sweatshop products generally. But when you *do* know, and you *do* have a reasonable alternative, I don't think there's any question whatsoever what is the best choice.
I didn't say using 100GB was hard. Infact I said spending several hundred GB is easy.
My point was just that your math seemed wrong, that's all.
Audio is reasonably low-bandwith. movies is probably *the* thing that takes space currently. vo-ip is ignorable, it takes less space than listening to radio (typically only 64kbps/mono or less), and few people speak over vo-ip 12 hours a day 30 days a month. Gaming tends to be low-bandwith too, but can be very sensitive to lag. Most games are perfectly playable over 64kbps ISDN, provided the lag is low.
True, but that's only half the story.
It's not hard to keep the audio-buffer large enough that you've got time to handle an incoming packet and return before the buffer under-runs.
Hell, if you where seriously low on power you could even respond to a network-interrupt by doing "fill-buffer, handle network-packet, return" which only requires you to have a buffer large enough that you've got time to handle 1 packet.
This is a completely moot point though, because a modern dual or quad-core has so insanely much power compared to what is needed to say decode mp3, that it can literally keep the audio-buffer filled by using 1% of one of the cores, which leaves it with 99.5% of its power for other stuff. Ok, so this means a 1% degradation of "other stuff" while listening to mp3s is acceptable.
But what people (including me) see, is a degradation of 1-2 orders of magnitude, which would be justified if the CPU spent 90% - 99% of its time keeping the audio-buffer full. Which is insane.
Must've been bad software, or an abundance of interrupt-generating devices on slow (nonpci) busses. A DX4 100 should have no problem whatsoever realtime-decoding mp3s, not even reasonably high-bitrate ones. Your typical (at the time) 128kbps mp3-file should play perfectly skip-free on a DX4 66, possibly even on 33.
That wasn't the problem. The problem is the numbers are stored with rounding-errors *AND* Excel contains some undocumented method of consistently correcting this and display the number as originally entered.
This method is not documented in the standard. Thus *other* programs that want to read Excel-files have to resort to guesswork to do a very basic thing that Excel does: Display a number that was entered by the user, the way the user entered it.
This means if we both get sent a valid OOXML-document, and you open it with Excel to read out some numbers, I open it with a program implemented by this so-called standard, the end-result is, you get the correct numbers, as entered by the user, I may or may not get the same numbers, depending on luck. This is a problem. It makes everyone else than Excel a second-class citizen.
Happened to me in Germany, Darmstadt. The cabbie managed to mess around for half an hour trough half the town in order to arrive at a major hotel that is located 100m off the highway. I had never been in Darmstadt before, but it was still obvious to me. So, I offered to pay half his meter. Which got him pissed, until I asked if we should together go into the hotel-lobby and ask how long the ride to the airport normally is.
I guess "average decent-quality" means 192kbps, agreed ? (even that is significantly above "average" webradios that are still at 128kbps, a lot of them is even lower, but that can't reasonably be called "decent-quality")
That is 24kB/second, 1.5MB/minute, 90MB/hour, 720MB/workday (8hr). On average, there's about 20 workdays a month, so I get 15GB/month.
Even if you listen 12hr/day, as you say, that upps it to 22GB/month. Which ain't 80GB. Even if you worked 12 hours, 30 days a month, that would be 32GB/month, which is a lot but still not 80GB.
Oh, it's not hard to use several hundred GB/month. But you're unlikely to do so by listening to webradio during your workday.
My point was, the "loss of capital" doesn't generally put their competitors ahead. Remember, the *owners* are the ones who hold the stock, they are the ones who benefit (primarily) when the stock go up. The company itself generally doesn't own much own stock, and isn't really influenced by changes in the stock-price.
It depends what you mean by "dangerous". Dangerous as in you'd be likely to lose a one-on-one figth with one of these monkeys and die, no.
Dangerous as in being attacked by a *group* of these monkeys may well get you severly scratched and bitten, yes. And remember, these are rural regions, you don't know if there's a friendly doctor over the street to patch you up.
The question is, how much risk to self is it reasonable to expect people to accept in order to *avoid* killing or potentially seriously injurying one or more animals that are a) harming your food-supply and b) being physically threathening to the point where people are scared of tending their own fields.
There's also the question about how much resources are available. These are poor farmers. Monkeys are quick. You *don't* want them to learn that humans are harmless, as would likely be the case if you tried something ineffective, like throwing stuff at them. (you'd never hit, other than by pure luck)
Oh ! I love that argument, it's just *so* compelling !
If someone does something bad, there's *always* going to be something *worse* you can point to, and that magically makes the bad thing ok.
If the article is rigth, then the monkeys are threathening the food-supply of the villagers. This ain't a "minor" problem. Taking care of the environment is important, but that doesn't mean self-defence ain't allowed.
Norway, for example, has in general very strict hunting-regulations, mosts predators are completely legally protected. That does, however, not mean that a farmer isn't allowed to defend his animals. By lethal force if nessecary. The moment a bear starts attacking your dog, cow or sheep, you're in your perfect rigth to shoot it dead.
Non-lethal is better offcourse, but in any case the monkeys need to learn that human food-supply and humans themselves are not to be messed with. Philosophical dicussion is unlikely to convince them, they need to have negative experiences. A problem is that the monkeys are large enough to be potentially dangerous, most non-lethal defence-systems would be risky for the women to employ. An electric cattle-prod for example would certainly teach the monkeys a lesson, but you need to be very close, close enough that you won't get away if the monkey goes for an all-out attack. Quite possibly, as from the description it doesn't appear the monkeys have much fear.
The thing is, stock-price doesn't have any direct influence on day-to-day running of a company whatsoever. It matters if you want to raise capital for new investments, because a high market-cap will let you raise more money for a smaller dillution of the current shareholders, but other than that the influence is sligth.
If the companys "bad quarter" is in the red, then it depends on how soon they can get back in the black, and the size of their war-coffers. If they're really likeky to go bankrupt before they manage to return to black, then they *deserve* the low stock-price.
True. But the extra delay from going trough a list of, say, 100 extents should be completely trivial compared to the current cost: seek-time. But you're rigth, the cost of fragmentation ain't zero. It's just a tiny fraction of the *current* cost of fragmentation.
Housing is an investment in the sense that owning a house gives you a very significant low-interest buffer-zone for that rainy day. If you *do* suddenly need a larger amount of cash, owning a house means you can borrow it cheaply.
:-))
If you are rigth, by the way, you can yourself get rich by putting your money where your mouth is. If, as you say, people "jump ship" after a bad quarter more than is justified from a rational long-term perspective. Then you should make a point of buying stock 2 days after a company announces a bad quarter. If you are correct, this should net you a nice profit long-term.
I'm 100% in agreement with you that most people are financial idiots. So what ? That's nothing new. Indeed I suspect it was even more so in the 80ies than it is now. And it's generally speaking a *good* thing for those of us who aren't. If people sell things that are really valuable for a too low price because of a bad quarter: buy it ! If people are willing to pay too high a price for something because of a really good quarter, sell it to them !
Similar thinking applies to most other areas. If people are willing to pay more, and do infact pay more, for LCD-tvs than they are worth, buy stock in whoever produces, distributes and retail them. If someone is paying overprice, someone else is making a killing, rigth ?
Stock can even be used as insurance. Are you vulnerable to increases in gas-price ? Buy stock in a oil-company ! If the price of oil goes up, sure you'll pay more for gas. But you'll be able to since your stock will go up. Don't like rain ? Buy stock in a hydro power-plant. It won't keep you dry, but it's more pleasant with rain when you can imagine it's raining cent-coins on you. (I did, AFK.OL tripled its share-price in the last 3 years, I still don't like rain though
It is stored in one of the extra data-areas I mentioned. And yes, this block is itself subject to wear-levelling.
Allthough, notice, that won't happen that often, even if you do a lot of writes to one block. Because after you do 1000 (or 10000) writes to a single block, that block gets physically moved. Which nessecitates *1* write to the block containing the mapping.
Which means that *that* block would get 1000 writes after 1 *million* writes to a single block. address. An in practice much later, nobody has a usage-pattern that consists of *only* writing repeatedly to a single location. In practice, the mapping-block is exceedingly unlikely to need moving at all. But yeah, it too will be moved if that is indeed nessecary. There are several simple tricks that are used for finding it (the simplest being having a double-indirect mapping where the first tells you where the mapping-block is)
That would offcourse mean that eventually, the block that tells you where the mapping-block is needs to be moved. Except in practice that ain't so. The mapping-block is updated at the soonest after 1000 written blocks. (or 10.000), which mean that *that* block may need to be moved after 1 million written blocks (or 100 million if using 10.000), so the *double* indirect block, the one that says where the mapping-block is, may need to be moved after 1000000000 to 1000000000000 blocks are written. But that won't happen, because the flash is worn out long before. (and besides, it'd take hundreds of years to write that much at the possible speed)
Another possibility would be having a "magic-value" that signifies the block that is mappings, and to scan for it on startup. (thereafter storing it's physical adress in a tiny bit of ram) that solution has the advantage that it extends indefinitely. (which is no advantage in practice but for those that are theoretical purists...) on the other hand this means slower startup, which sometimes matter. Your mileage may vary, as they say.
In any case, this problem is solved.
It's claimed in the article that the gap is narrowing. I don't know if this is true. I suspect it isn't, really. As you point out, flash is still a factor of 100, aproximately, more expensive than rotating platters.
What *is* however happening is that flash is getting to be "cheap enough" that is is attractive for other reasons. Silence. Performance. Physical-size. Ruggedness.
It's more like, if you can get a flash-disc of acceptable size for say a laptop for $300, it may be worth it if that compares to a $100 conventional drive with 5 times the capacity. Networks mean you can today store most of your data somewhere else (on magnetic platters) and so a 32GB flash-disk may be sufficient. (rather than the 160GB magnetic one say)
The cache-thing is already done -- by the OS. Unless you spesifically ask the OS to sync the file, it will return success to your application writing data before even initiating disk-io. If you write repeatedly to the same location, quite likely only the last write will ever hit the platters at all.
Internal ram-cache would be a win -- if it could be made dependable enough that you could consider stuff there "written" and not worry that it'd go lost say by powerloss. Perhaps the drive could have a small internal battery sufficient to write the ram to flash in the event of power-failure, I suspect that'd make it costly though.
There are many techniques. If you really want details, get a book or hit wikipedia. But to give you a general idea:
Each block has, infact, a bit more storage than the amount exposed. There are error-correcting checksums and stuff, allowing the drive to detect (and sometimes correct) errors, among these are, typically, a counter saying how many times the block has been written to.
If the drive notices that one block has a lot more writes than the average block, it can swap the contents of those two blocks internally, and then make a note of this swap. (just a simple mapping 0x000FE37 isat: 0x00A32B) The host-OS never even notices this, it keeps asking for block 0x000FE37, and keeps getting the same content that was always there, only that content is now *physically* stored somewhere else.
It's a lot as if your office is worn-down and needs to be redone, and management puts you in a different office, but let you keep your old phone-number. People calling for you won't even notice that *physically* you're now somewhere else, all they know is, they dial that number, they reach you.
Every OS with virtual memory does the same thing to RAM (though for a different reason) the logical adresses that the programs see are related to the *physical* (actual) ram-locations by a lookup-table.
It's *really* not a hard problem to solve, and it's been thoroughly solved for literally decades. Thus you really *CAN* assume that the entire drive is (aproximately) evenly used. Which means those calculations aren't bullshit afterall. Even if you just constantly rewrote a single block, what would happen is after a while (say 1000 writes) that block would, internally in the flash, be replaced with another physical block, if you write another 1000 times that'd happen again and so on.
Yes, there's a sligth overhead: Every time you do 1000 writes, the flash needs to do (aproximately) 1003 writes and 3 reads. That is a small overhead though, and it can be reduced by upping the constant from 1000 to 10000 say. (which would result in wear being sligthly less evenly distributed, but nevermind)