I'm just telling you what I'm seeing based on comparisons of flights around the holidays. From 1999-2001, flights were cheap. Immediately after 9/11, the prices jumped up a notch, and have been climbing steadily ever since, with the biggest increase being 2008, where they were nearly double the 2007 prices. Yes, there are some cheap tickets available, but the airlines block out the cheap seats completely for more than a week on either side of the holidays, thus even though the average ticket prices are not increasing rapidly, the ticket prices when people tend to want to fly are skyrocketing, or at least they are for flights between SJC and either BNA or MEM on every airline that flies out of those cities. YMMV.
What somebody else said about the best prices being 2-3 weeks before is interesting, and I may revisit giving the airlines money so far ahead of time. The prices are certainly not likely to go up much from $750 each way.... In a time when the airlines are strapped for cash and need money sooner rather than later, screwing over early ticket purchasers probably isn't the brightest idea....
Eh. I've gotten a letter back from Feinstein. Of course, she said that at the time, they were not planning any bills similar to the one that had just passed the house and she would consider my POV if they did, then promptly (within two or three weeks) voted for a bill that had precisely the same fundamental flaws I pointed out, which we are now stuck with.
The problem is this: politicians are in the pockets of big corporations---ALL politicians as far as I can tell. They don't care what their constituents have to say on issues because they've already made up their minds based on what lobbyists tell them to think. (I can pretty much guarantee that there were few of Feinstein's individual constituents on the opposite side of the issue in question, if any.) As long as we have lobbyists, we will never have a government of the people because it allows certain people or companies with more money to be able to buy influence.
I assume you're referring to chromatic aberration? Bear in mind that the James Webb satellite I mentioned already detects the red and possibly orange portions of the visible spectrum. If you're not getting too much chromatic aberration there, then even in the worst case, I'd expect the amount you'd get in the visible spectrum to be minor enough to be correctable in software. It should be easily correctable with an additional lens, with a modified sensor design, or with some combination of those two plus software post-processing.
Even with something designed for the far infrared, though, I would think you could still correct for the aberration with an additional lens on the front of the camera. Less ideal than using a full set of lenses specifically optimized for focusing visible light? Sure. I still think it would be better than nothing.
Besides, AFAIK, the Herschel telescope is fundamentally a reflecting telescope, not a refracting telescope. There's no chromatic aberration from a mirror, and I suspect that any lenses involved are likely to be per-instrument, not shared among the various detectors. Thus, dropping in a visible light detector should be pretty much a non-issue. Am I missing something fundamental here?
I don't know what flights you're flying on, but I used to be able to consistently get flights from CA to TN for $200-$300 round trip, and sometimes as low as $150. Last Christmas, I actually flew with frequent flyer miles first class because it was going to be somewhere around $1,000 for the round trip and it took 5,000 fewer frequent flyer miles for first class than it did for coach. This Christmas, it's looking like I'm going to have a hard time getting below $1500 for that round trip---more expensive by a factor of 10 in the same number of years.
There's something really appalling about spending $1,500 for a round trip cattle car flight with no food that previously cost $200, came with a meal, allowed us to check two bags at no charge (instead of zero now), and didn't require us to strip down, remove our shoes, and generally be treated to subhuman conditions all for the "privilege" of having to wolf down dinner while running to catch the connecting flight because your first leg was delayed an hour due to a problem with one of the engines that should have been caught by routine maintenance but wasn't because they aren't paying their maintenance people well enough....
If you buy tickets months ahead of time, you used to be able to get good deals. These days, by the time the return flight becomes available for sale, all the cheap seats on the flight over are sold out because everybody is buying so far ahead trying to keep from getting utterly screwed by the skyrocketing prices.
What it looks like from my perspective is that airlines are cutting the flights on minor routes to the bare minimum and gouging the passengers to limit ridership, then using the excess profits from the gouging to cover the operating losses on long haul flights like LA to NYC nonstop and on non-holiday-season flights to other places
Maybe things have gotten cheaper in comparison with... say 1980... but compared with the late 90s or the first part of this decade pre-9/11, they've gone through the roof. It has gone from Greyhound being more expensive to being a fourth as much almost overnight. Trains are now cheaper. Buying a cheap used car and driving across is sometimes cheaper, including the fuel costs. IIRC, flights to Europe last year were running almost double what they were just three years earlier. And so on. I'm just not seeing these price drops that you're seeing except when flying in the middle of the year. If you fly anywhere from mid-November through mid-January, expect to pay a lot more than you did just a few years ago.
That still doesn't answer the question about why they didn't include the visible spectrum. The way I see it, if you're building this giant space telescope with multiple sensors anyway, why not throw in a visible light camera for good measure? Sure, it may not be the main focus of the mission, but when Hubble's electronics finally bite the dust (and at this rate, that may be before too long), there's going to be a real lack of these breathtaking images of space for many years. All of that could be avoided by simply having something to take over that load when the time comes. And remember, while the infrared instruments may give you data that scientists want, public funding comes only in response to pretty pictures....
The same argument applies to the U.S. telescope scheduled to replace Hubble in 2013 for the same reason. It's all infrared as well.
And then you say "stop doing that or I'll block you" and your friends stop... or you block them. Either way, that's not spam. That's your friend being a jerk. It's a fine line.
No, no, no, you misunderstand. Servers have to sign the communication. Users have to connect to a server. The only thing a spammer could do is to masquerade as a user or masquerade as a server.
If they masquerade as a server, that's a crypto key that the spammers have to pay for, which gets rapidly banned by every mail server in the world after the first email message from it.
If they masquerade as a user, the person receiving the spam could reliably contact that user's ISP, forward the message, and that user would get cut off until he/she could disinfect his/her machine. In effect, this would nearly eliminate the viability of botnets as a delivery mechanism because it guarantees that the message's original sender is known and can be blocked. Even in the case of compromised computers, it ensures that those compromised computers can get taken down much, much faster than anonymous Winzombies known only by their IP numbers. Also, it means that you could send the sender an email telling them that their computer is sending you spam and they could fix it themselves.
The fundamental reason that botnets are so hard to deal with is that people shut their machines down when not in use, so these things are constantly hopping to new IP numbers, and thus tracking a spam message back to the computer that distributed it is a royal pain in the backside. Most ISPs just can't deal with that (or aren't willing to bother). Thus, a compromised machine can send countless spam email message before anyone knows something is wrong. If, however, the message contained a username at that ISP that could be reliably and conclusively linked to a single user on their system, the verification problem is no longer an issue (it can be handled programmatically as part of the protocol), so computers that spew spam can be rapidly thwarted merely by complaining about the spam message.
In fact, an ideal system would deal with some or all of the policy functionality for handling spam complaints as part o the design, making it trivial for ISPs to set up policies such that if several people complain about email from a single user within a certain period of time (via an automated "mark as spam" button that notifies the upstream server, which notifies the original user's mail server), the offending user gets cut off automatically. When an ISP received an automated complaint about a user, it would check the email message to make sure that it really was signed by that ISP's server (and has not been modified since it was signed), thus establishing that the identity of the user is trustworthy. After such verification, it would add to a "total complaints" counter for that user. After five complaints in a day, the user's email access would be shut off. When the customer calls the ISP to complain, the ISP tells them that they have a virus, the user downloads appropriate tools, cleans the machine up, and the spammer just lost a winzombie after sending only a handful of messages. Oh, and if you do it right, you could also have the ISP send out notices automatically recalling any unread messages that were sent by that user since the first spam was sent.
Even better, every email server along the mail chain could have similar policies with higher thresholds so that if the ISP can't be bothered to block a spamming user, the user will still get blocked until that ISP sends out a "user is clean" message. Maybe add a UUID for the customer that changes if the user is blocked and then unblocked. If an ISP starts abusing that, the ISP then can be blocked by its host key.
I guess the point is that the only reason we have spam is that email is so trivially forged. Were it nearly impossible to forge the sender, the balance of power would shift significantly. It wouldn't eliminate 100% of spam, but with a properly designed protocol, I think you could get pretty close.
As for the question of how to make only the lowest order bits have errors... analog computing, I guess.:-D
That said, I just thought of one way this might be useful. If (and only if) the amount of die shrinkage and processor speed increase you can get as a result of doing this is significantly more than a factor of three and the probability of an error is significantly less than a one-in-three chance, you could use a voting algorithm with "no two alike forces retry" logic. That said, I wouldn't hold my breath on getting such high boosts and still being able to make those sorts of probability guarantees.
Depends on what part of the FFT calculation, but sure. That said, what I'd be more concerned about is an assembly language instruction in the language of your choice along these lines: "Jump to address (0x31000 + value of register 1)". Make an off-by one in that calculation and you probably wedge the chip, or at best, get completely useless operation....
For something like this to even be plausible, it would have to be in some very special-case vector pipeline like a GPU, not in a general-purpose CPU with jumps and branches and stuff. I'm really not convinced this has any practical use in a general-purpose CPU, randomness in quantum computing notwithstanding.
Actually, if email gets replaced by some other messaging system, it very easily could eliminate spam. The sole reason email spam can't be taken care of with a technological solution is that the infrastructure changes would be too massive and you couldn't get 100% opt-in for any new scheme by such a large number of players.
If SMTP were replaced by something else entirely, that whole problem goes away and you can design proper security into the protocol. All you really need is a protocol that enforces end-to-end authentication (and possibly encryption) and requires that every host involved in the transaction except for client machines sign every message that passes through and include a public key in their DNS record that can verify that signature. This would completely eliminate any possibility of endpoint forging, which would mean that spammers would have to keep registering domains to get new non-blocked source domains. Eliminate domain tasting, and those new domains = $$$ that the spammers would have to pay... frequently... all without introducing any per-message costs. If you take it one step further and require an SSL cert (not self-signed), that action by itself would be pretty much be the end of bulk spam as we know it, but would have the advantage of not harming legitimate business-to-user communication, email discussion lists, etc. like a per-message cost would.
Such a change would radically alter the balance of power in the spam wars. It would ensure that a spammer, once identified, could be trivially blocked, and would make it much harder and more expensive for the spammer to recover from such blocking. Unfortunately, while it would be possible to retrofit this onto SMTP, that compatibility with legacy systems would ultimately make it a waste of effort to do so; spammers would merely continue using the legacy compatibility mode to deliver the spam. It really has to be a clean break from SMTP for this to work and gain any traction whatsoever, and has to be 100% spam-free by design from day one.
There are at least three major problems with your logic.
First, the customer is not always in control of their traffic. We're seeing games that use P2P for updates, etc., viruses trying to spread themselves, etc., and I'm sure we'll be seeing a lot more of that in the short-term future.
Second, a lot of folks are getting more and more of their video content through either video streaming or downloads from services like iTunes or Unbox. I would expect that within five years, a 100GB wall will seem unreasonably low, particularly with HD movies. An HD movie trailer takes over 100 megs at 720p. For TV shows @ 20 minutes apiece, between the high definition version and the standard-def version for iPod, you're at a gigabyte. That means a single two hour movie should average about 6 gigs. Watching one movie per day would require nearly double that bandwidth cap. That's legal movie downloading I'm talking about here. Heaven help you if you decide to download some TV shows, too.
As bandwidth becomes more ubiquitous, expect services like YouTube to also increase their quality and the bandwidth associated with it. They've already done that a couple of times. It won't be long before many frequent YouTube users start hitting that 100GB wall as well.
Third, things like video chat can push 1.5Mbps at high quality settings. Even at that rate, with a 100GB cap, you hit a wall just beyond 6 days. Thus, your teenager communicating with his/her girlfriend/boyfriend all day on Saturday and part of Sunday each week can easily eat more than half your month's bandwidth allotment by itself. If you make it a three-way or four-way video chat, you can burn through 100GB pretty easily.
Those are the facts as I see them. The upswell in high definition multimedia content, video messaging, etc. is pushing more and more users to move larger and larger amounts of data on a regular basis, and a 100GB limit is just too easy for a typical technologically inclined family to exceed. The networks are simply going to have to keep up. If they don't, they are likely to start burning through customers before too much longer.
After the Wii2 comes the Wii3... which comes in a special "R" edition with a digital video recorder, and also comes packaged with the ever popular game, Kings of Orient, making it the Wii3/Kings of Orient/R.
My bad. I got bits and bytes backwards in the google calculator. It's about 40 kBps, or 320 kbps, which is a slightly more respectable but still pathetically absurd number.
Even in large print, this borders on insanity. At 15 Mbps, 100 GB is only 15 hours at full bore in one direction. That's not an "always on" connection. That's an "occasionally on" connection, limited to a mere 2% of the advertised bandwidth when averaged over the month. The next tier is limited at 3%.
I think this falls under "truth in advertising". Companies advertising Internet service should be required to sell based on the average sustainable bandwidth, not peak bandwidth. That means for this 15 Mbps connection, it should be advertised as a 38.4 kbps connection---slower than dialup. There's something to tell your friends. "I'm paying $68 a month for this expensive internet connection, but to avoid getting cut off, I have to rate limit myself at half the speed of your $10 dialup line."
If I were on Charter cable, I would immediately drop service. Fortunately, I'm not. I've seen what that sleazy company is doing in terms of slashing local TV channels in TN. My parents used to get stations from Nashville and Memphis as part of their cable service at the insistence of the local government. Then, Tennessee decided to enact legislation that provides for statewide licensing of the right to provide cable service. Almost immediately after this, Charter cut those channels. The mayor is currently appealing to the FCC to force them to carry at least one station that provides decent coverage of Tennessee state news. Currently, all but two of the stations they carry are from Kentucky or Missouri, and only one of the TN stations has news at all (and it barely qualifies as news coverage, IMHO). As a result, they are basically cut off from all real coverage of local and regional events. Worse, they no longer get Memphis weather radar to warn about impending storms early on (the Paducah weather radar is okay, but it doesn't provide nearly as early a warning), so Charter is putting the health and safety of residents at risk as well, all to save them from having to maintain their repeater towers.
In the grand scheme of things, the bandwidth caps are just the tip of the iceberg when talking about what a disaster Charter cable has been---poor analog signals, frequent loss of cable modem service, a couple of days ago they lost all their upstream so you could get to sites in town but no farther out, etc. This company clearly got in way, way over their heads by acquiring more local cable companies than they could handle, and as a result, they're bleeding red and are struggling desperately to keep from completely collapsing. I say let them collapse, and moreover, demand that they sell their facilities to the local governments so that the wires can be managed and maintained as they should be---by a municipally-owned cable company. That's the only way residents are ever going to get good (or even adequate) cable service in rural communities.
FWIW, even in quantities of 1, an OEM copy of Windows Vista Home Basic is only $99. If you're building your own machine, odds are good that you'll decide to go that route, as the support isn't likely to be worth much to you, nor the pretty retail box, nor the transferability. About the only reason you might buy the retail copy as a builder is that you get both the 32-bit and 64-bit versions.
I'm probably picking nits here, but I think it's a bit disingenuous to describe Aperture as "professional" while calling Lightroom "intermediate". The programs are very comparable to one another. Each has a number of very useful features that people complain about being missing in the other. They definitely are not radically different tiers of software in terms of features, price, or performance.
That's a huge stretch. We all know that coal power is nasty, and over time, we should expect it to decrease to zero. At that point, the CFLs will still be polluting our environment by creating mercury plumes from garbage dumps, while those incandescent bulbs won't be releasing any mercury. Besides, not everybody's power comes from coal. There are a lot of places where the majority of power comes from nuclear and hydro. Some folks have solar panels on their roofs. And so on.
Also, the real risk of CFLs is caused by the fact that any pollution from it is local and concentrated as a point source, e.g. the risk of contaminating your house by dropping one. Mercury is nasty stuff. I don't even use lead-based solder inside my house. Mercury? Yikes. Also, consider that the plumes from garbage dumps invading your water supply are likely to be a heck of a lot more problematic for the human population than trace mercury emitted into the atmosphere a hundred miles away from the city....
Right now I'm in a market for a 15" MBP to replace a PPC Powerbook but the glossy screen is preventing me from purchasing it. For starters, glossy screens are unusable in a properly illuminated room with unequal multiple light sources. Its even worse outside on a sunny day.
Indoors, on a desktop panel, for graphics purposes, sure. You'd ideally like the panel to be consistently washed out even if it is slightly washed out. In terms of general usability of a laptop, and particularly for use outdoors, though, I couldn't disagree more. I've used Macs for many years, and recently got a MacBook. I tried both screens and concluded that the glossy screen worked much, much better for me than the matte display when used outdoors, so long as the sun is not directly behind me (at which point it is blinding, of course).
With matte displays, any significant source of light behind me resulted in poor contrast across the entire panel because of the diffuse reflection off the mate screen, to such a degree that I found the matte displays to be very difficult to use outdoors (without a sun hood) except on the darkest, cloudiest days. With the glossy display, by contrast, light and dark areas behind me remain in sharp focus, so I can more easily ignore them; I can always move around to see the portions of the screen I need to see if some part is obscured by a light source. WIth a matte screen, no amount f moving will make the glare go away. I still sometimes use a sun hood, but at least now it is about making me more comfortable rather than being a necessity to be able to make out anything at all.:-)
I'm not going to disagree with complaints about the color reproduction, though. I've never seen an LCD panel in my life from any manufacturer where certain gradients didn't look like absolute excrement, and that's almost bordering on cruelty to the excrement. I'm sure there are some panels that are good, but I certainly haven't seen them. At this point, I'm convinced that the panel manufacturers aren't even trying anymore. Color accuracy hasn't improved significantly in five or six years, and in most cases, has actually gotten worse over that time period.
I blame the panel manufacturers for focusing so heavily on the mass market by constantly trying to make screens brighter. Every time the screens get brighter and increase in contrast ratio, they seem to consistently do so at a cost to the accuracy of their color reproduction. Most consumers, however, seem to care more about brightness than accuracy, and outside of the graphics world, I can see how that would be more useful in many cases. That said, IMHO we've reached a point where the brightness of most modern panels is basically sufficient for most purposes, so I think it's time for the panel manufacturers to take a step back and start working to fix the color accuracy of modern panels.
My ultimate preference would be to see all of the ISPs upgrade how much bandwidth they can actually handle, instead of getting more and more customers, and then bitching when they don't have enough bandwidth to handle all of the customers they got, while they still go out to get more customers. Would any of this really be a problem, then? I suppose that costs money, but we're all giving them money every month; and many of us got their service expecting no such problems as this. I guess I, like a lot of people I know, just expected too much...
That still won't eliminate the fundamental problem, though. The only way you can solve it through adding bandwidth would be for the path between the two VoIP users to be large enough to handle all the BitTorrent traffic on it plus the VoIP calls. Sadly, that's not really feasible because BitTorrent naturally tends to consume as much bandwidth as it can (barring users setting bandwidth limits). The explanation is a little complicated, so I need to define some terms up front.
Let "wide path" represent the path between two VoIP callers, whose bandwidth has been increased sufficiently to handle that call plus BT traffic such that the link never saturates.
Let "BT seeders" represent the set of all of the BitTorrent seeders whose content anyone along the VoIP call's path is currently requesting if that traffic flows through any hop along the VoIP call's path.
Let "BT users" represent the set of all of the people downloading any of that content from any computer whose route to one of the BT seeders flows through any hop along the VoIP call's path.
For the wide path to be wide enough, we know that for every link along the wide path, the bandwidth for that link must be greater than the total amount of traffic flowing between BT users on either end of that link and BT seeders on every single path out of the router at the opposite end of that link put together. This means that if there is a single BT user and seeder on opposite sides of that link whose traffic must pass through that link, the only way the BitTorrent traffic will not grow to fill that link is if either the BT user or the BT seeder is be attached to the router through a link that is narrower than the link on the wide path. This means three things:
If the BT user and seeder happen to be connected to the same endpoint routers as the VoIP user, it is not possible for that pipe to be narrower than the wide path (because the route their traffic takes is identical to the wide path) unless either the BT user's connection to the ISP or the BT seeder's connection to the ISP is artificially throttled.
If the BT user or seeder (but not both) is connected through some other link off of any node along the wide path, that link must be narrower than the wide path, thus constraining the throughput to the width of that narrower link and preventing the wide path from filling up. Although the bandwidth along the wide path between those points is limited by that slower link, the BitTorrent traffic will fill that slower link to capacity. This means that any VoIP connection that needs to go through that link will fail. So by widening the wide path, you have just moved the problem to an adjacent link.
If both the BT user and the BT seeder are on different slow links off of the wide path, the BitTorrent traffic will be limited by the narrower of those two links. The BitTorrent traffic will fill that slower link to capacity, and again, a VoIP connection through that pipe will fail, so once again, you have just moved the problem to an adjacent link.
Thus, no matter where you increase the bandwidth, no matter how much you increase the bandwidth, you don't solve the problem; at best, you merely move it somewhere else. I suppose if you could get to the point where BitTorrent downloads were nearly instantaneous, that would do it, if only because the amount of time that BitTorrent users spent looking for another torrent to download wou
Either way, the effect is the same. If you make it per user, the only way to make that fair would be for each user's slice of the total bandwidth to be allowed to grow until it hits a point at which that user is using as much bandwidth as all of the other users. If you do that, however, the VoIP user starting a call still steals the exact same amount of bandwidth from each of the BitTorrent users as he/she would if prioritization were protocol-based (unless the VoIP app uses more bandwidth than the average BitTorrent user, which is rather unlikely). The only real difference is in the user experience of the VoIP user; with service-based priority, you can ensure low jitter for VoIP, while with user-based balancing, you can't. The BitTorrent user still gets the same bandwidth either way, and the extra latency to reduce VoIP jitter should have a negligible impact on downloads so long as the latency isn't so great that there are no outstanding requests waiting to come in and you get a pipeline stall....
Or, put another way, for the VoIP user to not reduce the bandwidth of BitTorrent users in other houses, those BitTorrernt users would have to be capped at a particular percentage of total available bandwidth. That's wasteful and inefficient, as there is no reason to cap bulk bandwidth usage if that bandwidth would otherwise go to waste. Protocol-based priority ensures that BitTorrent users get more bandwidth than they would with pretty much any other viable scheme (other than disallowing VoIP/streaming audio and video entirely) because it minimizes how much BitTorrent traffic must be throttled for anybody else to be able to get any use out of the network at all. It's really a much, much better solution than any of the alternatives... short of adding more pipes, of course.
That said, realistically, even adding pipes doesn't really help. No matter how big the pipes get, bandwidth will always be a problem because things like BitTorrent will always try to consume every bit of bandwidth that they can, up to the limits of the upstream network, so you just move the bottleneck a little farther out, until eventually the bottleneck is at the ISP of the VoIP user on the other end of the call, at which point they start increasing the size of their pipes farther and farther out, and the bottleneck moves back the other way. The only effective way to make networks viable for latency-sensitive traffic is to discriminate between latency-sensitive traffic and non-latency-sensitive traffic, and prioritize the two appropriately. Fortunately, if you do it correctly, it has little to no impact on bulk traffic.
It is pretty easy to define it, at least in this case. You define it as a maximum number of decibels measured from a distance of 30 feet from the vehicle. If you exceed that limit, whether because you don't have a decent muffler or because your radio shakes the entire freaking neighborhood, you get fined, and if you do not correct the problem in a timely manner, your vehicle is impounded. Simple as that.
Coal has killed FAR more than is attributed to it. Right now, nearly all the deaths attributed to coal is based on coal mine deaths, which IS much greater than nuclear power deaths (even when including all the uranium mining for weapons). But what is not added in there is the mercury poisoning that we get. Most of the mercury in our water is from coal. Likewise, much of our acid rains, etc are from coal. In a nutshell, Coal is far far worse than nukes.
If by coal mine deaths, you mean deaths from mining accidents, etc., then you also forgot black lung disease, which on average kills over 1,000 Americans per year by itself (on average over the last decade). Source: Wikipedia.
I'm told that it makes certain really bad music concerts seem palatable, so I suppose in that regard, it enhances performances... for some definition of "enhance"... and "performance"....
Slashdot Mirror
I'm just telling you what I'm seeing based on comparisons of flights around the holidays. From 1999-2001, flights were cheap. Immediately after 9/11, the prices jumped up a notch, and have been climbing steadily ever since, with the biggest increase being 2008, where they were nearly double the 2007 prices. Yes, there are some cheap tickets available, but the airlines block out the cheap seats completely for more than a week on either side of the holidays, thus even though the average ticket prices are not increasing rapidly, the ticket prices when people tend to want to fly are skyrocketing, or at least they are for flights between SJC and either BNA or MEM on every airline that flies out of those cities. YMMV.
What somebody else said about the best prices being 2-3 weeks before is interesting, and I may revisit giving the airlines money so far ahead of time. The prices are certainly not likely to go up much from $750 each way.... In a time when the airlines are strapped for cash and need money sooner rather than later, screwing over early ticket purchasers probably isn't the brightest idea....
Eh. I've gotten a letter back from Feinstein. Of course, she said that at the time, they were not planning any bills similar to the one that had just passed the house and she would consider my POV if they did, then promptly (within two or three weeks) voted for a bill that had precisely the same fundamental flaws I pointed out, which we are now stuck with.
The problem is this: politicians are in the pockets of big corporations---ALL politicians as far as I can tell. They don't care what their constituents have to say on issues because they've already made up their minds based on what lobbyists tell them to think. (I can pretty much guarantee that there were few of Feinstein's individual constituents on the opposite side of the issue in question, if any.) As long as we have lobbyists, we will never have a government of the people because it allows certain people or companies with more money to be able to buy influence.
I assume you're referring to chromatic aberration? Bear in mind that the James Webb satellite I mentioned already detects the red and possibly orange portions of the visible spectrum. If you're not getting too much chromatic aberration there, then even in the worst case, I'd expect the amount you'd get in the visible spectrum to be minor enough to be correctable in software. It should be easily correctable with an additional lens, with a modified sensor design, or with some combination of those two plus software post-processing.
Even with something designed for the far infrared, though, I would think you could still correct for the aberration with an additional lens on the front of the camera. Less ideal than using a full set of lenses specifically optimized for focusing visible light? Sure. I still think it would be better than nothing.
Besides, AFAIK, the Herschel telescope is fundamentally a reflecting telescope, not a refracting telescope. There's no chromatic aberration from a mirror, and I suspect that any lenses involved are likely to be per-instrument, not shared among the various detectors. Thus, dropping in a visible light detector should be pretty much a non-issue. Am I missing something fundamental here?
I don't know what flights you're flying on, but I used to be able to consistently get flights from CA to TN for $200-$300 round trip, and sometimes as low as $150. Last Christmas, I actually flew with frequent flyer miles first class because it was going to be somewhere around $1,000 for the round trip and it took 5,000 fewer frequent flyer miles for first class than it did for coach. This Christmas, it's looking like I'm going to have a hard time getting below $1500 for that round trip---more expensive by a factor of 10 in the same number of years.
There's something really appalling about spending $1,500 for a round trip cattle car flight with no food that previously cost $200, came with a meal, allowed us to check two bags at no charge (instead of zero now), and didn't require us to strip down, remove our shoes, and generally be treated to subhuman conditions all for the "privilege" of having to wolf down dinner while running to catch the connecting flight because your first leg was delayed an hour due to a problem with one of the engines that should have been caught by routine maintenance but wasn't because they aren't paying their maintenance people well enough....
If you buy tickets months ahead of time, you used to be able to get good deals. These days, by the time the return flight becomes available for sale, all the cheap seats on the flight over are sold out because everybody is buying so far ahead trying to keep from getting utterly screwed by the skyrocketing prices.
What it looks like from my perspective is that airlines are cutting the flights on minor routes to the bare minimum and gouging the passengers to limit ridership, then using the excess profits from the gouging to cover the operating losses on long haul flights like LA to NYC nonstop and on non-holiday-season flights to other places
Maybe things have gotten cheaper in comparison with... say 1980... but compared with the late 90s or the first part of this decade pre-9/11, they've gone through the roof. It has gone from Greyhound being more expensive to being a fourth as much almost overnight. Trains are now cheaper. Buying a cheap used car and driving across is sometimes cheaper, including the fuel costs. IIRC, flights to Europe last year were running almost double what they were just three years earlier. And so on. I'm just not seeing these price drops that you're seeing except when flying in the middle of the year. If you fly anywhere from mid-November through mid-January, expect to pay a lot more than you did just a few years ago.
Darl, is that you?
That still doesn't answer the question about why they didn't include the visible spectrum. The way I see it, if you're building this giant space telescope with multiple sensors anyway, why not throw in a visible light camera for good measure? Sure, it may not be the main focus of the mission, but when Hubble's electronics finally bite the dust (and at this rate, that may be before too long), there's going to be a real lack of these breathtaking images of space for many years. All of that could be avoided by simply having something to take over that load when the time comes. And remember, while the infrared instruments may give you data that scientists want, public funding comes only in response to pretty pictures....
The same argument applies to the U.S. telescope scheduled to replace Hubble in 2013 for the same reason. It's all infrared as well.
And then you say "stop doing that or I'll block you" and your friends stop... or you block them. Either way, that's not spam. That's your friend being a jerk. It's a fine line.
No, no, no, you misunderstand. Servers have to sign the communication. Users have to connect to a server. The only thing a spammer could do is to masquerade as a user or masquerade as a server.
If they masquerade as a server, that's a crypto key that the spammers have to pay for, which gets rapidly banned by every mail server in the world after the first email message from it.
If they masquerade as a user, the person receiving the spam could reliably contact that user's ISP, forward the message, and that user would get cut off until he/she could disinfect his/her machine. In effect, this would nearly eliminate the viability of botnets as a delivery mechanism because it guarantees that the message's original sender is known and can be blocked. Even in the case of compromised computers, it ensures that those compromised computers can get taken down much, much faster than anonymous Winzombies known only by their IP numbers. Also, it means that you could send the sender an email telling them that their computer is sending you spam and they could fix it themselves.
The fundamental reason that botnets are so hard to deal with is that people shut their machines down when not in use, so these things are constantly hopping to new IP numbers, and thus tracking a spam message back to the computer that distributed it is a royal pain in the backside. Most ISPs just can't deal with that (or aren't willing to bother). Thus, a compromised machine can send countless spam email message before anyone knows something is wrong. If, however, the message contained a username at that ISP that could be reliably and conclusively linked to a single user on their system, the verification problem is no longer an issue (it can be handled programmatically as part of the protocol), so computers that spew spam can be rapidly thwarted merely by complaining about the spam message.
In fact, an ideal system would deal with some or all of the policy functionality for handling spam complaints as part o the design, making it trivial for ISPs to set up policies such that if several people complain about email from a single user within a certain period of time (via an automated "mark as spam" button that notifies the upstream server, which notifies the original user's mail server), the offending user gets cut off automatically. When an ISP received an automated complaint about a user, it would check the email message to make sure that it really was signed by that ISP's server (and has not been modified since it was signed), thus establishing that the identity of the user is trustworthy. After such verification, it would add to a "total complaints" counter for that user. After five complaints in a day, the user's email access would be shut off. When the customer calls the ISP to complain, the ISP tells them that they have a virus, the user downloads appropriate tools, cleans the machine up, and the spammer just lost a winzombie after sending only a handful of messages. Oh, and if you do it right, you could also have the ISP send out notices automatically recalling any unread messages that were sent by that user since the first spam was sent.
Even better, every email server along the mail chain could have similar policies with higher thresholds so that if the ISP can't be bothered to block a spamming user, the user will still get blocked until that ISP sends out a "user is clean" message. Maybe add a UUID for the customer that changes if the user is blocked and then unblocked. If an ISP starts abusing that, the ISP then can be blocked by its host key.
I guess the point is that the only reason we have spam is that email is so trivially forged. Were it nearly impossible to forge the sender, the balance of power would shift significantly. It wouldn't eliminate 100% of spam, but with a properly designed protocol, I think you could get pretty close.
As for the question of how to make only the lowest order bits have errors... analog computing, I guess. :-D
That said, I just thought of one way this might be useful. If (and only if) the amount of die shrinkage and processor speed increase you can get as a result of doing this is significantly more than a factor of three and the probability of an error is significantly less than a one-in-three chance, you could use a voting algorithm with "no two alike forces retry" logic. That said, I wouldn't hold my breath on getting such high boosts and still being able to make those sorts of probability guarantees.
Depends on what part of the FFT calculation, but sure. That said, what I'd be more concerned about is an assembly language instruction in the language of your choice along these lines: "Jump to address (0x31000 + value of register 1)". Make an off-by one in that calculation and you probably wedge the chip, or at best, get completely useless operation....
For something like this to even be plausible, it would have to be in some very special-case vector pipeline like a GPU, not in a general-purpose CPU with jumps and branches and stuff. I'm really not convinced this has any practical use in a general-purpose CPU, randomness in quantum computing notwithstanding.
Actually, if email gets replaced by some other messaging system, it very easily could eliminate spam. The sole reason email spam can't be taken care of with a technological solution is that the infrastructure changes would be too massive and you couldn't get 100% opt-in for any new scheme by such a large number of players.
If SMTP were replaced by something else entirely, that whole problem goes away and you can design proper security into the protocol. All you really need is a protocol that enforces end-to-end authentication (and possibly encryption) and requires that every host involved in the transaction except for client machines sign every message that passes through and include a public key in their DNS record that can verify that signature. This would completely eliminate any possibility of endpoint forging, which would mean that spammers would have to keep registering domains to get new non-blocked source domains. Eliminate domain tasting, and those new domains = $$$ that the spammers would have to pay... frequently... all without introducing any per-message costs. If you take it one step further and require an SSL cert (not self-signed), that action by itself would be pretty much be the end of bulk spam as we know it, but would have the advantage of not harming legitimate business-to-user communication, email discussion lists, etc. like a per-message cost would.
Such a change would radically alter the balance of power in the spam wars. It would ensure that a spammer, once identified, could be trivially blocked, and would make it much harder and more expensive for the spammer to recover from such blocking. Unfortunately, while it would be possible to retrofit this onto SMTP, that compatibility with legacy systems would ultimately make it a waste of effort to do so; spammers would merely continue using the legacy compatibility mode to deliver the spam. It really has to be a clean break from SMTP for this to work and gain any traction whatsoever, and has to be 100% spam-free by design from day one.
There are at least three major problems with your logic.
First, the customer is not always in control of their traffic. We're seeing games that use P2P for updates, etc., viruses trying to spread themselves, etc., and I'm sure we'll be seeing a lot more of that in the short-term future.
Second, a lot of folks are getting more and more of their video content through either video streaming or downloads from services like iTunes or Unbox. I would expect that within five years, a 100GB wall will seem unreasonably low, particularly with HD movies. An HD movie trailer takes over 100 megs at 720p. For TV shows @ 20 minutes apiece, between the high definition version and the standard-def version for iPod, you're at a gigabyte. That means a single two hour movie should average about 6 gigs. Watching one movie per day would require nearly double that bandwidth cap. That's legal movie downloading I'm talking about here. Heaven help you if you decide to download some TV shows, too.
As bandwidth becomes more ubiquitous, expect services like YouTube to also increase their quality and the bandwidth associated with it. They've already done that a couple of times. It won't be long before many frequent YouTube users start hitting that 100GB wall as well.
Third, things like video chat can push 1.5Mbps at high quality settings. Even at that rate, with a 100GB cap, you hit a wall just beyond 6 days. Thus, your teenager communicating with his/her girlfriend/boyfriend all day on Saturday and part of Sunday each week can easily eat more than half your month's bandwidth allotment by itself. If you make it a three-way or four-way video chat, you can burn through 100GB pretty easily.
Those are the facts as I see them. The upswell in high definition multimedia content, video messaging, etc. is pushing more and more users to move larger and larger amounts of data on a regular basis, and a 100GB limit is just too easy for a typical technologically inclined family to exceed. The networks are simply going to have to keep up. If they don't, they are likely to start burning through customers before too much longer.
After the Wii2 comes the Wii3... which comes in a special "R" edition with a digital video recorder, and also comes packaged with the ever popular game, Kings of Orient, making it the Wii3/Kings of Orient/R.
My bad. I got bits and bytes backwards in the google calculator. It's about 40 kBps, or 320 kbps, which is a slightly more respectable but still pathetically absurd number.
Even in large print, this borders on insanity. At 15 Mbps, 100 GB is only 15 hours at full bore in one direction. That's not an "always on" connection. That's an "occasionally on" connection, limited to a mere 2% of the advertised bandwidth when averaged over the month. The next tier is limited at 3%.
I think this falls under "truth in advertising". Companies advertising Internet service should be required to sell based on the average sustainable bandwidth, not peak bandwidth. That means for this 15 Mbps connection, it should be advertised as a 38.4 kbps connection---slower than dialup. There's something to tell your friends. "I'm paying $68 a month for this expensive internet connection, but to avoid getting cut off, I have to rate limit myself at half the speed of your $10 dialup line."
If I were on Charter cable, I would immediately drop service. Fortunately, I'm not. I've seen what that sleazy company is doing in terms of slashing local TV channels in TN. My parents used to get stations from Nashville and Memphis as part of their cable service at the insistence of the local government. Then, Tennessee decided to enact legislation that provides for statewide licensing of the right to provide cable service. Almost immediately after this, Charter cut those channels. The mayor is currently appealing to the FCC to force them to carry at least one station that provides decent coverage of Tennessee state news. Currently, all but two of the stations they carry are from Kentucky or Missouri, and only one of the TN stations has news at all (and it barely qualifies as news coverage, IMHO). As a result, they are basically cut off from all real coverage of local and regional events. Worse, they no longer get Memphis weather radar to warn about impending storms early on (the Paducah weather radar is okay, but it doesn't provide nearly as early a warning), so Charter is putting the health and safety of residents at risk as well, all to save them from having to maintain their repeater towers.
In the grand scheme of things, the bandwidth caps are just the tip of the iceberg when talking about what a disaster Charter cable has been---poor analog signals, frequent loss of cable modem service, a couple of days ago they lost all their upstream so you could get to sites in town but no farther out, etc. This company clearly got in way, way over their heads by acquiring more local cable companies than they could handle, and as a result, they're bleeding red and are struggling desperately to keep from completely collapsing. I say let them collapse, and moreover, demand that they sell their facilities to the local governments so that the wires can be managed and maintained as they should be---by a municipally-owned cable company. That's the only way residents are ever going to get good (or even adequate) cable service in rural communities.
FWIW, even in quantities of 1, an OEM copy of Windows Vista Home Basic is only $99. If you're building your own machine, odds are good that you'll decide to go that route, as the support isn't likely to be worth much to you, nor the pretty retail box, nor the transferability. About the only reason you might buy the retail copy as a builder is that you get both the 32-bit and 64-bit versions.
I'm probably picking nits here, but I think it's a bit disingenuous to describe Aperture as "professional" while calling Lightroom "intermediate". The programs are very comparable to one another. Each has a number of very useful features that people complain about being missing in the other. They definitely are not radically different tiers of software in terms of features, price, or performance.
And yet most manufacturers are moving from CCFLs to LEDs. Hmm. Guess people are getting worked up about it after all.
That's a huge stretch. We all know that coal power is nasty, and over time, we should expect it to decrease to zero. At that point, the CFLs will still be polluting our environment by creating mercury plumes from garbage dumps, while those incandescent bulbs won't be releasing any mercury. Besides, not everybody's power comes from coal. There are a lot of places where the majority of power comes from nuclear and hydro. Some folks have solar panels on their roofs. And so on.
Also, the real risk of CFLs is caused by the fact that any pollution from it is local and concentrated as a point source, e.g. the risk of contaminating your house by dropping one. Mercury is nasty stuff. I don't even use lead-based solder inside my house. Mercury? Yikes. Also, consider that the plumes from garbage dumps invading your water supply are likely to be a heck of a lot more problematic for the human population than trace mercury emitted into the atmosphere a hundred miles away from the city....
Indoors, on a desktop panel, for graphics purposes, sure. You'd ideally like the panel to be consistently washed out even if it is slightly washed out. In terms of general usability of a laptop, and particularly for use outdoors, though, I couldn't disagree more. I've used Macs for many years, and recently got a MacBook. I tried both screens and concluded that the glossy screen worked much, much better for me than the matte display when used outdoors, so long as the sun is not directly behind me (at which point it is blinding, of course).
With matte displays, any significant source of light behind me resulted in poor contrast across the entire panel because of the diffuse reflection off the mate screen, to such a degree that I found the matte displays to be very difficult to use outdoors (without a sun hood) except on the darkest, cloudiest days. With the glossy display, by contrast, light and dark areas behind me remain in sharp focus, so I can more easily ignore them; I can always move around to see the portions of the screen I need to see if some part is obscured by a light source. WIth a matte screen, no amount f moving will make the glare go away. I still sometimes use a sun hood, but at least now it is about making me more comfortable rather than being a necessity to be able to make out anything at all. :-)
I'm not going to disagree with complaints about the color reproduction, though. I've never seen an LCD panel in my life from any manufacturer where certain gradients didn't look like absolute excrement, and that's almost bordering on cruelty to the excrement. I'm sure there are some panels that are good, but I certainly haven't seen them. At this point, I'm convinced that the panel manufacturers aren't even trying anymore. Color accuracy hasn't improved significantly in five or six years, and in most cases, has actually gotten worse over that time period.
I blame the panel manufacturers for focusing so heavily on the mass market by constantly trying to make screens brighter. Every time the screens get brighter and increase in contrast ratio, they seem to consistently do so at a cost to the accuracy of their color reproduction. Most consumers, however, seem to care more about brightness than accuracy, and outside of the graphics world, I can see how that would be more useful in many cases. That said, IMHO we've reached a point where the brightness of most modern panels is basically sufficient for most purposes, so I think it's time for the panel manufacturers to take a step back and start working to fix the color accuracy of modern panels.
That still won't eliminate the fundamental problem, though. The only way you can solve it through adding bandwidth would be for the path between the two VoIP users to be large enough to handle all the BitTorrent traffic on it plus the VoIP calls. Sadly, that's not really feasible because BitTorrent naturally tends to consume as much bandwidth as it can (barring users setting bandwidth limits). The explanation is a little complicated, so I need to define some terms up front.
Let "wide path" represent the path between two VoIP callers, whose bandwidth has been increased sufficiently to handle that call plus BT traffic such that the link never saturates.
Let "BT seeders" represent the set of all of the BitTorrent seeders whose content anyone along the VoIP call's path is currently requesting if that traffic flows through any hop along the VoIP call's path.
Let "BT users" represent the set of all of the people downloading any of that content from any computer whose route to one of the BT seeders flows through any hop along the VoIP call's path.
For the wide path to be wide enough, we know that for every link along the wide path, the bandwidth for that link must be greater than the total amount of traffic flowing between BT users on either end of that link and BT seeders on every single path out of the router at the opposite end of that link put together. This means that if there is a single BT user and seeder on opposite sides of that link whose traffic must pass through that link, the only way the BitTorrent traffic will not grow to fill that link is if either the BT user or the BT seeder is be attached to the router through a link that is narrower than the link on the wide path. This means three things:
Thus, no matter where you increase the bandwidth, no matter how much you increase the bandwidth, you don't solve the problem; at best, you merely move it somewhere else. I suppose if you could get to the point where BitTorrent downloads were nearly instantaneous, that would do it, if only because the amount of time that BitTorrent users spent looking for another torrent to download wou
Either way, the effect is the same. If you make it per user, the only way to make that fair would be for each user's slice of the total bandwidth to be allowed to grow until it hits a point at which that user is using as much bandwidth as all of the other users. If you do that, however, the VoIP user starting a call still steals the exact same amount of bandwidth from each of the BitTorrent users as he/she would if prioritization were protocol-based (unless the VoIP app uses more bandwidth than the average BitTorrent user, which is rather unlikely). The only real difference is in the user experience of the VoIP user; with service-based priority, you can ensure low jitter for VoIP, while with user-based balancing, you can't. The BitTorrent user still gets the same bandwidth either way, and the extra latency to reduce VoIP jitter should have a negligible impact on downloads so long as the latency isn't so great that there are no outstanding requests waiting to come in and you get a pipeline stall....
Or, put another way, for the VoIP user to not reduce the bandwidth of BitTorrent users in other houses, those BitTorrernt users would have to be capped at a particular percentage of total available bandwidth. That's wasteful and inefficient, as there is no reason to cap bulk bandwidth usage if that bandwidth would otherwise go to waste. Protocol-based priority ensures that BitTorrent users get more bandwidth than they would with pretty much any other viable scheme (other than disallowing VoIP/streaming audio and video entirely) because it minimizes how much BitTorrent traffic must be throttled for anybody else to be able to get any use out of the network at all. It's really a much, much better solution than any of the alternatives... short of adding more pipes, of course.
That said, realistically, even adding pipes doesn't really help. No matter how big the pipes get, bandwidth will always be a problem because things like BitTorrent will always try to consume every bit of bandwidth that they can, up to the limits of the upstream network, so you just move the bottleneck a little farther out, until eventually the bottleneck is at the ISP of the VoIP user on the other end of the call, at which point they start increasing the size of their pipes farther and farther out, and the bottleneck moves back the other way. The only effective way to make networks viable for latency-sensitive traffic is to discriminate between latency-sensitive traffic and non-latency-sensitive traffic, and prioritize the two appropriately. Fortunately, if you do it correctly, it has little to no impact on bulk traffic.
It is pretty easy to define it, at least in this case. You define it as a maximum number of decibels measured from a distance of 30 feet from the vehicle. If you exceed that limit, whether because you don't have a decent muffler or because your radio shakes the entire freaking neighborhood, you get fined, and if you do not correct the problem in a timely manner, your vehicle is impounded. Simple as that.
If by coal mine deaths, you mean deaths from mining accidents, etc., then you also forgot black lung disease, which on average kills over 1,000 Americans per year by itself (on average over the last decade). Source: Wikipedia.
I'm told that it makes certain really bad music concerts seem palatable, so I suppose in that regard, it enhances performances... for some definition of "enhance"... and "performance"....