I got one of these at a job fair from Microsoft recently. I'd never actually been able to solve one before, but now with the power of Google, I thought I'd try for a first.
Turns out it's easy to find links to speed cubing pages, but for people like me who just want an easy-to-understand (as opposed to super-fast) solution you have to dig a bit. The best explanation I found was Denny's 3x3 cubing page, which uses a layer-by-layer approach that's pretty intuitive. The only drawback was that it doesn't cover what to do about logos, which need to be oriented in a specific way (as opposed to just being on the right face); for that one, try Matt Monroe's page.
What's to stop this from becoming real? Why can't I bet on things like this in Vegas or Atlantic City?
Much of this is already "real". You can already go to the financial markets today to try to profit from what you think is going to happen to the price of commodities (via futures contracts), interest rates (swaps), stocks (options), etc. The key line from the article is:
Innovation Futures is a prediction game.
Similar to fantasy stock market games, this one lets players trade on all kinds of events.
The point of this seems to be to take a "real" game (the financial markets) and apply it in a virtual form, not vice versa. I'm sure Vegas and Atlantic City both have stockbrokers sitting around somewhere that would be happy to take your cash.
What I disagree with is this statement later in the article:
Innovation Futures, like any speculative market, aggregates the individual expertise and opinions of well-informed and well-motivated traders into a single easily understood number: a stock's trading price. This number can be taken as indicating the collectively agreed-upon likelihood of a particular outcome in the real-world. Even non-participants may thus benefit from such market signals.
Real markets (and the "terror market" which the US proposed earlier) contain information because people work very hard to make sure their investments perform well and that they don't suffer financial losses. In stock market games, on the other hand, participants aren't penalized for losing money, only for winning huge amounts of it. (The article even prevents you from going bankrupt: "When your account's net worth is below a certain level... you'll be given a choice of contracts that you can get for free and then resell on the markets against fresh cash.".) What tends to happen here (and I've seen it plenty of times in stock market games where they offer cash prizes to the winners) is that people start loading up on risky ventures, essentially turning the whole contest into a lottery rather than a reasoned investment strategy. I really doubt that under this sort of a situation any valid information can be read from the virtual prices.
I always imagined that at some point someone would come up with a standard cheap widget that everyone could plug into their POTS jack which would enable a distributed P2P style of VoIP system
How many people would like making phone calls over a network where other people could listen in on your calls by simply picking up an extension to the line the destination computer is using?
That aside, the real problem is that this doesn't solve the POTS to VoIP interface. If all you're doing is VoIP, you don't care about any of these services -- there's enough free voice chat available (tied into things like Yahoo, Messenger, Netmeeting etc) that you wouldn't need a for-pay service. The problem is when someone with a regular phone (and btw, that's still most of the world) tries to call you. What phone # are they supposed to dial to get to you? That's the real service that companies like Vonage are providing: translating from a regular POTS number to your IP and getting the call to you, regardless of your physical location. Not only is this a routing issue (one of the advantages of companies like Vonage is that you can carry your phone with you when you travel, and plug into anywhere there's a broadband outlet -- great for avoiding those ridiculous hotel phone charges), but it provides a seamless interface into POTS so that Joe Public doesn't have to do anything different to call you.
Looks like another Media Lab project; you can link to the project homepage here.
Personally I'd find it great if they could add voice recognition to it. One of my biggest weaknesses is remembering new names, especially when I'm introduced to a whole bunch of people one after the other. (I remember a job interview where I was taken on a tour of the building, and met around 10 people in 15 minutes. Then near the end of the tour, one of those people joined us for the rest of the interview, and I was trying desperately to remember which one he was:) ). Being able to have it dynamically associate people's faces with names and display a prompt would be a huge assist.
What's with the NSA tapping a Canadian company to do their classified encryptions? Most government research labs (Lincoln, Draper, Sandia...) won't even consider hiring non-US employees for security purposes.
it's not unreasonable to think a group of students might get together and pay $15 or $20 to print a couple hundred pages of textbook
This same opinion was expressed in the article, and it makes very little sense to me. Removing this feature from Amazon isn't going to affect textbook copiers anyhow. See, in most universities, there are these conveniently-located buildings called libraries, that have copies of just about every book for every class you'll take there. Many even have copies on reserve, so that they're never all checked out at once. The same building also has these fancy devices called photocopiers, which are good at high-volume duplication of paper.
It strikes me that the effort involved in scamming all the scanned pages out of Amazon would be as great or greater than making the initial copy from a hardcopy by hand. Trying to guess keywords for each set of 5 pages, frankly, sounds like a lot of work. Subsequent copies are both equally easy regardless of whether you're using a printer to spit out scans from Amazon or a sheet-feeder on your photocopier.
There are valid reasons for worrying about this technology (the point about cookbooks and reference books, where the relevant information really does only span a few pages, is especially well made), but this particular one is just the knee-jerk reflex to blame college students for yet more copyright-related legal measures.
sue not just the spammers, but the companies who hire them and ban their imports to the Unites States
Oh good! Then instead of hiring spammers to advertise their own products, companies will just hire'em to advertise their competitors' products. I wouldn't be surprised if that works out to be faster and cheaper than trying to push your own products.
Not to mention the fact that often spammers are resellers of no-name crap products which could easily be relabelled and sold under a different name. Banning products only works if your product name has some value.
If you can't enough make money in the first few years (and ideally the first year)of realease, why bother at all?
This is an excellent point. Movies which have not already made a large chunk of money in the first life+50 years aren't going to suddenly going to turn into cash cows in the next 20 years after that.
However, there are cases where movies which were never popularized under copyright flourish after the copyrights expire. The classic example of this is, of course, "It's A Wonderful Life". Although now one of the most-watched Christmas movies and most-imitated storylines, the initial reception for the movie was cool at best, with the movie barely making back the production costs. It wasn't until the copyrights expired that it began showing regularly every December and saw its popularity skyrocket.
If there'd been the same trend towards perpetual copyrights back then as there are today, we all would have missed out on this kind of gem. The public (especially later generations) got to see a movie that would probably have otherwise disappeared into the discount bin, and he studios lost nothing from the experience (it wasn't until IAWL was out of copyright that it took off). In fact, given how often the formula for IAWL has been copied, it could be argued that they've benefited by using its popularity to spawn a bunch of derivative works, which they later capitalized on. (I've lost count of how many TV shows and movies have used this theme.) It's unfortunate that the studios seem to prefer to take the shortsighted "milk-it-to-the-last-drop" view to a more longer-term and broader perspective.
The built-in 25W laser is powerful enough to burn quickly through paper and cardboard without leaving so much as a charred edge. It can cut through wood 1/4 inch thick or more. The thin beam barely discolors the edges of the top surface but leaves the sides of the cut a rich, dark brown. Acrylic plastics such as Plexiglas cut easily, too. The beam moves with surprising speed for wood and paper but slower and at lower power for plastics, so the heat from the beam can melt the plastic edge and leave a smooth, polished surface.
Give'em another year to improve it, and us budget mad scientists will be able to afford one of these for the next time Halle Berry breaks into our secret ice-covered labs, too.
More like they had a nasty shock when they discovered that they don't have as much power as they thought. The reaction of those running the root servers and their move towards circumventing Sitefinder via the BIND patch made it clear that there are still checks to the power Verisign currently wields. It's not surprising that they're advocating moves which will remove some of those checks so that they won't be as easily stopped next time.
Actually, it looks like you're referencing the work of a different group, not the one in the CNN article. From the NCDM link you provided:
A new milestone was reached in trans-Atlantic data transmission today by researchers at the
University of Illinois at Chicago (UIC) who demonstrated the practicality of transferring even very large data sets over high-speed production networks... In the test, 1.4 terabytes of astronomical data was transmitted from Chicago to Amsterdam in 30 minutes using UDT, a new protocol developed by the NCDM at the University of Illinois at Chicago.
From the CNN article:
The European Organisation for Nuclear Research, CERN, said the feat, doubling the previous top speed, was achieved in a nearly 30-minute transmission over 7,000 kms of network between
Geneva and a partner body in California.
It appears the link you provided was for a UIUC test between Chicago and Amsterdam. The joint CERN/CalTech test was of a Geneva/California link.
I understand there was some sort of computing conference recently, involving high-speed transfer, which had a transatlantic testbed for participants to use to run their tests. CalTech's test was part of this, and shared bandwidth with other projects (you actually see it in their Fast TCP transfer rate graphs when someone else's project starts using the fiber too). So I wouldn't be surprised if a few other organizations were testing similar high-rate schemes at around the same time.
Steven Low of CalTech's Netlabgave a talk at MIT yesterday regarding the modified TCP protocol they used to achieve this transfer. Those who are curious about the details can check out the Fast TCP homepage.
Basically they showed that conventional TCP is not very good at scaling to large flows like the ones in the article. He described a typical broadband Internet connection as being able to utilize only about 27 percent of the available bandwidth, while their modified FAST TCP connection reached 95 percent efficiency. He had some nice test results showing how the protocols reacted to having to share bandwidth with other flows, and pointed out how when other flows finished and more bandwidth opened up, conventional TCP was very slow to take advantage of the increased bandwidth.
There's an older Economist article describing the protocol in more detail for those who are interested.
Normally I'd agree with you about WMA being a necessary evil to get an online music retailer launched. For better or for worse, in most countries the RIAA (or their appropriate country-specific clone) will only feel comfortable peddling their wares digitally if they can have their DRM locked into the format. Since Microsoft is a big, well-known player, I'm guessing the tendency is for the tech-phobic peeps at RIAA to reflexively go toward them (name-brand recognition) when buying into a DRM scheme.
Having said that, I'd argue that it's unfair and legally sketchy for a Canadian online company to restrict your ability to use your online music purchases. Puretracks.com makes a big deal out of being Canada-only at the moment, which means that they should be more tuned in to Canadian copyright regulation. The thing is, because of CD taxation, Canadians are allowed to make as many copies as they want for personal use. Restricting the number of copies you can make via DRM prevents you from exercising that right. Note that this is completely separate from the question of whether the personal-use argument applies to P2P networks -- with Puretracks, you've already paid to get the songs digitally.
It'd be interesting to see if anyone challenges their DRM restrictions legally based on these arguments. I'd suspect it's something they'll inevitably see this issue raised sooner or later... Canadians haven't been too happy about the CD tax in general (especially when they see Americans getting blank CDs for free or almost free after rebate); at least before they were getting something in return for the extra price. If precendents like this continue to be set, the CD tax is just going to be free music industry money without returning any value to the consumer.
Why would Canadians need an online music service when Canadians are legally allowed to make copies for personal use? The whole point of the 21-cent CD tax (PDF) over there is to explicitly allow this sort of thing. After all, Canadians are already paying an ever-increasing amount for this very privilage.
Granted, this currently only applies to sneakernet and the application of this argument to peer-to-peer networks has yet to be tested in court. This is IMHO a very important point that needs to be clarified in Canadian law for the good of the public. Otherwise, everyone there runs the risk of having to double-pay for every song they get online -- once to download the song, and once again to burn it onto CD.
All Dilbert jokes aside, which would you rather have, a technologically-impaired boss who recognizes his/her shortcomings and works to remedy it, or one who doesn't care and just keeps faking it?
Now what I'm curious about is how you'd figure out which of these consultants are the "good" ones, both in terms of being good at how the latest technology works and being able to explain it to the layperson. After all, when all your clients demand the level of secrecy described in the article, it's not like you can ask your potential tutor for references.
Tracks will be available in the Microsoft-developed.wma format. The levels of copyright protection afforded by the service were "at a level where we are very comfortable with it," he said.
I just can't understand the insistance of so many online music retailers to go with WMA instead of a more popular and widespread format like MP3. It can't be because of a fear of pirating -- this isn't going to help keep new (downloadable) music from Kazaa and the like because there are still too many simple ways to circumvent it, starting with looping your analog output back into your line input, and all it takes is a few people to realize this before the music gets out "into the wild".
In the meantime, it's just one more annoyance for their paying customers. Old mp3 players tend not to support WMA, and there's also the distastefulness of your music being tied to a Microsoft proprietary format, which have a tendency to have backward-compatibility issues (ever try migrating Office documents between versions?) and to be changed without much customer input. Given these issues, I don't think their security would suffer much by going with mp3, and it would be a good selling point for their legit customers. All in all, I see this as an annoying trend towards a business model where companies continue to try to impose their controls on things you've already paid for, rather than just letting you have full use of your purchases after you've forked over your cash.
Exactly. Whoever was responsible for writing such anti-spam software would be the first person to get hit with a massive lawsuit the first time some spammer found a way to "aim" this sort of scheme at an innocent bystander. If that bystander happens to be a big company with deep pockets, the programmer could be looking at some serious pain. Knowing that such a risk exists, it would be interesting to see if anyone would still be willing to develop such software.
The article tries to combat false positives with blacklists. A couple of problems with this come to mind right away. The first is that centrally-maintained blacklists are easy to take offline via DDOS, as we've already seen with sites like SPEWS. The second, and IMHO more serious, problem is that this would give the blacklist maintainers huge power over the rest of the internet -- if you ever got on their bad side, or if they were just plain inefficient/not conscientious about accidentally listing innocent bystanders, your site could potentially be shut down until they felt like taking you off the blacklist, just by some spammer spoofing you. Given the poor history of responsiveness that many blacklist maintainers have shown historically, I don't think giving them more power is the answer. Bad enough not being able to send people email if you accidentally get blacklisted -- imagine not being able to get net access at all.
First off, let me say that I agree that whether we're talking about 10^35 or 10^120 is largely moot; both of those are ridiculously large numbers. I'm only trying to address some of the points you brought up in this post, not to refute your conclusion that mapping the entire decision tree is infeasible.
On the other hand, the 10^120 sequences of moves is also a killer; sure, you only have to store 10^35 boards at 256 bits each (which is still an outrageously large number), but since you're turning it into a decision tree you need to keep track of which moves take you into which other states.
It's the actual solution of the problem that requires the tree, not the part about writing a program which implements the solution once it is known. Basically solving this type of an optimization problem involves calculating a "score" for each of the states that tells you the relative value of being in a given state versus another. Suppose the computer is in state #1. When it's the computer's turn, the computer (depending on the board position and pieces) has several choices of moves, which correspond to moving to states #2, 3, 4,... , N. Basically all it has to do is consult a look-up table to see which state has the highest score, and make the corresponding move. The difficulty is in coming up with the "scores" for each state.
The point of neurodynamic programming is that instead of having to store scores for each state, you only have to store scores for each "feature". For example, a simplistic program might decide that the only important feature is which pieces are left on the board. Then suppose during the program's turn it has the choice of either taking a pawn, a bishop, or a rook. It would consult its (small) lookup table and see (for example) that a pawn has score 1, a bishop has score 3, and a rook has score 4. It would then conclude that taking a rook leads to the state with the highest subsequent score, and make the corresponding move.
In the preceding example, you'd only have to have a 6 entry lookup table, corresponding to the 6 different pieces, instead of a 10^32-entry one. In reality, there are plenty of other features which are important (mostly positional), but the final size of your lookup table will still be much smaller than if you hadn't performed feature extraction.
Of course, the hard part is scoring each feature. As I mentioned in my earlier post, this is where the "training" part comes in. By the way, choosing features intelligently is also a big part of making a program perform well, which is the primary reason why chess programs perform so well while Go programs do not. The understanding of how to quantify which features on a Go board are important is not nearly as well understood; this is evident if you ever listen to skilled Go players watching a game in progress, where they'll be lots of comments about "shape" and "influence" which even the players themselves will find hard to quantify.
Correct me if I'm wrong, but this does not account for repetetive moves, in which case there are infinite number of possible moves.
In a move tree, repetitive moves are represented by cycles on the 'tree', and don't add any more nodes. In other words, I just return to a node I visited before, so no new node is created; the tree just kind of "doubles back" on itself. The number of nodes in the tree is still finite.
(I use quotes around 'tree' because technically it then becomes a general graph... but that's a quibble.)
Just to nitpick a little (since you're a math person, I thought you'd appreciate having your terms right): there are 10^120 different possible sequences of moves. The number of different states is actually quite a bit smaller, only around 10^35 or so. (A rough approximation would be 64!/32!, or the number of different ways you can set up a chessboard.) As a side note, this figure originates from a paper by Claude Shannon, the so-called father of modern communications ("Programming a digital computer for playing chess", Phil. Mag., pp 356-375, 1950). All computer chess programs today are based on the fundamental principles from this paper.
However, noting that the state-space size is large isn't really a very useful observation, since chess programs these days don't try to map out the entire tree of possible outcomes. Instead, they operate on neurodynamic programming techniques, which basically try to extract which "features" of the game are important and weigh those features to decide which moves to make. This significantly reduces the complexity of the system, but requires that the person writing the program have some intuition about which "features" are important. In chess, for example, these include such things as material balance, piece mobility, king safety, and other positional factors. A period of training is usually required as well, where basically the computer goes over a lot of games that grandmasters have played and tries to "learn" how to weigh the different features in order to choose the optimal move.
For those who are interested in reading further about this (yeah, yeah, this is Slashdot, if people can't RTFA what are the odds they'll want to pick up a book?:) ) a good place to start would be Chapter 6 of Bertsekas' "Dynamic Programming and Optimal Control".
Re:No reliance on a single "tower"
on
The Smart Sensor Web
·
· Score: 2, Insightful
One thing I've always said a cell phone should have is the ability to pick up cell phone broadcast signals to determine which are nearby. Then, in the case of a tower outage, or straying too far away, you would switch over to a peer-to-peer version where your signal would get passed on via other phones to the next nearest tower. Hopefully this functionality would use a low quality, low bandwidth signal, so as not to disrupt other callers on the phones it's passed through.
Unsurprisingly given how many smart people are working on wireless these days, systems like this have been discussed frequently in the past. The primary reason we don't see this isn't actually the channel allocation and spectral efficiency issues you mention; it's a much more simple problem. While most new cell phones have standby times on the order of up to a week, the actual talk time (by which they usually mean when data is being transmitted) is usually only a few hours. How many users would go for a system where just a few hours after recharging their phone they ran out of juice because it was busy relaying someone else's call to the nearest tower?
Not quite. Smart dust (a project that started back in 1999 at the Robotics lab at Berkeley, and which reached the prototype testing stage earlier this year) was never intended to be a global, long-term sensor network. Its strengths are that it can be easily deployed in areas which have been traditionally difficult to fit with conventional wireless sensor networks (such as battlefields) and that it is self-organizing so minimal setup time is required (again, important in combat applications -- there was a reason Smart Dust research is funded by DARPA). Neither feature is essential to the global sensor network that this story is discussing.
The primary reason Smart Dust wouldn't be a good fit (aside from the relatively high cost of deploying it, compared to using a cheaper, less miniaturized commercial solution) is the power problem. A big challenge for networking researchers involved with this type of sensor net is that each dust "mote" has very limited power reserves, which once consumed are typically not replenishable. (There have been ideas tossed around about recharging by harvesting solar or vibrational energy, but those are just idle speculation at the moment.) This is great for something like a battlefield network, which only needs to be up for the duration of your conflict, but is unsuitable for a persistant network.
It used to be that one of the reasons artists would sign away their soul for a recording contract was because they needed cash to finance the high up-front cost of studio time, recording, and editing to put out an album. With cheap digital studios on the rise, and as artists become increasingly computer-savvy, they'll be able to do more with less up-front capital, and be able to release songs more easily on their own. All this will let them sit down at the negotiating table with a bit more bargaining leverage.
(Of course, the next part of the story is promotions...)
Slashdot Mirror
Turns out it's easy to find links to speed cubing pages, but for people like me who just want an easy-to-understand (as opposed to super-fast) solution you have to dig a bit. The best explanation I found was Denny's 3x3 cubing page, which uses a layer-by-layer approach that's pretty intuitive. The only drawback was that it doesn't cover what to do about logos, which need to be oriented in a specific way (as opposed to just being on the right face); for that one, try Matt Monroe's page.
What I disagree with is this statement later in the article:
Real markets (and the "terror market" which the US proposed earlier) contain information because people work very hard to make sure their investments perform well and that they don't suffer financial losses. In stock market games, on the other hand, participants aren't penalized for losing money, only for winning huge amounts of it. (The article even prevents you from going bankrupt: "When your account's net worth is below a certain levelThat aside, the real problem is that this doesn't solve the POTS to VoIP interface. If all you're doing is VoIP, you don't care about any of these services -- there's enough free voice chat available (tied into things like Yahoo, Messenger, Netmeeting etc) that you wouldn't need a for-pay service. The problem is when someone with a regular phone (and btw, that's still most of the world) tries to call you. What phone # are they supposed to dial to get to you? That's the real service that companies like Vonage are providing: translating from a regular POTS number to your IP and getting the call to you, regardless of your physical location. Not only is this a routing issue (one of the advantages of companies like Vonage is that you can carry your phone with you when you travel, and plug into anywhere there's a broadband outlet -- great for avoiding those ridiculous hotel phone charges), but it provides a seamless interface into POTS so that Joe Public doesn't have to do anything different to call you.
Personally I'd find it great if they could add voice recognition to it. One of my biggest weaknesses is remembering new names, especially when I'm introduced to a whole bunch of people one after the other. (I remember a job interview where I was taken on a tour of the building, and met around 10 people in 15 minutes. Then near the end of the tour, one of those people joined us for the rest of the interview, and I was trying desperately to remember which one he was :) ). Being able to have it dynamically associate people's faces with names and display a prompt would be a huge assist.
What's with the NSA tapping a Canadian company to do their classified encryptions? Most government research labs (Lincoln, Draper, Sandia ...) won't even consider hiring non-US employees for security purposes.
It strikes me that the effort involved in scamming all the scanned pages out of Amazon would be as great or greater than making the initial copy from a hardcopy by hand. Trying to guess keywords for each set of 5 pages, frankly, sounds like a lot of work. Subsequent copies are both equally easy regardless of whether you're using a printer to spit out scans from Amazon or a sheet-feeder on your photocopier.
There are valid reasons for worrying about this technology (the point about cookbooks and reference books, where the relevant information really does only span a few pages, is especially well made), but this particular one is just the knee-jerk reflex to blame college students for yet more copyright-related legal measures.
Not to mention the fact that often spammers are resellers of no-name crap products which could easily be relabelled and sold under a different name. Banning products only works if your product name has some value.
However, there are cases where movies which were never popularized under copyright flourish after the copyrights expire. The classic example of this is, of course, "It's A Wonderful Life". Although now one of the most-watched Christmas movies and most-imitated storylines, the initial reception for the movie was cool at best, with the movie barely making back the production costs. It wasn't until the copyrights expired that it began showing regularly every December and saw its popularity skyrocket.
If there'd been the same trend towards perpetual copyrights back then as there are today, we all would have missed out on this kind of gem. The public (especially later generations) got to see a movie that would probably have otherwise disappeared into the discount bin, and he studios lost nothing from the experience (it wasn't until IAWL was out of copyright that it took off). In fact, given how often the formula for IAWL has been copied, it could be argued that they've benefited by using its popularity to spawn a bunch of derivative works, which they later capitalized on. (I've lost count of how many TV shows and movies have used this theme.) It's unfortunate that the studios seem to prefer to take the shortsighted "milk-it-to-the-last-drop" view to a more longer-term and broader perspective.
Conspiracy theorests trade tin-foil hats for head-mounted mirrors.
More like they had a nasty shock when they discovered that they don't have as much power as they thought. The reaction of those running the root servers and their move towards circumventing Sitefinder via the BIND patch made it clear that there are still checks to the power Verisign currently wields. It's not surprising that they're advocating moves which will remove some of those checks so that they won't be as easily stopped next time.
I understand there was some sort of computing conference recently, involving high-speed transfer, which had a transatlantic testbed for participants to use to run their tests. CalTech's test was part of this, and shared bandwidth with other projects (you actually see it in their Fast TCP transfer rate graphs when someone else's project starts using the fiber too). So I wouldn't be surprised if a few other organizations were testing similar high-rate schemes at around the same time.
Basically they showed that conventional TCP is not very good at scaling to large flows like the ones in the article. He described a typical broadband Internet connection as being able to utilize only about 27 percent of the available bandwidth, while their modified FAST TCP connection reached 95 percent efficiency. He had some nice test results showing how the protocols reacted to having to share bandwidth with other flows, and pointed out how when other flows finished and more bandwidth opened up, conventional TCP was very slow to take advantage of the increased bandwidth.
There's an older Economist article describing the protocol in more detail for those who are interested.
Having said that, I'd argue that it's unfair and legally sketchy for a Canadian online company to restrict your ability to use your online music purchases. Puretracks.com makes a big deal out of being Canada-only at the moment, which means that they should be more tuned in to Canadian copyright regulation. The thing is, because of CD taxation, Canadians are allowed to make as many copies as they want for personal use. Restricting the number of copies you can make via DRM prevents you from exercising that right. Note that this is completely separate from the question of whether the personal-use argument applies to P2P networks -- with Puretracks, you've already paid to get the songs digitally.
It'd be interesting to see if anyone challenges their DRM restrictions legally based on these arguments. I'd suspect it's something they'll inevitably see this issue raised sooner or later ... Canadians haven't been too happy about the CD tax in general (especially when they see Americans getting blank CDs for free or almost free after rebate); at least before they were getting something in return for the extra price. If precendents like this continue to be set, the CD tax is just going to be free music industry money without returning any value to the consumer.
Granted, this currently only applies to sneakernet and the application of this argument to peer-to-peer networks has yet to be tested in court. This is IMHO a very important point that needs to be clarified in Canadian law for the good of the public. Otherwise, everyone there runs the risk of having to double-pay for every song they get online -- once to download the song, and once again to burn it onto CD.
Now what I'm curious about is how you'd figure out which of these consultants are the "good" ones, both in terms of being good at how the latest technology works and being able to explain it to the layperson. After all, when all your clients demand the level of secrecy described in the article, it's not like you can ask your potential tutor for references.
In the meantime, it's just one more annoyance for their paying customers. Old mp3 players tend not to support WMA, and there's also the distastefulness of your music being tied to a Microsoft proprietary format, which have a tendency to have backward-compatibility issues (ever try migrating Office documents between versions?) and to be changed without much customer input. Given these issues, I don't think their security would suffer much by going with mp3, and it would be a good selling point for their legit customers. All in all, I see this as an annoying trend towards a business model where companies continue to try to impose their controls on things you've already paid for, rather than just letting you have full use of your purchases after you've forked over your cash.
Try Waybackmachine instead; it's a little older but at least they archive the pics too.
The article tries to combat false positives with blacklists. A couple of problems with this come to mind right away. The first is that centrally-maintained blacklists are easy to take offline via DDOS, as we've already seen with sites like SPEWS. The second, and IMHO more serious, problem is that this would give the blacklist maintainers huge power over the rest of the internet -- if you ever got on their bad side, or if they were just plain inefficient/not conscientious about accidentally listing innocent bystanders, your site could potentially be shut down until they felt like taking you off the blacklist, just by some spammer spoofing you. Given the poor history of responsiveness that many blacklist maintainers have shown historically, I don't think giving them more power is the answer. Bad enough not being able to send people email if you accidentally get blacklisted -- imagine not being able to get net access at all.
The point of neurodynamic programming is that instead of having to store scores for each state, you only have to store scores for each "feature". For example, a simplistic program might decide that the only important feature is which pieces are left on the board. Then suppose during the program's turn it has the choice of either taking a pawn, a bishop, or a rook. It would consult its (small) lookup table and see (for example) that a pawn has score 1, a bishop has score 3, and a rook has score 4. It would then conclude that taking a rook leads to the state with the highest subsequent score, and make the corresponding move.
In the preceding example, you'd only have to have a 6 entry lookup table, corresponding to the 6 different pieces, instead of a 10^32-entry one. In reality, there are plenty of other features which are important (mostly positional), but the final size of your lookup table will still be much smaller than if you hadn't performed feature extraction.
Of course, the hard part is scoring each feature. As I mentioned in my earlier post, this is where the "training" part comes in. By the way, choosing features intelligently is also a big part of making a program perform well, which is the primary reason why chess programs perform so well while Go programs do not. The understanding of how to quantify which features on a Go board are important is not nearly as well understood; this is evident if you ever listen to skilled Go players watching a game in progress, where they'll be lots of comments about "shape" and "influence" which even the players themselves will find hard to quantify.
(I use quotes around 'tree' because technically it then becomes a general graph ... but that's a quibble.)
However, noting that the state-space size is large isn't really a very useful observation, since chess programs these days don't try to map out the entire tree of possible outcomes. Instead, they operate on neurodynamic programming techniques, which basically try to extract which "features" of the game are important and weigh those features to decide which moves to make. This significantly reduces the complexity of the system, but requires that the person writing the program have some intuition about which "features" are important. In chess, for example, these include such things as material balance, piece mobility, king safety, and other positional factors. A period of training is usually required as well, where basically the computer goes over a lot of games that grandmasters have played and tries to "learn" how to weigh the different features in order to choose the optimal move.
For those who are interested in reading further about this (yeah, yeah, this is Slashdot, if people can't RTFA what are the odds they'll want to pick up a book? :) ) a good place to start would be Chapter 6 of Bertsekas' "Dynamic Programming and Optimal Control".
The primary reason Smart Dust wouldn't be a good fit (aside from the relatively high cost of deploying it, compared to using a cheaper, less miniaturized commercial solution) is the power problem. A big challenge for networking researchers involved with this type of sensor net is that each dust "mote" has very limited power reserves, which once consumed are typically not replenishable. (There have been ideas tossed around about recharging by harvesting solar or vibrational energy, but those are just idle speculation at the moment.) This is great for something like a battlefield network, which only needs to be up for the duration of your conflict, but is unsuitable for a persistant network.
(Of course, the next part of the story is promotions ...)