It seems like there are two different issues in play here. The RFC referenced in the article talks about link-local addressing, which is simply a way to assign an address in the 169.254/16 subnet if no DHCP server is found. It is not wireless-specific at all.
What we have here is that, in addition to doing this, Windows is also offering to set up an ad-hoc (i.e. computer-to-computer) network on the link-local subnet with the same SSID as that of the last network the laptop connected to. I wonder what the rationale for doing this could have been. It seems to me that a machine should not offer to set up an ad-hoc network unless specifically directed to do so by the user. When such a network is set up then it is appropriate to use link-local addressing to auto-configure the interface.
Textpattern may be what you are looking for. Although it is used for blogs, it is very easy to setup as a general purpose CMS. And it seems to have good unicode support for your multilingual needs.
Actually, Intel has not yet issued a formal proposal to IEEE, but since they are playing a key role in the 11s task group, they will almost certainly be issuing their proposal in response to the call for proposals that just went out in late January. Proposals are due June 15th, I believe.
As for the details of what has been discussed so far in the 11s task group, anyone can sign up for an account at 802wirelessworld, and obtain access to all the documents submitted for consideration to the task group so far. (Once you register and login, a link for Documents shows up under 802.11 WLAN WG on the left of the page.)
Various usage scenarios have been considered, from the scale of the home ( a few devices) to larger scale community meshes. The standard will work on any "mesh-aware" point, which may be an AP or a client device. It will likely run at layer 2 (below the IP layer) and provide a standards based mechanism for multi-hop access to a wired gateway (or "mesh portal" as they refer to it).
There's an old computer science saw that all problems can be solved by adding a layer of indirection. In that spirit, how about the following:
Create a P2P system that requires a password for access (Windows file sharing should do!)
Encrypt the password
Leave the encrypted password and the decryption key accessible
Now, the password itself is a copyrighted work of my creation. So using the decryption key to crack the password should be a DMCA violation. Thus if people (including the RIAA!) use my decrypted password to access my files, it is they who are violating the law, not me, i.e. I cannot be accused of making copyrighted works publicly available.
Thoughts? (and apologies if this is how some of the more advanced systems already work; I am not familiar with them)
Navigation is an embarrassment. Using bookmarks and back and forth buttons -- we had about eighteen different things we had in mind for the browser.
Besides the use of tabs that most/.ers are familiar with now, there are also other new approaches to navigation as evidenced by the iRider browser. It's IE based, non-free, and Windows only, but they have some nice ideas. In particular, they have a left hand navigation pane that shows all visited websites in a tree fashion (with thumbnails), that works quite well.
When it comes to discussing irony, my favorite piece was in the Economist's Millenium edition, entitled British Irony. An excerpt:
The ironic tone has become a staple of British literary style, but can still cause considerable confusion overseas--as writers for this newspaper occasionally discover. A recent article on Paul Gascoigne, a footballer who had been caught beating his wife, began: "It could happen to anybody, really. Go out for a meal with the wife, have a few too many, she starts to nag, and before you know what's happened, she's lying on the floor covered in bruises." It did not occur to the author that anybody might seriously regard this as an endorsement of wife-beating--at least not until the outraged letters began arriving from the United States.
The article discusses some of the finer points of irony, noting for instance that "...irony is much more subversive than sarcasm, and also much more fun--those who realise that an ironic remark has been made are instantly complicit, and they can enjoy the fact that there are others who have missed the joke."
Take a look at Club Photo. They have limited but decent free services, and also some pay services. You have to subscribe to be able to download at high resolutions. But somehow I have more faith in the longevity of a service that charges something than in the free ones!
I've used them once by mailing in film; the prints were decent and I also got a CD of the hi-res scans.
While power control is definitely a big issue in IS-95, I am told that progress has been made since then in alternate solutions to the near-far problem. In particular, there is an "onion-peeling" approach that involves decoding the most powerful received signal first, and then subtracting that out of the received signal in order to decode the weaker signals.
All of this still leaves us within the bounds of Shannon capacity. The multiple antenna stuff has promise of orders of magnitude better performance.
The article does a terrible job of describing some remarkable recent progress.
CDMA systems showed us that it is possible to transmit two signals at the same time and the same frequency and distinguish them at the receiver; a task which at first might seem impossible. However, Shannon's theory still imposes limits on the maximum possible transmission rate.
What's new today is that by using multiple antennas it is actually possible to go beyond the limits Shannon established for point-to-point communication! This is not snake oil; it is well established, refereed research. In fact, it is already demonstrated technology!
I still think it is a long, long way from these ideas to an unregulated spectrum.
I'm Indian, and English is my native language. It's the language I think in, and the language I speak with my parents. Admittedly, only a small (but growing) population would consider English their primary language, but consider the following:
Most major colleges use English as their medium of instruction. So the students don't just study the English language, they also study their economics, science, and history in English.
Most businesses (not just the IT industry) use English as the medium of communication. My dad worked for years in a cement company, and every last memo he ever wrote was in English.
With all the languages that exist (most states have their own), English is really the only common thread. Most college grads are also comfortable with Hindi, but Hindi is less commonly spoken in the South. As an aside, it is not unusual to witness a conversation where one person is speaking Hindi and the other English without either even realizing it.
Just about the only "language" an American might usefully learn to communicate with Indian colleagues would be British English; that way you won't get shocked when we ask you for a rubber...
(For those not in the know, a rubber is an eraser in India, but a condom in the States.)
Remember the Slashdot story from the other day about the RIAA press release touting the capture of the equivalent of 421 CD burners, when only 156 had actually been found?
Well, in the CNN article this story links to, that has become simply "421 compact disc burners." No mention of "equivalency" anywhere.
All streaming players support this to some extent, you just switch the buffering time to be very low, or zero.
"Instant-on" is not quite the same thing as setting your buffer size to zero. Typically, setting the buffer size to zero in a player means that there is no buffering for the entire duration of the program. This means that if you want to watch without interruptions, the available bandwidth between the server and you must exceed the bit rate of the stream at all times; since the encoding is at a variable bit rate, you will often see peaks in the bit rate which need really high bandwidth to sustain the stream.
However with "instant-on", the playback begins immediately, but the buffer continues to build. This means that you need a fairly high bandwidth initially in order to start the playback and build up the buffer, but--after that initial period--the buffer prevents interruptions which would otherwise have been caused by the peak rate of the stream going beyond the available bandwidth. So "instant-on" is not the same as setting your buffer to a fixed size of zero. Now this isn't really revolutionary; to researchers in the field the question is why everyone hasn't been doing this all along.
As an aside, the "skip protection" feature of Quicktime streaming is simply over-buffering; i.e. using the available bandwidth to build up the client buffer as much as possible instead of maintaining a fixed size buffer.
QuickTime doesn't really have any major advantages over other technologies as far as I can tell, other than it being made by Apple and therefore given lots of marketing. Why should I choose QuickTime over Real, or Windows Media, or hell even standard MPEG?
Quicktime streaming has an advantage to broadcasters: the server is open-source and free as in beer. You can download the latest snapshot via CVS, and contributions to the source from the public are welcome. The streaming protocols are all standards based (RTP/RTSP). Of course, you could always encode the data with a proprietary codec, but if you used an open codec, then the streams could be played back by any standards compliant player, not just Quicktime Player.
Finally, although ideas like "skip-protection" and "instant-on" are fairly obvious to anyone who spends some time thinking about these issues, the fact remains that only Apple seems to be taking the initiative to incorporate these into its servers and clients. What new features (from a streaming perspective) have Real and Microsoft offered in the new releases of their products? While Apple does have a big marketing push, that does not take away from the fact that there is solid work going on behind the scenes.
Specifically, our hearing demonstrated that there are three discrete problem areas that merit government intervention.
The three areas he proceeds to list correspond exactly to the three goals Jack Valenti stated in his testimony. (You can read Valenti's testimony.) I got in too late to hear everyone's testimony (Valenti went last), but I doubt that those three areas were a general consensus. That section of Hollings' statement was essentially written by the MPAA.
One other point: the parent post is right - this bill naming thing is really important. Remember the USA PATRIOT act? I attended a talk by Russ Feingold (D-Wisconsin) who spoke about how difficult it was to oppose a bill named that way (I believe he was the only senator who voted against it.) The current bill's title sounds like something
everyone would want, making it that much more difficult to oppose.
I agree with the parent that the tool for the job depends on what you want to do, but I must disagree with the parent's evaluation of Matlab and Mathematica. I have used both, and here are my thoughts on them.
Both have their own language backed by a huge collection of built-in functions. Mathematica's strength is in symbolic calculations: Solve[a x^2 + b x + c ==0,x] will give you the standard formula for a quadratic in a,b, and c. Matlab excels in numerical computations, especially on vectors and matrices. You can do numerical stuff in Mathematica and symbolic stuff in Matlab, but those aren't their strengths.
I would not say that either is easier to program than the other. You can write programs for either using your favorite text editor. One thing common to both languages is that though they have "for" loops, such loops are generally an inefficient way of doing things. Applying functions to entire lists in Mathematica or to vectors in Matlab will be a lot more efficient. You will have to get used to this "vector" way of thinking.
Whether either of these is applicable to the ask slashdotter's situation depends very much on his goals. Solutions to algebraic equations become one-line commands in these languages; if you are seeking the thrill of programming the solution yourself from first principles, use one of the standard programming languages others have described. If you want to make deeper investigations, you will likely get there sooner using one of these programs than starting from scratch.
Both packages also allow you to call (and call from) C programs - useful for intensive stuff that is better done directly in C.
Also worth noting is that both these software packages cost money! Your university may have them on their lab computers, but if you want to use them (legally) at home, you will have to shell out $100-$150 for a student license. Thankfully the student versions are pretty much full-featured. Your university might have special deals with these companies; mine sells student Matlab for about $30. Single non-student licenses are very expensive (about $1000 for Mathematica and $2300 for Matlab, I believe)
Also, the release notes for AppleScript Studio states as a "known issue" that "AppleScript Studio does not currently support other OSA languages."(emphasis mine) This holds out hope that this excellent tool will support Perl etc in the future.
MANETs have been a "hot" research area for a while in the academic community. As others have pointed out, there are many problems to surmount.
You'll find that a lot of the research is funded by the army. They are very interested in these as a way to set up communications among an isolated troop in the middle of nowhere. They really want small non-power-hungry devices with multimedia capabilities that communicate in peer-to-peer fashion.
I suspect that you are referring to the issue of 2600 posting "links" to DeCSS code. This is a fascinating issue
Of late, the NYTimes online has stopped posting "clickable" links along with their stories (i.e. no href tags). They simply include the URL in the text (like this: www.nytimes.com). As a side effect of Smart Tags, we will undoubtedly have the browser converting such strings into active links, much like many email clients do nowadays. If the browser on the vast majority of the population's computer's does this, it becomes effectively impossible to post a link which requires a copy and paste to activate.
Now, this might well be regarded as a good thing by many; but it removes the loophole that 2600 exploited to get around Judge Kaplan's ruling. Similarly, I'm sure the NYT does the "non-linking" only to protect their legal backside, not to annoy.
Here's hoping that if Smart Tags come to pass, we won't have some judge ruling that the 2600 page is now in violation of the law.
When OS X Final was released, a lot of folks with dual processor Macs reported kernel panics when getting online via dial-up. A Darwin developer (Louis Gerbarg), who is not an Apple employee, tracked down the bug in an open source kernel extension, and fixed it. No big deal in the Linux world, but possibly a first for a commercial OS (again, this was a kernel level bug!)
End result: Lots of happy OS X users (who paid upto $129 a pop to buy the software), without any involvement from Apple! Talk about a win-win...
Re:Proof of optimality (was Re:Mod this way up)
on
The Three Hat Problem
·
· Score: 1
Your proof also explains the following quote from the NYT article:
Still, it all comes down to making sure that most of the time no one is wrong and occasionally everyone is wrong at once.
As it turns out, this requirement can be perfectly met only when the number of players is one less than a power of two (three, seven, 15 and so on.)
To achieve the N/(N+1) bound, we must have W=NL. Since W+L=2^N, this gives W=N 2^N / (N+1). But W must be an integer. Since N and N+1 are relatively prime (they have no common factors), the only way this can happen is if N+1=2^m, for some m.
Re:Proof of optimality (was Re:Mod this way up)
on
The Three Hat Problem
·
· Score: 1
That's nice. It took me a minute to convince myself that no matter what, the number of correct guesses must equal the number of incorrect guesses. If anyone else harbors doubts, consider the two assignments 0000000 and 1000000. Since player 1 must base his decision on only the other hats, he must make the same guess (or pass) in both cases...which means that for every correct guess there is an incorrect one, regardless of strategy.
Oh, and thanks for the nice comments, GodSpiral; you're very welcome!
Yup, I think you're right on both counts: the two conditions imply that the code is perfect, and non-linear perfect codes (such as your example)exist.
For those wondering, Blahut(*) defines a perfect code as "one for which there are equal-radius spheres about the codewords that are disjoint and that completely fill the space." In our case, that radius is of course, 1.
As an aside, I haven't given any thought to showing that the Hamming code strategy is optimal...ideas, anyone?
(*) R.E. Blahut,"Theory and Practice of Error Control Codes," Addison-Wesley, 1983.
I suspect the moderators have all left the building...oh well.
Your construction does not generate a valid Hamming code, for reasons I'll get to in a minute. I haven't given this much thought, but I suspect that the two facts I listed in my post do not completely specify Hamming codes, i.e. there might be non-Hamming codes that meet those criteria. However since the proposed strategy fails for precisely those hat assignments that are associated with codewords, improvement over the Hamming code can only be achieved if a code with fewer than 16 codewords meets those criteria. (By "improvement", I mean a higher percentage of hat assignments that result in a reward for the team).
On why your construction does not represent a valid Hamming code: Hamming codes are linear codes, in that the bitwise, modulo-2 sum of 2 codewords is also a codeword. In particular, the sum of a codeword with itself must be a codeword; but that sum is always the all-zero word, because addition is modulo-2 (an XOR). So the all zero word is a codeword. Taken with Fact 1 of the previous post, this means that words with a single 1 in them are not codewords in a Hamming code.
The general definition of Hamming codes is in terms of matrices. However, the n=7 case is easily described in terms of parity check bits. Call the first 4 bits i1 through i4--these are the information bits. Call the next three bits p1 through p3--the parity bits. To construct the Hamming code take the 16 four bit sequences formed by i1 through i4, and append the 3 parity bits generated according to the formulas: p1=i1+i2+i3, p2=i1+i2+i4, and p3=i1+i3+i4, where all additions are again modulo 2.
The general construction of binary Hamming codes of length n=2^k-1 specifies the contents of a matrix G with n rows and n-k columns, from which the codewords may be generated by the matrix multiplication Gv, where v is any length n-k column vector. The structure of G is fairly simple, but a little too much trouble to specify in HTML, so I'll just refer anyone interested to any coding theory text.
Slashdot Mirror
What we have here is that, in addition to doing this, Windows is also offering to set up an ad-hoc (i.e. computer-to-computer) network on the link-local subnet with the same SSID as that of the last network the laptop connected to. I wonder what the rationale for doing this could have been. It seems to me that a machine should not offer to set up an ad-hoc network unless specifically directed to do so by the user. When such a network is set up then it is appropriate to use link-local addressing to auto-configure the interface.
Textpattern may be what you are looking for. Although it is used for blogs, it is very easy to setup as a general purpose CMS. And it seems to have good unicode support for your multilingual needs.
As for the details of what has been discussed so far in the 11s task group, anyone can sign up for an account at 802wirelessworld, and obtain access to all the documents submitted for consideration to the task group so far. (Once you register and login, a link for Documents shows up under 802.11 WLAN WG on the left of the page.)
Various usage scenarios have been considered, from the scale of the home ( a few devices) to larger scale community meshes. The standard will work on any "mesh-aware" point, which may be an AP or a client device. It will likely run at layer 2 (below the IP layer) and provide a standards based mechanism for multi-hop access to a wired gateway (or "mesh portal" as they refer to it).
- Create a P2P system that requires a password for access (Windows file sharing should do!)
- Encrypt the password
- Leave the encrypted password and the decryption key accessible
Now, the password itself is a copyrighted work of my creation. So using the decryption key to crack the password should be a DMCA violation. Thus if people (including the RIAA!) use my decrypted password to access my files, it is they who are violating the law, not me, i.e. I cannot be accused of making copyrighted works publicly available.Thoughts? (and apologies if this is how some of the more advanced systems already work; I am not familiar with them)
The article discusses some of the finer points of irony, noting for instance that "...irony is much more subversive than sarcasm, and also much more fun--those who realise that an ironic remark has been made are instantly complicit, and they can enjoy the fact that there are others who have missed the joke."
If you like this sort of thing, check out the PostScript Fractals page. You can print out very detailed images from tiny PostScript files.
I've used them once by mailing in film; the prints were decent and I also got a CD of the hi-res scans.
All of this still leaves us within the bounds of Shannon capacity. The multiple antenna stuff has promise of orders of magnitude better performance.
CDMA systems showed us that it is possible to transmit two signals at the same time and the same frequency and distinguish them at the receiver; a task which at first might seem impossible. However, Shannon's theory still imposes limits on the maximum possible transmission rate.
What's new today is that by using multiple antennas it is actually possible to go beyond the limits Shannon established for point-to-point communication! This is not snake oil; it is well established, refereed research. In fact, it is already demonstrated technology!
I still think it is a long, long way from these ideas to an unregulated spectrum.
- Most major colleges use English as their medium of instruction. So the students don't just study the English language, they also study their economics, science, and history in English.
- Most businesses (not just the IT industry) use English as the medium of communication. My dad worked for years in a cement company, and every last memo he ever wrote was in English.
- With all the languages that exist (most states have their own), English is really the only common thread. Most college grads are also comfortable with Hindi, but Hindi is less commonly spoken in the South. As an aside, it is not unusual to witness a conversation where one person is speaking Hindi and the other English without either even realizing it.
Just about the only "language" an American might usefully learn to communicate with Indian colleagues would be British English; that way you won't get shocked when we ask you for a rubber...(For those not in the know, a rubber is an eraser in India, but a condom in the States.)
Well, in the CNN article this story links to, that has become simply "421 compact disc burners." No mention of "equivalency" anywhere.
Sigh.
I can see it now.
"No, your honour, I didn't make an obscene gesture. I had just gotten done counting to 27..."
"Instant-on" is not quite the same thing as setting your buffer size to zero. Typically, setting the buffer size to zero in a player means that there is no buffering for the entire duration of the program. This means that if you want to watch without interruptions, the available bandwidth between the server and you must exceed the bit rate of the stream at all times; since the encoding is at a variable bit rate, you will often see peaks in the bit rate which need really high bandwidth to sustain the stream.
However with "instant-on", the playback begins immediately, but the buffer continues to build. This means that you need a fairly high bandwidth initially in order to start the playback and build up the buffer, but--after that initial period--the buffer prevents interruptions which would otherwise have been caused by the peak rate of the stream going beyond the available bandwidth. So "instant-on" is not the same as setting your buffer to a fixed size of zero. Now this isn't really revolutionary; to researchers in the field the question is why everyone hasn't been doing this all along.
As an aside, the "skip protection" feature of Quicktime streaming is simply over-buffering; i.e. using the available bandwidth to build up the client buffer as much as possible instead of maintaining a fixed size buffer.
QuickTime doesn't really have any major advantages over other technologies as far as I can tell, other than it being made by Apple and therefore given lots of marketing. Why should I choose QuickTime over Real, or Windows Media, or hell even standard MPEG?
Quicktime streaming has an advantage to broadcasters: the server is open-source and free as in beer. You can download the latest snapshot via CVS, and contributions to the source from the public are welcome. The streaming protocols are all standards based (RTP/RTSP). Of course, you could always encode the data with a proprietary codec, but if you used an open codec, then the streams could be played back by any standards compliant player, not just Quicktime Player.
Finally, although ideas like "skip-protection" and "instant-on" are fairly obvious to anyone who spends some time thinking about these issues, the fact remains that only Apple seems to be taking the initiative to incorporate these into its servers and clients. What new features (from a streaming perspective) have Real and Microsoft offered in the new releases of their products? While Apple does have a big marketing push, that does not take away from the fact that there is solid work going on behind the scenes.
Hollings' statement states
The three areas he proceeds to list correspond exactly to the three goals Jack Valenti stated in his testimony. (You can read Valenti's testimony.) I got in too late to hear everyone's testimony (Valenti went last), but I doubt that those three areas were a general consensus. That section of Hollings' statement was essentially written by the MPAA.One other point: the parent post is right - this bill naming thing is really important. Remember the USA PATRIOT act? I attended a talk by Russ Feingold (D-Wisconsin) who spoke about how difficult it was to oppose a bill named that way (I believe he was the only senator who voted against it.) The current bill's title sounds like something everyone would want, making it that much more difficult to oppose.
Both have their own language backed by a huge collection of built-in functions. Mathematica's strength is in symbolic calculations: Solve[a x^2 + b x + c ==0,x] will give you the standard formula for a quadratic in a,b, and c. Matlab excels in numerical computations, especially on vectors and matrices. You can do numerical stuff in Mathematica and symbolic stuff in Matlab, but those aren't their strengths.
I would not say that either is easier to program than the other. You can write programs for either using your favorite text editor. One thing common to both languages is that though they have "for" loops, such loops are generally an inefficient way of doing things. Applying functions to entire lists in Mathematica or to vectors in Matlab will be a lot more efficient. You will have to get used to this "vector" way of thinking.
Whether either of these is applicable to the ask slashdotter's situation depends very much on his goals. Solutions to algebraic equations become one-line commands in these languages; if you are seeking the thrill of programming the solution yourself from first principles, use one of the standard programming languages others have described. If you want to make deeper investigations, you will likely get there sooner using one of these programs than starting from scratch.
Both packages also allow you to call (and call from) C programs - useful for intensive stuff that is better done directly in C.
Also worth noting is that both these software packages cost money! Your university may have them on their lab computers, but if you want to use them (legally) at home, you will have to shell out $100-$150 for a student license. Thankfully the student versions are pretty much full-featured. Your university might have special deals with these companies; mine sells student Matlab for about $30. Single non-student licenses are very expensive (about $1000 for Mathematica and $2300 for Matlab, I believe)
Also, the release notes for AppleScript Studio states as a "known issue" that "AppleScript Studio does not currently support other OSA languages."(emphasis mine) This holds out hope that this excellent tool will support Perl etc in the future.
You'll find that a lot of the research is funded by the army. They are very interested in these as a way to set up communications among an isolated troop in the middle of nowhere. They really want small non-power-hungry devices with multimedia capabilities that communicate in peer-to-peer fashion.
Of late, the NYTimes online has stopped posting "clickable" links along with their stories (i.e. no href tags). They simply include the URL in the text (like this: www.nytimes.com). As a side effect of Smart Tags, we will undoubtedly have the browser converting such strings into active links, much like many email clients do nowadays. If the browser on the vast majority of the population's computer's does this, it becomes effectively impossible to post a link which requires a copy and paste to activate.
Now, this might well be regarded as a good thing by many; but it removes the loophole that 2600 exploited to get around Judge Kaplan's ruling. Similarly, I'm sure the NYT does the "non-linking" only to protect their legal backside, not to annoy.
Here's hoping that if Smart Tags come to pass, we won't have some judge ruling that the 2600 page is now in violation of the law.
End result: Lots of happy OS X users (who paid upto $129 a pop to buy the software), without any involvement from Apple! Talk about a win-win...
Oh, and thanks for the nice comments, GodSpiral; you're very welcome!
For those wondering, Blahut(*) defines a perfect code as "one for which there are equal-radius spheres about the codewords that are disjoint and that completely fill the space." In our case, that radius is of course, 1.
As an aside, I haven't given any thought to showing that the Hamming code strategy is optimal...ideas, anyone?
(*) R.E. Blahut,"Theory and Practice of Error Control Codes," Addison-Wesley, 1983.
Your construction does not generate a valid Hamming code, for reasons I'll get to in a minute. I haven't given this much thought, but I suspect that the two facts I listed in my post do not completely specify Hamming codes, i.e. there might be non-Hamming codes that meet those criteria. However since the proposed strategy fails for precisely those hat assignments that are associated with codewords, improvement over the Hamming code can only be achieved if a code with fewer than 16 codewords meets those criteria. (By "improvement", I mean a higher percentage of hat assignments that result in a reward for the team).
On why your construction does not represent a valid Hamming code: Hamming codes are linear codes, in that the bitwise, modulo-2 sum of 2 codewords is also a codeword. In particular, the sum of a codeword with itself must be a codeword; but that sum is always the all-zero word, because addition is modulo-2 (an XOR). So the all zero word is a codeword. Taken with Fact 1 of the previous post, this means that words with a single 1 in them are not codewords in a Hamming code.
The general definition of Hamming codes is in terms of matrices. However, the n=7 case is easily described in terms of parity check bits. Call the first 4 bits i1 through i4--these are the information bits. Call the next three bits p1 through p3--the parity bits. To construct the Hamming code take the 16 four bit sequences formed by i1 through i4, and append the 3 parity bits generated according to the formulas: p1=i1+i2+i3, p2=i1+i2+i4, and p3=i1+i3+i4, where all additions are again modulo 2.
The general construction of binary Hamming codes of length n=2^k-1 specifies the contents of a matrix G with n rows and n-k columns, from which the codewords may be generated by the matrix multiplication Gv, where v is any length n-k column vector. The structure of G is fairly simple, but a little too much trouble to specify in HTML, so I'll just refer anyone interested to any coding theory text.
You could think of the players as being tagged, and the info available to each player being: Al's hat is red, Betsy's hat is blue, and so on...
Shameless plug: see my solution for the general case using Hamming codes.