I honestly think Windows is harder to install than most distributions of Linux. I have been dual-booting Linux and BeOS for a couple years now. Recently I wanted to install Unreal, so I started to install Windows 95. Halfway through the installation, it simply hangs. No error messages, no information about what it was doing when it crashed, nothing. That is not ease of use.
Back to the topic under discussion, easy to use graphical tools are fine if designed properly. Tools should encapsulate the complexity of the entire problem, and then organize this complexity into an understandable format. Simple operations should be simple to perform using the tool, and complex operations should be doable but complex using the tool. Tools should not be a black-box. They should be like a shotgun shell loader - you pull the level, and lots of things happen to reload the shell. But you can see everything that happens, because sooner or later things will get jammed and you'll have to reach in and fix it.
Comments on Slashdot should be moderated by smell rather than number. Think a comment sucks? Make it stinky! Is it another karma whore? Make it smell like cheap perfume.
This is a niggly point, but a mistake that many people make. It is not legal to redistribute MP3's you download from mp3.com! The terms of use specifically prohibit you from redistributing the files.
He makes his money running a computer security company that provides security solutions. He might not make a dime when someone implements his algorithm, but how valuable is it to the company to be able to say that they developed a major cryptographic standard?
There is nothing wrong with being proud of your ideas and advocating them. The problem is when he uses his cryptography newsletter to advocate his own pet algorithms. Nothing wrong with that, but maybe he should be more open and mention when he's talking about products he developed.
Can someone answer me this question please? I really want to know. Why don't people make browsers for Windows that are as good as IE by using the "built-in" routines that are "part of the OS"? Why can't Netscape use the IE routines (since they are part of the OS and are freely available) and then add cool stuff on top of that? I know that Gecko is competing with the rendering engine of IE, but in the meantime until Mozilla is finished why not embrace and extend IE?
People bat around the word "conservative" quite a lot without having a clue what it means. Mandating the use of filtering software in public libraries is not conservative. True conservative towns don't keep computers in their libraries at all, let alone ones connected to the Internet. The true conservative solution is the one the town is already implementing. When some idiot looks at pornography in the library, the librarian walks by with a stern look.
Jan. 8, 2000 - The ACM has just announced that the emminent computer scientist Fred Brooks has won the Turing Test Award. Alan Turing is the originator of the celebrated "Turing test" in which a judge must distinguish between a computer program and a real human being.
The actual test took place last week, with Donald Knuth playing the role of the judge. After 10 grueling hours of questioning, Knuth declared that Fred Brooks was virtually indistinguishable from a computer program.
"It gives me great pleasure to accept this award on behalf of computer scientists everywhere./* Fix up speech here, add some thank-you's and stuff. Re-check grammar. */," said Dr. Brooks on being presented with his prize. "I have always had a knack for fast arithmetic, and a lifetime of dealing with computers has made me doubt if I was human myself." Some members of the press commented that Dr. Brook's lack of social skills couldn't have hurt his chances either.
Fred Brooks is the first human to legitimately pass the Turing test. Last month a man known only as HemostheRobotMan claimed to have passed the Turing test, but was later disqualified after he was found using a Palm Pilot hidden in his shorts. Fred Brooks caused a stir when he said, "I need to change my batteries." It turns out he was merely hungry.
Wavelets are an exciting idea. Standard lossy compression algorithms (like JPEG) use Fourier transforms to transform the data to the frequency domain. Then low coefficients are ignored, which degrades the image slightly and saves space.
Wavelets work in a similar fashion, using the wavelet transform. The difference is that wavelets are localised in space, unlike frequencies which are infinite sine waves. Thus when you forget the low coefficients and only save the highest wavelet coefficients, you can still retain things like discontinuities and local detail. The wavelets representing the local discontinuity have high coefficients and so are kept. In the frequency domain (used in Fourier transforms), discontinuities have components in all frequencies. When JPEG filters out the low coefficients, sharp discontinuities become blurred.
This is very cool to see wavelet compression start to move into the mainstream. One application already in use: the FBI uses wavelet compression to store fingerprint data (source: Wavelet Toolbox for Matlab documentation). Another common application of wavelets is removing noise from signals. Standard frequency filtering removes the noise and part of the signal. Wavelet de-noising can remove just the noise (it's really pretty amazing). I had an internship at the University of Washington where I used wavelets to remove RF interference from experimental physics data.
I would like to see a system like you describe, with posts bouncing between many nodes totally anonymously before being sent to the final destination. It would be nice if there were a network run by many people, and a message sent to any one address would bounce through as many links as possible. If one site got shut down, it wouldn't matter. It would also be hard to press legal action against any one site, because only one site would know the message origination, and one different site would know the message destination.
If the network was designed correctly, it would be impossible for any one site to know where the traffic is coming from or going. Someone trying to break the anonymity through legal means would have to subpoena all the sites. Messages could be limited to 10K each, and only 1 message every 10 minutes per IP address (or something to prevent spam).
How severe should pain be before you see a doctor? I occassionally feel sore in the wrists after heavy bouts of typing (usually from typing up something due the next day). I get some sleep and rest, and the pain goes away. With my new keyboard this happens maybe once a month. Is this enough pain to warrant a trip to the doctor? Should you really have a "zero tolerance" for pain?
I think people have some misunderstanding of how analog->digital->analog reproduction happens. There are two kinds of distortion introduced in converting an analog signal into a digital signal and back again. The basic idea behind the conversion is sampling. A sample is taken every t seconds (where t might be 1/44100). Each sample is represented by b bits.
Quantization error - this happens because b is not infinite. There is less quantization error in a 16-bit sample than in an 8-bit sample.
Aliasing artifacts - these are signals introduced at the output that were not present at the input of the system. They arise because any frequency in the input higher than the Nyquist frequency will be reflected back into the 0-Nyquist range.
By increasing the number of bits used per sample, b, you help eliminate quantization errors. By increasing the sampling frequency (decreasing t) you eliminate aliasing effects.
The Nyquist limit is half of the sampling frequency. To represent a frequency at the Nyquist frequency, it is necessary to have samples alternate like +1,-1,+1,-1,etc.
The previous poster wondered about phase information when you get close to the Nyquist limit. The answer is that if you have infinite precision samples (each sample is a real number), then all phase and frequency information below the Nyquist frequency is preserved. Suppose you have 10 sample points on a high-frequency sine wave (but that is still below the Nyquist frequency). Shifting the sine wave left and right slightly (altering the phase) makes the sample points go up and down. If you have accurate sample point resolution, you can exactly restore the phase of the signal (and its frequency).
If you try to sample a frequency that is higher than the Nyquist frequency, you will get sample points that look like another frequency between 0 frequency and the Nyquist frequency. There is no way to determine if the sample points represent the lower or the higher frequency. The solution is to filter out all frequencies higher than the Nyquist frequency at the input.
My friend's parents ordered an iMac, and didn't open the box until my friend returned from college. When she opened the box, one of the plastic pieces was broken. When she called UPS, they said it was too late to claim shipping damages. When she called Apple, they said that the plastic pieces are not under warranty, because they are supposed to last the life of the computer. It took her ALL DAY long on the phone with Apple to finally get them to replace the computer.
What exactly is the point of banning strong crypto products? I was under the impression that the reason crypto is "bad" is because "bad people" can use it for nefarious purposes. Aren't there any "bad people" in the US? If people in the US can use crypto for benign purposes, why can't people in other countries do the same?
It might make sense to restrict the sale of cryptographic products going to governments with a known track record of abusive behavior. Anything more than this is lunacy. It appears (from my reading of the article) that Netscape won't be able to ship a strong-crypto version of Navigator to people in Cuba. Where is the logic there? The Cuban government might get a hold of it and be able to order missiles from www.nukes.com and not have to worry about their credit card number being stolen?
I agree with you that encryption must be really easy to use before it becomes standard. As mentioned, the way public keys work is that you encrypt something with someone's public key, and then only that person can decrypt it. So even if email software automatically grabbed keys, if clueless users did not have public keys there is no way they could receive encrypted email.
I think the real solution is to add encryption to the underlying mail protocols. This probably won't happen for a while. The idea would be that to have an email address, you must have a public key.
When clueless Joe User signs onto the Internet via his local ISP, he wants an email account. To get his email account, he must enter a passphrase into his computer. This becomes his private key (which he can remember and use anywhere). His computer then computes the corresponding public key, and sends it to his ISP. The ISP then creates his mail account, and sends his address plus public key to the nearest trusted keyserver.
Whenever email is sent, the appropriate keyserver is queried and the email is automatically encrypted. This way encryption is the default, not plain text. Of course this way there are problems with people forgetting their passphrase, etc. This would also rely on the keyserver's accuracy. So someone could put up an insincere keyserver that always responds, "Yes, I know the public key for that address, it is 0xdeadbeef" where 0xdeadbeef is the public key of the owner of the keyserver.
Widespread encryption is inevitable given the way people (and governments) abuse the situation now. I predict that within 3-5 years most computer-savvy people with use PGP or something equivalent for their normal correspondence. Within 8 years encryption will be the default, and it will be harder to send an email without encryption than with encryption.
What I would like to see is an organization that has a contractual obligation to revoke its endorsement of any company found to violate the public's trust. Individual consumers should be able to donate money to the organization. In return they get the assurance that any site they visit with the TRUSTe logo really is following good privacy guidelines. If some company is found to violate the guidelines, then TRUSTe must revoke its endorsement or face lawsuits by the individuals. This way the companies are obligated to stay honest, and TRUSTe is obligated to stay honest or face lawsuits. -Nathan Whitehead
Here is a possible protocol for an open-source program to authenticate itself.
Concatenate the binary of the program with the current time (in Universal Time), put in X
Hash X using a one-way hash function
Send hash value as certificate
This protocol does not work, because if you have the source code for the hashing function and a copy of the "real" binary, you can spoof the authenticator. Is there a solution?
If there is a kernel or library function that hashes the binary code of the calling process, it is possible. The hashing function could be open-source, but the compiled function would have to include an embedded constant known only to the authenticator.
I don't know how feasible that is. The idea is if you hacked the source of the hash function and recompiled, you would lose the embedded constant. Of course if you hack the original program, when you call the hash function it will return the wrong value.
Put simply, TeX brought typesetting to the masses. Knuth invented TeX because he was tired of sending manuscripts with complicated equations to the typesetter and getting back garbage. TeX allows anyone with a little patience and willpower to typeset virtually anything they want. (I use LaTeX, an extension to TeX, so my comments might only be applicable to LaTeX).
Advantages of TeX
There are many advantages of (La)TeX over other document creation systems.
TeX output looks great. Nothing looks more professional than TeX.
Once you know the basics it is faster to create complicated documents with LaTeX than with anything else. LaTeX automatically handles hyphenation, line spacing, section title placement, equation numbering, figure placement and numbering, etc. When you want to refer to "equation 5", you refer to it by name. When you later add a new equation before "equation 5", your reference is automatically updated. Some word processors have this feature, but it is more of a "tacked on" thing than in TeX.
Typing in equations is much faster in TeX. Once you know the names of the symbols, it is possible to type mathematical equations faster than writing them on paper. TeX has every symbol imaginable. With Word (for example), you must hunt-and-peck for symbols on a chart. If your symbol isn't there, you're SOL.
TeX includes many many packages that other people have written for just about anything you could possibly want TeX to do. Want to typeset some Java code so keywords are bold, comments are italics, and everything else is roman font? There's a package to do it. How about a package to draw graphs? Yes. All this is possible because TeX is a programming language for typesetting. Fortunately, you don't need to use all its features just to produce a document. But the power is there, waiting to be used.
Customizability. If you find yourself typing "with regard to" over and over, you can define a command \wrt that expands to "with regard to". Suppose you invent some new notation. Then just define a command for it and never worry about how the notation is generated. Then if you ever want to change your notation, you can change the definition. All your notation will instantly change, without any of the bother of searching for each time you used the notation. This is incredibly useful. Before I printed out my thesis, I previewed it with vectors written with an arrow above the vector quantity, and with vectors in boldface type. It was a one-line change.
I think that TeX is most appropriate for creating documents that clearly and simply convey information. It is the ideal tool for writing papers to be published in academic journals. Everything about TeX is designed with communication in mind. The default margins are wider than most other word processors. This is because people have a hard time reading wide columns of text. Look at any good book and count the number of words across a column. It will be the same as the number of words in a default TeX column.
Disadvantages of TeX
There is a learning curve. TeX is a deep system. Learning TeX is a never-ending journey. Initially it can be painful and unproductive. Many people get discouraged.
Using TeX is not interactive. TeX by itself is a "compiler". Given a source file, it produces an output file suitable for printing. Writing TeX can often become more frustrating than programming. Silly syntactic errors can stop the compile from completing. It gets old reTeXing after every change in your file.
It is often hard to make formatting changes. The formatting system is complex. For example, to change margins or line spacing is not as simple as grabbing a slider.
It is sometimes hard to tell what a given section of source file produces. \delta is different than seeing the delta symbol.
There are several expensive commercial systems based on TeX that provide an easy-to-use interface. There is also LyX, which is GPL. I do not have much experience with these, so I don't know effective they are.
When I bought my latest computer last year, I naively decided to get a SB Live! since it seemed like such a cool card, and it was by Creative, so how could there not be Linux drivers for it?
When I discovered there were no drivers, and that Creative was not releasing the specifications, I was understandably disappointed. After the first beta drivers came out, I could at least get some sound out of the card. But some games didn't work with it, and it still had less functionality than the AWE32 driver. So my younger brother got a free upgrade from AWE32 to SB Live!, and I got a free downgrade that gave me more functionality. If the driver at that time had been open source, I honestly would have spent my whole summer hacking it to do cool stuff.
My point? This is a VERY SMART MOVE by Creative. There are a LOT of DSP hackers in college, just itching to write cool DSP code. Sound has been an often neglected field of hacking. We are on the verge of a Renaissance in sound. Under Linux, there is a good abstraction layer between the programmer and the soundcard. All the demo programmers, trackers, general code hackers, etc. are coming to Linux. The Linux crew has traditionally been made up of more 'mature' programmers. But with the rise in popularity of Linux, the migration is inevitable. And these people will bring new ideas of what is "cool" to program. Look for lots of cool sound programs in the next couple years.
There are reasons it is better to use the GPL for projects that aren't designed to be libraries. For example, suppose I develop a specialized mathematics program. If I release it under the LGPL, it is possible for Wolfram (for example) to rip out the interesting routines and make it into a library. They LGPL their changes to my program, then use that library in Mathematica without my consent or even mentioning that I wrote that code!
This might not be what I intended. If I release it as GPL software, they would have to release Mathematica as GPL software to do this. Wolfram would never do something this, but the possibility that some company might is something to consider. Of course, you might WANT anyone and everyone to take advantage of the code, so then you would release it under a LGPL or BSD license. I think it is entirely reasonable to have both the GPL and LGPL licenses.
Back in the "good old days" of cryptography, the algorithms used were understandable by non-mathematicians. Most modern cryptographic systems in use are still mathematically "simple". By this I mean that once you understand the complexities of the algorithm, the mathematical basis is understandable to someone who has, say, a college degree in mathematics or physics.
The cryptographic systems being developed today are often based on much more sophisticated mathematical ideas. Elliptic and hyper-elliptic curves spring to mind. The algorithms may be understandable, but the mathematical basis may be complicated enough that it takes a PhD in mathematics to understand.
These systems are the future generation of cryptography. Some have suggested that their security is based more on mathematical obscurity than anything else (i.e. the number of people able to even understand what the algorithm is doing is very small). Do you think this is accurate? Do you see cryptography moving exclusively into the domain of mathematicians, so that it is totally inaccessible to motivated non-mathematicians (such as yourself)?
My girlfriend broke up with me a couple weeks ago. This article was exactly what I needed to hear. It is guy-talk at its finest. Based on my own experience, this article matches with reality pretty closely. The biggest mistake 'geek' guys make is not catching social signals. They either over- or under-interpret things.
Thank you for the cool article RobLimo! Although maybe it should have included the disclaimer, "FOR GUYS ONLY!" If women only knew what guys REALLY thought...
Microsoft made its millions by being chosen as the operating system for IBM's PC. It made its billions by developing Windows and branching into application programming (Microsoft Office). Now it seems like Microsoft is trying to grab its piece of many different pies.
What is Microsoft's core business? If it is operating systems, why are Win2000 and NT losing respect so rapidly? If it's application development, why isn't Microsoft opening up Windows to be a standard for everybody?
Slashdot Mirror
Back to the topic under discussion, easy to use graphical tools are fine if designed properly. Tools should encapsulate the complexity of the entire problem, and then organize this complexity into an understandable format. Simple operations should be simple to perform using the tool, and complex operations should be doable but complex using the tool. Tools should not be a black-box. They should be like a shotgun shell loader - you pull the level, and lots of things happen to reload the shell. But you can see everything that happens, because sooner or later things will get jammed and you'll have to reach in and fix it.
-Nathan Whitehead
Comments on Slashdot should be moderated by smell rather than number. Think a comment sucks? Make it stinky! Is it another karma whore? Make it smell like cheap perfume.
This is a niggly point, but a mistake that many people make. It is not legal to redistribute MP3's you download from mp3.com! The terms of use specifically prohibit you from redistributing the files.
There is nothing wrong with being proud of your ideas and advocating them. The problem is when he uses his cryptography newsletter to advocate his own pet algorithms. Nothing wrong with that, but maybe he should be more open and mention when he's talking about products he developed.
-Nathan Whitehead
-Nathan Whitehead
People bat around the word "conservative" quite a lot without having a clue what it means. Mandating the use of filtering software in public libraries is not conservative. True conservative towns don't keep computers in their libraries at all, let alone ones connected to the Internet. The true conservative solution is the one the town is already implementing. When some idiot looks at pornography in the library, the librarian walks by with a stern look.
Many prominent physicists seem to think that the revolutions of the next century will be in biology and not in physics. What do you think?
He says they are Acme Klein bottles, so I think they are the same as on www.kleinbottle.com (I couldn't get through to that link).
-Nathan Whitehead
The actual test took place last week, with Donald Knuth playing the role of the judge. After 10 grueling hours of questioning, Knuth declared that Fred Brooks was virtually indistinguishable from a computer program.
"It gives me great pleasure to accept this award on behalf of computer scientists everywhere. /* Fix up speech here, add some thank-you's and stuff. Re-check grammar. */," said Dr. Brooks on being presented with his prize. "I have always had a knack for fast arithmetic, and a lifetime of dealing with computers has made me doubt if I was human myself." Some members of the press commented that Dr. Brook's lack of social skills couldn't have hurt his chances either.
Fred Brooks is the first human to legitimately pass the Turing test. Last month a man known only as HemostheRobotMan claimed to have passed the Turing test, but was later disqualified after he was found using a Palm Pilot hidden in his shorts. Fred Brooks caused a stir when he said, "I need to change my batteries." It turns out he was merely hungry.
-Nathan Whitehead
Wavelets work in a similar fashion, using the wavelet transform. The difference is that wavelets are localised in space, unlike frequencies which are infinite sine waves. Thus when you forget the low coefficients and only save the highest wavelet coefficients, you can still retain things like discontinuities and local detail. The wavelets representing the local discontinuity have high coefficients and so are kept. In the frequency domain (used in Fourier transforms), discontinuities have components in all frequencies. When JPEG filters out the low coefficients, sharp discontinuities become blurred.
This is very cool to see wavelet compression start to move into the mainstream. One application already in use: the FBI uses wavelet compression to store fingerprint data (source: Wavelet Toolbox for Matlab documentation). Another common application of wavelets is removing noise from signals. Standard frequency filtering removes the noise and part of the signal. Wavelet de-noising can remove just the noise (it's really pretty amazing). I had an internship at the University of Washington where I used wavelets to remove RF interference from experimental physics data.
-Nathan Whitehead
If the network was designed correctly, it would be impossible for any one site to know where the traffic is coming from or going. Someone trying to break the anonymity through legal means would have to subpoena all the sites. Messages could be limited to 10K each, and only 1 message every 10 minutes per IP address (or something to prevent spam).
-Nathan Whitehead
-Nathan Whitehead
I think people have some misunderstanding of how analog->digital->analog reproduction happens. There are two kinds of distortion introduced in converting an analog signal into a digital signal and back again. The basic idea behind the conversion is sampling. A sample is taken every t seconds (where t might be 1/44100). Each sample is represented by b bits.
- Quantization error - this happens because b is not infinite. There is less quantization error in a 16-bit sample than in an 8-bit sample.
- Aliasing artifacts - these are signals introduced at the output that were not present at the input of the system. They arise because any frequency in the input higher than the Nyquist frequency will be reflected back into the 0-Nyquist range.
By increasing the number of bits used per sample, b, you help eliminate quantization errors. By increasing the sampling frequency (decreasing t) you eliminate aliasing effects.The Nyquist limit is half of the sampling frequency. To represent a frequency at the Nyquist frequency, it is necessary to have samples alternate like +1,-1,+1,-1,etc.
The previous poster wondered about phase information when you get close to the Nyquist limit. The answer is that if you have infinite precision samples (each sample is a real number), then all phase and frequency information below the Nyquist frequency is preserved. Suppose you have 10 sample points on a high-frequency sine wave (but that is still below the Nyquist frequency). Shifting the sine wave left and right slightly (altering the phase) makes the sample points go up and down. If you have accurate sample point resolution, you can exactly restore the phase of the signal (and its frequency).
If you try to sample a frequency that is higher than the Nyquist frequency, you will get sample points that look like another frequency between 0 frequency and the Nyquist frequency. There is no way to determine if the sample points represent the lower or the higher frequency. The solution is to filter out all frequencies higher than the Nyquist frequency at the input.
-Nathan Whitehead
-Nathan Whitehead
What exactly is the point of banning strong crypto products? I was under the impression that the reason crypto is "bad" is because "bad people" can use it for nefarious purposes. Aren't there any "bad people" in the US? If people in the US can use crypto for benign purposes, why can't people in other countries do the same?
It might make sense to restrict the sale of cryptographic products going to governments with a known track record of abusive behavior. Anything more than this is lunacy. It appears (from my reading of the article) that Netscape won't be able to ship a strong-crypto version of Navigator to people in Cuba. Where is the logic there? The Cuban government might get a hold of it and be able to order missiles from www.nukes.com and not have to worry about their credit card number being stolen?
-Nathan Whitehead
I agree with you that encryption must be really easy to use before it becomes standard. As mentioned, the way public keys work is that you encrypt something with someone's public key, and then only that person can decrypt it. So even if email software automatically grabbed keys, if clueless users did not have public keys there is no way they could receive encrypted email.
I think the real solution is to add encryption to the underlying mail protocols. This probably won't happen for a while. The idea would be that to have an email address, you must have a public key.
When clueless Joe User signs onto the Internet via his local ISP, he wants an email account. To get his email account, he must enter a passphrase into his computer. This becomes his private key (which he can remember and use anywhere). His computer then computes the corresponding public key, and sends it to his ISP. The ISP then creates his mail account, and sends his address plus public key to the nearest trusted keyserver.
Whenever email is sent, the appropriate keyserver is queried and the email is automatically encrypted. This way encryption is the default, not plain text. Of course this way there are problems with people forgetting their passphrase, etc. This would also rely on the keyserver's accuracy. So someone could put up an insincere keyserver that always responds, "Yes, I know the public key for that address, it is 0xdeadbeef" where 0xdeadbeef is the public key of the owner of the keyserver.
Widespread encryption is inevitable given the way people (and governments) abuse the situation now. I predict that within 3-5 years most computer-savvy people with use PGP or something equivalent for their normal correspondence. Within 8 years encryption will be the default, and it will be harder to send an email without encryption than with encryption.
-Nathan Whitehead
What I would like to see is an organization that has a contractual obligation to revoke its endorsement of any company found to violate the public's trust. Individual consumers should be able to donate money to the organization. In return they get the assurance that any site they visit with the TRUSTe logo really is following good privacy guidelines. If some company is found to violate the guidelines, then TRUSTe must revoke its endorsement or face lawsuits by the individuals. This way the companies are obligated to stay honest, and TRUSTe is obligated to stay honest or face lawsuits. -Nathan Whitehead
Here is a possible protocol for an open-source program to authenticate itself.
This protocol does not work, because if you have the source code for the hashing function and a copy of the "real" binary, you can spoof the authenticator. Is there a solution?
If there is a kernel or library function that hashes the binary code of the calling process, it is possible. The hashing function could be open-source, but the compiled function would have to include an embedded constant known only to the authenticator.
I don't know how feasible that is. The idea is if you hacked the source of the hash function and recompiled, you would lose the embedded constant. Of course if you hack the original program, when you call the hash function it will return the wrong value.
Just some random thoughts.
Nathan Whitehead
Put simply, TeX brought typesetting to the masses. Knuth invented TeX because he was tired of sending manuscripts with complicated equations to the typesetter and getting back garbage. TeX allows anyone with a little patience and willpower to typeset virtually anything they want. (I use LaTeX, an extension to TeX, so my comments might only be applicable to LaTeX).
Advantages of TeX
There are many advantages of (La)TeX over other document creation systems.
I think that TeX is most appropriate for creating documents that clearly and simply convey information. It is the ideal tool for writing papers to be published in academic journals. Everything about TeX is designed with communication in mind. The default margins are wider than most other word processors. This is because people have a hard time reading wide columns of text. Look at any good book and count the number of words across a column. It will be the same as the number of words in a default TeX column.
Disadvantages of TeX
- There is a learning curve. TeX is a deep system. Learning TeX is a never-ending journey. Initially it can be painful and unproductive. Many people get discouraged.
- Using TeX is not interactive. TeX by itself is a "compiler". Given a source file, it produces an output file suitable for printing. Writing TeX can often become more frustrating than programming. Silly syntactic errors can stop the compile from completing. It gets old reTeXing after every change in your file.
- It is often hard to make formatting changes. The formatting system is complex. For example, to change margins or line spacing is not as simple as grabbing a slider.
- It is sometimes hard to tell what a given section of source file produces. \delta is different than seeing the delta symbol.
There are several expensive commercial systems based on TeX that provide an easy-to-use interface. There is also LyX, which is GPL. I do not have much experience with these, so I don't know effective they are.-Nathan Whitehead
When I bought my latest computer last year, I naively decided to get a SB Live! since it seemed like such a cool card, and it was by Creative, so how could there not be Linux drivers for it?
When I discovered there were no drivers, and that Creative was not releasing the specifications, I was understandably disappointed. After the first beta drivers came out, I could at least get some sound out of the card. But some games didn't work with it, and it still had less functionality than the AWE32 driver. So my younger brother got a free upgrade from AWE32 to SB Live!, and I got a free downgrade that gave me more functionality. If the driver at that time had been open source, I honestly would have spent my whole summer hacking it to do cool stuff.
My point? This is a VERY SMART MOVE by Creative. There are a LOT of DSP hackers in college, just itching to write cool DSP code. Sound has been an often neglected field of hacking. We are on the verge of a Renaissance in sound. Under Linux, there is a good abstraction layer between the programmer and the soundcard. All the demo programmers, trackers, general code hackers, etc. are coming to Linux. The Linux crew has traditionally been made up of more 'mature' programmers. But with the rise in popularity of Linux, the migration is inevitable. And these people will bring new ideas of what is "cool" to program. Look for lots of cool sound programs in the next couple years.
There are reasons it is better to use the GPL for projects that aren't designed to be libraries. For example, suppose I develop a specialized mathematics program. If I release it under the LGPL, it is possible for Wolfram (for example) to rip out the interesting routines and make it into a library. They LGPL their changes to my program, then use that library in Mathematica without my consent or even mentioning that I wrote that code!
This might not be what I intended. If I release it as GPL software, they would have to release Mathematica as GPL software to do this. Wolfram would never do something this, but the possibility that some company might is something to consider. Of course, you might WANT anyone and everyone to take advantage of the code, so then you would release it under a LGPL or BSD license. I think it is entirely reasonable to have both the GPL and LGPL licenses.
Back in the "good old days" of cryptography, the algorithms used were understandable by non-mathematicians. Most modern cryptographic systems in use are still mathematically "simple". By this I mean that once you understand the complexities of the algorithm, the mathematical basis is understandable to someone who has, say, a college degree in mathematics or physics.
The cryptographic systems being developed today are often based on much more sophisticated mathematical ideas. Elliptic and hyper-elliptic curves spring to mind. The algorithms may be understandable, but the mathematical basis may be complicated enough that it takes a PhD in mathematics to understand.
These systems are the future generation of cryptography. Some have suggested that their security is based more on mathematical obscurity than anything else (i.e. the number of people able to even understand what the algorithm is doing is very small). Do you think this is accurate? Do you see cryptography moving exclusively into the domain of mathematicians, so that it is totally inaccessible to motivated non-mathematicians (such as yourself)?
My girlfriend broke up with me a couple weeks ago. This article was exactly what I needed to hear. It is guy-talk at its finest. Based on my own experience, this article matches with reality pretty closely. The biggest mistake 'geek' guys make is not catching social signals. They either over- or under-interpret things.
Thank you for the cool article RobLimo! Although maybe it should have included the disclaimer, "FOR GUYS ONLY!" If women only knew what guys REALLY thought...
Microsoft made its millions by being chosen as the operating system for IBM's PC. It made its billions by developing Windows and branching into application programming (Microsoft Office). Now it seems like Microsoft is trying to grab its piece of many different pies.
What is Microsoft's core business? If it is operating systems, why are Win2000 and NT losing respect so rapidly? If it's application development, why isn't Microsoft opening up Windows to be a standard for everybody?