Well, it's really more that the pipelines are very long. On the order of 600 pipelinestages, and that's pretty damned long. (P4 which is a CPU with a deep pipeline has 21 stages IIRC.)
They do of course store data between those stages, and there are caches on the chip. Otherwise performance would be shot all to hell.
I doubt that the original statement that GPU designs don't count the on chip memory is correct. That just seems like an odd way to do it.
And you are just as bad yourself by claiming that all readers of Slashdot have a fanatical hate towards Microsoft. Bashing of Slashdot is in my experience typical of people who belong to one of the other community sites and feel they must distinct themselves from Slashdot to feel better (or something).
Slashdot is probably the biggest community site in the tech area, and I would not be surprised if it is one of the top 5 biggest community sites in the world. When you claim that all those people have the same opinions you are just getting rediculous.
Now on the topic. The point with Mozilla is (to me) that it is a free and portable system. You can run it on basically any platform you can run a modern OS on. This means that if you use Mozilla as a target then you are ensured that everyone can use it. If you begin implementing platform specific extentions then you lose that portability. And thus also one of the biggest benefits with Mozilla.
As others have pointed out, if it\s a good thing to do (XAML) then perhaps a new project should try to reimplement it. Naturally there are a lot of areas were Microsoft have previously demonstrated hostility against such project. (SMB, NTFS and file formats are all good examples of Microsoft trying to hinder the development of OSS by changing specs.)
And personally I see a lot of Windows users who are convinced of their platforms superiority. Just as there are many for OSX, Linux, BSD and just about any OS you can mention. Drinking the Kool-Aid is pretty much mandatory.
Of course the patent system has use today, and it's quite good at it as well. What you hav to ask yourself is
1) What is it we want to accomplish with it? 2) Is it actually accomplishing that today?
The first question is typically that you want to aid/the little guy/ in all this. It should be possible for a small company or an individual to gain protection from the big boys when they invent something. So if I invent something and start selling it then Some Big Company can't just copy what I have done and use their much larger market presence and resources to take me off the market.
Furthermore it should make it a good idea for me to share ideas with other people since I don't have to worry about them taking the idea and ripping me in the process.
Is the current patent system doing this? No, not really. It's far to expensive to apply for patents and protecting them for an individual or small company to do so successfully. Since there are so many incorrectly issued patents (which you still have to challenge in court at great cost if used against you) you are almost guaranteed to be breaking a few patents yourself amd thus under threat of litigation. (Which you can't fight since you don't have the economic capacity for it, even if you are right.)
So what we have in the current patent system is pretty much something which does the opposite of what it originally intended. At least as far as the "little guy" is concerned.
And patents are NOT tools. At least not in any useable way I see it. If you could use patents to make new things that would be true. As it is today it's more or less just "insurance" aginst future threats.
I agree with you that the benefits outweighing the problems is doubtful in the case of software patents. In fact, unless you happen to be a very larg corporation I doubt there are any benefits at all. (At least in Europe all small-medium developers have been extremely negative in the issue. Large corps are basically the only ones in favour.)
What whould have annoyed me with that example is that they didn't have the API description locally available. The entire point of having an API is to allow your coders to ignore irrelevant details like what type a certain function takes and such.
OTOH I think most test like these tend to fail their mark. Unless you get "out of the box" behaviour like you described and the interviewer recognices it. Unless you are hiring code monkeys there seems to be little point in having trivial coding assignments (input file, sort output file). It seems like a more complex problem without coding but instead reasoning is more relevant.
FLAC is NOT a lossy format, that's the entire point of it. That the source itself is not 100% of what was played in a studio is irrelevant.
Perhaps people like having an entirely open format? Perhaps people like not being tied down with patents? Perhaps people want to make SURE that they can still play the files 10-20 years down the road and not just cross their fingers and hope.
And a lot of the ripped MP3 are poor quality. When you start playing on high quality equipment it begins to be more apparent and since you can get higher quality today than say 5 years ago why would you want to put up with the old low quality stuff?
Let me put it like this, if you are completely satisfied with your old MP3 and don't plan on re-ripping fine. If you do take the time to re-rip why not do it to FLAC and save those as well? Then transcode to whatever quality and format you need right now. That way if your new MP3 player handles higher bitrates (my cheap one is best at 128k) you can reencode it as a batch job during the night.
Central point is this: space is cheap, even with FLAC storing my entire CD collection on disk is not particularly expensive. The time I have to spend to swap discs and ensure correct tagging and such is not cheap however. I have better things to do with my life than watch the progressbar.
Actually there are a lot of sites with WAP support. Google is just one. And while it's not the best in the world it's still kind of fun to do a Google search from your mobile phone. I've also used it sometimes to browse The Onion while bored. Those articles are pretty long and tend to be funny so it's a good thing to do on the phone.
Today using WAP on my T630 is no harder than pushing "Online" and then "more" -> "bookmarks" or "address". It's even pretty easy to type wap since it's the first letters if you use the multi-tap input function.
WAP2.0 has support for images and such too. It's just too bad that not all sites have jumped on it. I'd love to be able to use IMDB while at the video store / cinema or Amazon while at the bookstore.
I think that when there comes a format which stores even more information than CD which becomes established (perhaps DVD Audio?) then you can take that time. If you can actually get hold of new versions of the old music which is actually better than what you have on CD. (Not sure how high the quality is on the masters they use now.)
My point was that if you ripped a huge collection of CDs in the beginning of MP3s then you'd have to redo all that work in order to turn it into higher quality MP3 or Ogg (or any future format). If you instead rip it to FLAC you can just transcode it. Transcoding between lossy formats is just stupid.
After reading this it seems like it's a pretty good idea. Providing an add-on which can interoperate with existing software and hardware is always nice. However they claim that the next step will be to apply this to other areas such as Bluetooth and even mobile phones.
While changing the size of the packets is a good way to improve performance on systems with high level of abstractions (like WiFi) I doubt it can be successfully applied to systems such as mobile phones. In fact I would be surprised if this could even remotely compete with techniques such as turbo codes (or FEC) which are routinely used for such systems.
The described system is quite naive and uses a very simplistic way of controlling the system. Error codes instead model the channel and then add redundant information so that even if a packet is corrupted it can be corrected by the receiver. It has been shown that such codes can come close to the Shannon limit (maximum limit of information transfer) for a given channel.
Naturally construction such codes is quite a lot harder and it requires that both sides use it. This method is good since it's compatible with existing systems and even if not all participants use it it gives some benefits.
The problem to solve is that depending on the amount of traffic and noise the optimal packet length to send varies. If you send a long packet and fail you have to resend all that data, even the data which you managed to get through.
If instead you send two packets then only the packet with the error would have to be resent. However smaller packets mean larger overhead from the protocols used. This overhead to packet length ratio is optimised depending on the level of noise on the channel.
I'd recommend that you read the second article, from EE-times. It actually has some content which is something their own site is quite completely void of.
And people have been doing this before. The EE-times article mentions that. Apparently no-one has either not made so much progress or just not made so much of a fuss over it before. A quick search for "variable packet length and wireless" turns up quite a lot of results though. I'm fairly confident that you can find previous research in this area if you look around.
I'm not entirely convinced that Kalman filters would do a good job though. Or rather, they may need more additions in order to be efficient in this particular problem space. Thus making it bigger and less desireable to put into eg firmware. It's not unreasonable that they haven't considered it though. But they did seem to try some other systems, some of them seem a bit too optimistic to be reasonable in this problem space though. According to EE-times they did attempts with database lookup and expert systems eg. Seems like both of those could be "trivially" rejected to not be adaptive and accurate enough.
It's not impossible that there exist methods which are a lot more efficient than NN to solve this problem. NN seems to do a good job though, so I guess it warrants some more looking into.
I think the rule of thumb (har har) there is that you want to make sure that the "something you have/are" thing is something you can "throw to the ground and run".
So if you are going to spend a lot of money using biometrics you need to have physical guards there as well. (Naturally they can also work to help people.) These guards would make the systems better by stopping the biggest attacks on these system. To fool fingerprint detectors or to hold up an image in front of your face at the iris detector.
To me it seems like your initial instruction was about changing the size of the page file. That's not the same thing as changing the swappiness.
And personally I'd actually have to say that I find it easier to browse the/proc filesystem (since it's a FS you can use search and such just like normal) and then use "man" to find out what the things do. I bet you can find a lot of good guides too. But I don't think "tweak" is a good term to use when searching for Linux tips. It seems to be a term more frequently used with Windows. Finding a programs settings file is typically as easy as reading the man file (file references at the end) or doing "locate" for the program name.
That said, I think that basic tuning (ie swap size) is easier on Windows. But I think that more complex stuff (going through/proc or registry) is a bit easier on Linux. The reason being that on Linux it's actually designed so that you can use/proc and similar. (It has man pages to back it up.) On Windows you are pretty much forced to rely on searching online for hints of what the parameters do.
That said I don't really like the registry so I'm certainly not unbiased on the issue.
If this ever becomes interesting for the development I think a good way of testing it would be to work with a university. Since the goal is to produce open software in the end this would make it easier for the university to comply as well. (Ie it doesn't conflict with their interests of providing the results to the public in the end.)
Even if no real product come out of it you're bound to get feedback on some gotchas with the system. And hey, students are cheap labour.;-)
At least in Sweden McDonalds did a "comparison" where they compared the food in McD with that of real lunch restaurants. The conclusion was that the food at McD had less fat than a meal at a normal restaurant, thus making McD a healthy alternative.
If that's not lying it's at least grossly misinformational. Food at McD contain a lot more sugar (to compensate low fat at poor taste) which has been shown to be at least as big a problem to your health as fat. Furthermore a meal at McD will make you last an hour or so before you become hungry (because there's no nutrion in it) while I can eat a meal they compared to and feel full for the rest of the day.
So sure, companies do usually try not to lie outright. But advertising and marketing (advertising over time) sure isn't out to be honest with you. It's propaganda, so don't trust it.
BTW I think most animosity towards marketers is due to inhouse. Everyone in engineering know people who have had to work their asses off because a marketer has told a client that "Of course we have that feature!" when they in fact didn't. So I think that's part of the problem, people know that their own marketers don't tell you the whole story, so why should any other company do so?
I recommend that you play Max Payne and/or Prince of Persia instead. Both have a similar "time dilation" mode and both are vastly more playable than EtM.
The idea with EtM was pretty good, but it was extremely poorly made. The models looked like Max Payne 1 era, and that game is pretty old. The animations were so bad it was funny. Watch a guy "climb" a fence and you'll see what I mean.
If they had developed it for a few more months it could have been a good game though.
Well sure you can "reduce it" like that. Not sure what good it will do however. Personally I like 1/sqrt(3) more, less numbers in there to throw you off.;-) But it's all the same result in the end, so how you type it is a matter of formality.
When I was taking HS classes in math I had a pretty easy time. I used the calculator (my trusty TI-82) a lot. Then I started taking classes of math at university level and I quite quickly was clued in that I was pretty bad at math in reality. Furthermore since I no longer could use my calculator I had to think about the math for myself. It was not until that time I actually began understanding what I was doing.
If I had spent half the time in HS doing that I would have understood the math in HS better as well as had a better start in college. My point isn't that people who can't manage math in HS are stupid. My point is that the way math is tought in HS is stupid.
And while I'm sure that my memories of HS math (and similar subjects) have changed with time I have to say that it felt like 1 month of math at university was about the same as 1 year of math in HS. Naturally that pace is set as most people at university are inclined towards math, but it does put the things you learn in HS in perspective.
Oh I'm sure you can do it, but I think there is a rather big gap between the "small paper" which can use LaTeX/TeX since they have a really simple layout and the really big paper which has the kind of resources to make a fully automated system.
My experience from doing Highschool papers / yearbook designs was that there was a pretty tight feedback from layout to writer. So every now and then a few words were added or removed to make columns line up better and stuff like that. Also the individual page designs were pretty much freely made by hand. (Well we used Pagemaker, but it wasn't automatic.)
If you have a real newspaper I bet they use a lot of automated tools instead to do all that. Just tell the program the general layout of the paper and let it schedule the articles to best fill the space. Such a system could use LaTeX I bet, but it wouldn't be trivial to contruct. (It wouldn't be trivial in any language.)
Crap, I forgot the one biggest benefit I have encountered with LaTeX compared to eg Word. It doesn't completely fuck up references. Whenever I use references in a Word document I know that sooner or later it's going to get corrupt and I'll have to add them all again. It just seems like it can't properly handle that you move figures around and such.
Well put. There are some gotchas which can make your layout funky (empty newlines in a big math contex eg) but those are few and far between. If you are trying to make LaTeX look like something it's not then most likely you are trying to make it look "wrong". That's my experience at any rate.
And since it's a text document you can do a load of magic with it, just as you mention. Take a large document and break it into pieces and you can work on it concurrently. (Try that in any Word version or clone.) You can even put in in a version control system.
If you are doing things like API descriptions you can let scripts extract information from your source code and have an instantly updated API document each time you compile the project.
Really, LaTeX is a real killer when you are just "getting the work done". If you are doing layout for a paper it may not help you much. That's note really what it's for though.
Slashdot Mirror
Well, it's really more that the pipelines are very long. On the order of 600 pipelinestages, and that's pretty damned long. (P4 which is a CPU with a deep pipeline has 21 stages IIRC.)
They do of course store data between those stages, and there are caches on the chip. Otherwise performance would be shot all to hell.
I doubt that the original statement that GPU designs don't count the on chip memory is correct. That just seems like an odd way to do it.
Look at gpgpu.org I believe they have papers on doing FFT on GPUs. They also have a collection on papers regarding GPU as CPUs.
And you are just as bad yourself by claiming that all readers of Slashdot have a fanatical hate towards Microsoft. Bashing of Slashdot is in my experience typical of people who belong to one of the other community sites and feel they must distinct themselves from Slashdot to feel better (or something).
Slashdot is probably the biggest community site in the tech area, and I would not be surprised if it is one of the top 5 biggest community sites in the world. When you claim that all those people have the same opinions you are just getting rediculous.
Now on the topic. The point with Mozilla is (to me) that it is a free and portable system. You can run it on basically any platform you can run a modern OS on. This means that if you use Mozilla as a target then you are ensured that everyone can use it. If you begin implementing platform specific extentions then you lose that portability. And thus also one of the biggest benefits with Mozilla.
As others have pointed out, if it\s a good thing to do (XAML) then perhaps a new project should try to reimplement it. Naturally there are a lot of areas were Microsoft have previously demonstrated hostility against such project. (SMB, NTFS and file formats are all good examples of Microsoft trying to hinder the development of OSS by changing specs.)
And personally I see a lot of Windows users who are convinced of their platforms superiority. Just as there are many for OSX, Linux, BSD and just about any OS you can mention. Drinking the Kool-Aid is pretty much mandatory.
Of course the patent system has use today, and it's quite good at it as well. What you hav to ask yourself is
/the little guy/ in all this. It should be possible for a small company or an individual to gain protection from the big boys when they invent something. So if I invent something and start selling it then Some Big Company can't just copy what I have done and use their much larger market presence and resources to take me off the market.
1) What is it we want to accomplish with it?
2) Is it actually accomplishing that today?
The first question is typically that you want to aid
Furthermore it should make it a good idea for me to share ideas with other people since I don't have to worry about them taking the idea and ripping me in the process.
Is the current patent system doing this? No, not really. It's far to expensive to apply for patents and protecting them for an individual or small company to do so successfully. Since there are so many incorrectly issued patents (which you still have to challenge in court at great cost if used against you) you are almost guaranteed to be breaking a few patents yourself amd thus under threat of litigation. (Which you can't fight since you don't have the economic capacity for it, even if you are right.)
So what we have in the current patent system is pretty much something which does the opposite of what it originally intended. At least as far as the "little guy" is concerned.
And patents are NOT tools. At least not in any useable way I see it. If you could use patents to make new things that would be true. As it is today it's more or less just "insurance" aginst future threats.
I agree with you that the benefits outweighing the problems is doubtful in the case of software patents. In fact, unless you happen to be a very larg corporation I doubt there are any benefits at all. (At least in Europe all small-medium developers have been extremely negative in the issue. Large corps are basically the only ones in favour.)
What whould have annoyed me with that example is that they didn't have the API description locally available. The entire point of having an API is to allow your coders to ignore irrelevant details like what type a certain function takes and such.
OTOH I think most test like these tend to fail their mark. Unless you get "out of the box" behaviour like you described and the interviewer recognices it. Unless you are hiring code monkeys there seems to be little point in having trivial coding assignments (input file, sort output file). It seems like a more complex problem without coding but instead reasoning is more relevant.
Quite a lot harder to set up and evaluate though.
Adding finnish to the list of languages I'm not going to learn. ;-)
FLAC is NOT a lossy format, that's the entire point of it. That the source itself is not 100% of what was played in a studio is irrelevant.
Perhaps people like having an entirely open format? Perhaps people like not being tied down with patents? Perhaps people want to make SURE that they can still play the files 10-20 years down the road and not just cross their fingers and hope.
And a lot of the ripped MP3 are poor quality. When you start playing on high quality equipment it begins to be more apparent and since you can get higher quality today than say 5 years ago why would you want to put up with the old low quality stuff?
Let me put it like this, if you are completely satisfied with your old MP3 and don't plan on re-ripping fine. If you do take the time to re-rip why not do it to FLAC and save those as well? Then transcode to whatever quality and format you need right now. That way if your new MP3 player handles higher bitrates (my cheap one is best at 128k) you can reencode it as a batch job during the night.
Central point is this: space is cheap, even with FLAC storing my entire CD collection on disk is not particularly expensive. The time I have to spend to swap discs and ensure correct tagging and such is not cheap however. I have better things to do with my life than watch the progressbar.
Actually there are a lot of sites with WAP support. Google is just one. And while it's not the best in the world it's still kind of fun to do a Google search from your mobile phone. I've also used it sometimes to browse The Onion while bored. Those articles are pretty long and tend to be funny so it's a good thing to do on the phone.
Today using WAP on my T630 is no harder than pushing "Online" and then "more" -> "bookmarks" or "address". It's even pretty easy to type wap since it's the first letters if you use the multi-tap input function.
WAP2.0 has support for images and such too. It's just too bad that not all sites have jumped on it. I'd love to be able to use IMDB while at the video store / cinema or Amazon while at the bookstore.
I think that when there comes a format which stores even more information than CD which becomes established (perhaps DVD Audio?) then you can take that time. If you can actually get hold of new versions of the old music which is actually better than what you have on CD. (Not sure how high the quality is on the masters they use now.)
My point was that if you ripped a huge collection of CDs in the beginning of MP3s then you'd have to redo all that work in order to turn it into higher quality MP3 or Ogg (or any future format). If you instead rip it to FLAC you can just transcode it. Transcoding between lossy formats is just stupid.
How about never having to re-rip an album again because a better format was released?
I mean, when we have jpeg, why would we need a lossless format like PNG?
After reading this it seems like it's a pretty good idea. Providing an add-on which can interoperate with existing software and hardware is always nice. However they claim that the next step will be to apply this to other areas such as Bluetooth and even mobile phones.
While changing the size of the packets is a good way to improve performance on systems with high level of abstractions (like WiFi) I doubt it can be successfully applied to systems such as mobile phones. In fact I would be surprised if this could even remotely compete with techniques such as turbo codes (or FEC) which are routinely used for such systems.
The described system is quite naive and uses a very simplistic way of controlling the system. Error codes instead model the channel and then add redundant information so that even if a packet is corrupted it can be corrected by the receiver. It has been shown that such codes can come close to the Shannon limit (maximum limit of information transfer) for a given channel.
Naturally construction such codes is quite a lot harder and it requires that both sides use it. This method is good since it's compatible with existing systems and even if not all participants use it it gives some benefits.
No, that's not what this is about.
The problem to solve is that depending on the amount of traffic and noise the optimal packet length to send varies. If you send a long packet and fail you have to resend all that data, even the data which you managed to get through.
If instead you send two packets then only the packet with the error would have to be resent. However smaller packets mean larger overhead from the protocols used. This overhead to packet length ratio is optimised depending on the level of noise on the channel.
I'd recommend that you read the second article, from EE-times. It actually has some content which is something their own site is quite completely void of.
And people have been doing this before. The EE-times article mentions that. Apparently no-one has either not made so much progress or just not made so much of a fuss over it before. A quick search for "variable packet length and wireless" turns up quite a lot of results though. I'm fairly confident that you can find previous research in this area if you look around.
I'm not entirely convinced that Kalman filters would do a good job though. Or rather, they may need more additions in order to be efficient in this particular problem space. Thus making it bigger and less desireable to put into eg firmware. It's not unreasonable that they haven't considered it though. But they did seem to try some other systems, some of them seem a bit too optimistic to be reasonable in this problem space though. According to EE-times they did attempts with database lookup and expert systems eg. Seems like both of those could be "trivially" rejected to not be adaptive and accurate enough.
It's not impossible that there exist methods which are a lot more efficient than NN to solve this problem. NN seems to do a good job though, so I guess it warrants some more looking into.
I think the rule of thumb (har har) there is that you want to make sure that the "something you have/are" thing is something you can "throw to the ground and run".
So if you are going to spend a lot of money using biometrics you need to have physical guards there as well. (Naturally they can also work to help people.) These guards would make the systems better by stopping the biggest attacks on these system. To fool fingerprint detectors or to hold up an image in front of your face at the iris detector.
Knoppix does it, for obvious reasons. I quite like it on machines with lots of memory. (I have one with 1GB, and I've never filled that.)
To me it seems like your initial instruction was about changing the size of the page file. That's not the same thing as changing the swappiness.
/proc filesystem (since it's a FS you can use search and such just like normal) and then use "man" to find out what the things do. I bet you can find a lot of good guides too. But I don't think "tweak" is a good term to use when searching for Linux tips. It seems to be a term more frequently used with Windows. Finding a programs settings file is typically as easy as reading the man file (file references at the end) or doing "locate" for the program name.
/proc or registry) is a bit easier on Linux. The reason being that on Linux it's actually designed so that you can use /proc and similar. (It has man pages to back it up.) On Windows you are pretty much forced to rely on searching online for hints of what the parameters do.
And personally I'd actually have to say that I find it easier to browse the
That said, I think that basic tuning (ie swap size) is easier on Windows. But I think that more complex stuff (going through
That said I don't really like the registry so I'm certainly not unbiased on the issue.
If this ever becomes interesting for the development I think a good way of testing it would be to work with a university. Since the goal is to produce open software in the end this would make it easier for the university to comply as well. (Ie it doesn't conflict with their interests of providing the results to the public in the end.)
;-)
Even if no real product come out of it you're bound to get feedback on some gotchas with the system. And hey, students are cheap labour.
At least in Sweden McDonalds did a "comparison" where they compared the food in McD with that of real lunch restaurants. The conclusion was that the food at McD had less fat than a meal at a normal restaurant, thus making McD a healthy alternative.
If that's not lying it's at least grossly misinformational. Food at McD contain a lot more sugar (to compensate low fat at poor taste) which has been shown to be at least as big a problem to your health as fat. Furthermore a meal at McD will make you last an hour or so before you become hungry (because there's no nutrion in it) while I can eat a meal they compared to and feel full for the rest of the day.
So sure, companies do usually try not to lie outright. But advertising and marketing (advertising over time) sure isn't out to be honest with you. It's propaganda, so don't trust it.
BTW I think most animosity towards marketers is due to inhouse. Everyone in engineering know people who have had to work their asses off because a marketer has told a client that "Of course we have that feature!" when they in fact didn't. So I think that's part of the problem, people know that their own marketers don't tell you the whole story, so why should any other company do so?
Like the Canon EOS 1Ds or the EOS 1D mark 2? You are right that they are not consumer or prosumer prices though.
You were lucky, we used to dream about having a pinhole. We had to make do with square.
I recommend that you play Max Payne and/or Prince of Persia instead. Both have a similar "time dilation" mode and both are vastly more playable than EtM.
The idea with EtM was pretty good, but it was extremely poorly made. The models looked like Max Payne 1 era, and that game is pretty old. The animations were so bad it was funny. Watch a guy "climb" a fence and you'll see what I mean.
If they had developed it for a few more months it could have been a good game though.
Well sure you can "reduce it" like that. Not sure what good it will do however. Personally I like 1/sqrt(3) more, less numbers in there to throw you off. ;-) But it's all the same result in the end, so how you type it is a matter of formality.
When I was taking HS classes in math I had a pretty easy time. I used the calculator (my trusty TI-82) a lot. Then I started taking classes of math at university level and I quite quickly was clued in that I was pretty bad at math in reality. Furthermore since I no longer could use my calculator I had to think about the math for myself. It was not until that time I actually began understanding what I was doing.
If I had spent half the time in HS doing that I would have understood the math in HS better as well as had a better start in college. My point isn't that people who can't manage math in HS are stupid. My point is that the way math is tought in HS is stupid.
And while I'm sure that my memories of HS math (and similar subjects) have changed with time I have to say that it felt like 1 month of math at university was about the same as 1 year of math in HS. Naturally that pace is set as most people at university are inclined towards math, but it does put the things you learn in HS in perspective.
Oh I'm sure you can do it, but I think there is a rather big gap between the "small paper" which can use LaTeX/TeX since they have a really simple layout and the really big paper which has the kind of resources to make a fully automated system.
My experience from doing Highschool papers / yearbook designs was that there was a pretty tight feedback from layout to writer. So every now and then a few words were added or removed to make columns line up better and stuff like that. Also the individual page designs were pretty much freely made by hand. (Well we used Pagemaker, but it wasn't automatic.)
If you have a real newspaper I bet they use a lot of automated tools instead to do all that. Just tell the program the general layout of the paper and let it schedule the articles to best fill the space. Such a system could use LaTeX I bet, but it wouldn't be trivial to contruct. (It wouldn't be trivial in any language.)
Crap, I forgot the one biggest benefit I have encountered with LaTeX compared to eg Word. It doesn't completely fuck up references. Whenever I use references in a Word document I know that sooner or later it's going to get corrupt and I'll have to add them all again. It just seems like it can't properly handle that you move figures around and such.
Well put. There are some gotchas which can make your layout funky (empty newlines in a big math contex eg) but those are few and far between. If you are trying to make LaTeX look like something it's not then most likely you are trying to make it look "wrong". That's my experience at any rate.
And since it's a text document you can do a load of magic with it, just as you mention. Take a large document and break it into pieces and you can work on it concurrently. (Try that in any Word version or clone.) You can even put in in a version control system.
If you are doing things like API descriptions you can let scripts extract information from your source code and have an instantly updated API document each time you compile the project.
Really, LaTeX is a real killer when you are just "getting the work done". If you are doing layout for a paper it may not help you much. That's note really what it's for though.