This means you need one person who groks the entire project but you can have many people writing the code. See Feynman's critique of the design process used in for the Space Shuttle.
Back in WW-II, they added tube based analog computers to the anti-aircraft guns on some ships to help direct the shells/bullets to whatever was in the targeting sight.
IIRC, there were two problems with these devices. First, the tubes took too long to warm up because there was often very little advanced warning of an aerial attack. Second, there was a bug that got the sign of the Coriolis force wrong in the Southern Hemisphere.
The sailors ended up using fire axes to disable the automatic controls and return their guns to manual mode.
I heard this story from a fellow engineer almost 30 years ago and he was close to retirement age when he told it. He was also an artist and did the illustrations for the first edition of Ralph Nader's book, "Unsafe at any Speed".
My point was not "rational is good and non-rational is bad". I used the well documented fact that people respond non-rationally when faced with certain threats to suggest that another terror attack in the US would probably help the Republicans in November even though it happened "on their watch".
If anyone was questioning what I said in my original post, several of the responses, including yours, should erase all possible doubt. Thank you for your help by providing a specific example of the behavior I was discussing.
The best recent example (the book was written in 1989) is America's reaction to the 9/11 attacks. More people died of hunger that day than were killed in the attack. In the US? I'm calling BS on that one unless you have some stats that prove otherwise. The most commonly used number is 24,000 world-wide... No, of course not just in US. I didn't say more people died of hunger that day in the US. I just said more peopled died of hunger that day.
Your assumption and your misunderstanding of seven simple words combined with the visceral attacks in the rest of your response are an excellent example of the point I was trying to make. Thank you for your help.
Unfortunately, I think there are going to be powerful dark forces at work to try get the Republicans back in again.
People are easily swayed. Another terrorist attack in the USA I think could sway the elections. That after 8 years, Republicans can't protect America? You need to read
New World, New Mind by Robert E. Ornstein and Paul Ehrlich. Pdf's available
here.
The book explains that people are not rational or logical especially when it comes to risk assessment. The best recent example (the book was written in 1989) is America's reaction to the 9/11 attacks. More people died of hunger that day than were killed in the attack. The US response to the attacks was totally illogical because people felt threatened and this caused them to stop using the higher levels of their brains. They instead, reverted to their reptilian "flight or fight" instincts.
Another similar (or worse) attack will most likely produce a similar response from the American people. They will stop thinking rationally, which is probably the only way the Republicans can beat Obama on November 4th.
If you read all the comments following
this rant, you will discover that the person who created the offending patch tried to check it with the OpenSSL devs by posting the patch to the openssl-dev mailing list.
Unfortunately that list is not for OpenSSL devs, instead it is for users of OpenSSL. Therefore only other clueless users saw the patch. To reach the OpenSSL devs one needed to use the openssl-team mailing list instead of openssl-dev.
IMO, this problem was due to a communication problem and it is hard to blame just one person for that. If I had to place blame, I would say the fault was with the poorly chosen names for the OpenSSL mailing lists.
This phenomenon is explained in E. T. Jaynes' book
Probability Theory: the Logic of Science. In it he develops the concept of an ideal plausible reasoner who can use inference. He shows that given a small set of desired traits all such reasoner with those traits conform to the same mathematical model which turns to be pretty much the same thing as
Baysian probability theory.
Jaynes goes on to show that when two such ideal reasoners have differing views (due to their different past experiences) and are confronted with the same piece of new evidence, each one can believe the new evidence further reinforces their own view, thus polarizing the two ideal reasoners even further. This is a fundamental yet little known problem facing all human societies.
I think this is exactly what is going on in the fascinating study you brought up. It is nothing new. Even if humans were ideal reasoners (in the sense defined by Jaynes) we would still encounter these problems. Instead of blaming the audience for what is essentially a mathematical reality, it would be much more useful if we educated people about the underlying mathematics that causes this behavior.
I'm not familiar with the video game you are referring to. It looks like it is a lot of fun. Can you tell me where I can get a bootleg copy so I can try it out?
Obama didn't help by giving the impression that the Reverend was accurately presented in the media.
I disagree. I thought Obama made it clear that the clips shown over (and over again) in the media were not typical of Wright's sermons. But please also realize the reason why the media choose to run these particular clips over and over.
Wright's most radical comments are laced with anger and hatred so they inflame anger and hatred in many who hear them. This is why Gandhi, the Dalai Lama, and so many others teach that hatred and anger are not part of the path that will lead us to peace. The reason Obama needed to disavow Wright's words of
hate is because they represent the wrong path. The problem is not the
realities Wright exposed but the hate and anger he used in exposing
them.
Obama's great strength comes from what I would describe as the spiritual
balance of his campaign. He has eschewed the politics of hate and
division. This gives him such power that the only way his opponents
can stop him is to falsely attach to him a message of anger and hate.
This is not Obama's message and this is why it was essential for Obama
to disavow Wright's words of hate. The message Wright was giving in
those sermons was exactly the opposite of Obama's message. This is why Wright's words were used to attack Obama and this is why Obama needed to make it clear that Wright's message of anger is not his message.
The TiVo demonstrates that open source DRM is possible. As many, many others have pointed out, if the end-user has control of the code then they can easily circumvent DRM measures. The TiVo closed this loophole by using DRM/encryption technology to control the code that runs on the user's device, making it impossible for users to run modified software that is not approved by a central authority. Yet the source code is open for all to see.
It is much more difficult (impossible) to use Free (as in freedom) software to implement DRM. Free software ensures the end-user has control over the code running on their own machine and then all of the oxymoron arguments made above apply. All free software is open source but not all open source software is free. When discussing things such as implementating DRM, it is a mistake to assume "open source" and "free" are synonymous. They're not.
OSCE found the US elections to have only some minor problems,...
The 2004 presidential election was decided by a few key battleground states, most notably Ohio. Oddly enough there were strange
exit poll discrepancies in many of these states including Ohio where the outcome in hinged on less than 20,000 votes. Due to a host of peculiarities, a recount was ordered in Cuyahoga County. Last year the two people who performed that recount, Jacqueline Maiden and Kathleen Dreamer were
convicted of negligent misconduct for
rigging the recount:
They worked behind closed doors for three days to pick ballots they knew would not cause discrepancies when checked by hand, prosecutors said.
They were recently sentenced to 18 months in prison. The judge gave them the maximum because he did not believe their story that they were acting alone.
Let's recap:
Many states (such as NH and Ohio) still count votes
using machines with secret sauce source code that have been proved to be trivial to crack, making it easy for a single person to alter the outcome of an entire election.
The media via a private company have conspired to keep the raw exit poll data secret (see first link above) so it can't be used to check the official results.
A recount was ordered in one of the states that could possibly change the overall winner of the entire election but that recount was rigged and the ballots were destroyed so we have no idea of who actually won.
This is not proof the election was rigged, if the votes had been honestly recounted they may have matched the official results. But why on earth would the two official in charge of the recount go to the trouble and risk of rigging it if they thought the election was honest? Unfortunately we'll never know if it was honest or not. Never knowing if the outcome of an election was actually fair seems to be more than just a minor problem.
We are mostly in agreement but I should warn you that I'm a big fan of soft clipping and have used it in devices spanning from ultrasonic imaging equipment to research fish detectors.
I don't agree with your point (2):
2. Any extra harmonics it introduces are _before_ the final amplifier stage that actually drives the speakers. So they'll be amplified too. And any clipping that that final stage has, will come on top of it.
As I may have said before, you usually want the clipping stage to be early in the processing chain. I would agree with you that the best solution is to have the soft clipping as part of the power amp so it can be calibrated to eliminate any hard clipping in the final power stage. Perhaps I'm old fashioned but I prefer to use a linear power amp (with hard clipping) and put a line level soft clipper in front of it even if it is built into the
power amp.
You want the extra harmonics introduced by the soft clipper to be amplified by the power amp. Perhaps it is easiest to visualize this in the time domain. The clipper gracefully shaves off the high peaks and low valleys leaving a signal that is still fairly smooth but of limited range. It's the extra harmonics that do the rounding off. You want the power amp to make an exact replica of this signal except with a larger overall amplitude. That means you need to send the extra harmonics to the power amp. That's the whole point of the clipper.
Also, if the soft clipper is calibrated correctly then the power amp will never hard clip.
the chief drawback in having the soft clipper in what is essentially the preamp is that the soft clipping needs to be tuned to the specs of the power amp. For example, I think an extra volume control after the clipper may suffice. If the board with the tube on the end doesn't have a control to adjust the clipping then I would agree that the clipper won't be useful for all power amps. But there are probably some power amps that it works really well with.
My main point is that soft clipping is a very valuable tool to have in your bag of tricks. Also, IMO I think it is possible that the tube you described could actually do some good, especially if there is a way to tune the clipping to match the off-board power stage.
It says that some women who voted for Clinton might have been lying about it to avoid conflicts with their husbands. I hadn't consider this possibility but it makes sense to me and therefore I am less inclined to believe there was a problem with the official tally, although I'm still glad they're going to do a recount.
- audiophile motherboards with one vacuum tube at the end of an otherwise 100% digital chain, and again people swearing that their MP3's sound closer to the original with that (never mind that it's really just adding the tube's own soft-clipping kind and harmonics, to those that the digital chain already introduced),
Hard clipping is a problem with audio equipment especially in the final power stage driving the speakers. AFAIK, there are really only two solutions to this problem: 1) have enough extra headroom and a limited maximum volume so that clipping can never occur or, 2) add soft clipping, which is not as good as the headroom solution but is usually much less expensive. In fact, since many people want volume controls to go up to "11", most (but not all) audio systems would (or do) benefit from properly implemented soft clipping.
Good audio engineers follow some of the same basic principles as good software engineers and good web site designers including:
If you want to know how the law actually works, please simply read the EFF brief. They cite cases where the intent to infringe did not amount to a copyright violation. I don't think it can get any clearer than that. Fair use had nothing to do with it.
If you actual read the fine brief by the EFF you will see that according to the law, intent does not matter. If intent was what mattered then the RIAA would certainly prevail since intentionally
"making available" implies intent. The EFF cite the letter of the law and then give many case law examples that make clear that an actual transfer to a 3rd party must occur before it is a copyright violation. They cite cases where unauthorized people advertised copyrighted works and the defendants prevailed because there was no proof of an actual transfer.
They do mention one case where some (dead tree) libraries were successfully sued for copyright infringement without any proof of an actual transfer but that case seemed to be exceptional.
There are many unknowns in this case and in copyright law in general but one thing we do know is: intent does NOT matter in the eyes of the law.
I've been using Gentoo for years and I love it. In the great FOSS tradition, Gentoo is all about choice (read: control). In the early years the stable branch was not very stable and the forums were filled with emerge problems in the stable branch. In the last few years those problems have almost all gone away. My system is mostly from the stable branch and I almost never run into a problem when I do updates. This is a very good thing[TM] IMO. I'm very glad the devs keep packages out of stable until they are almost always problem free.
But with Gentoo I can keyword-unmask specific packages if I want to be more on the bleeding edge and if I am really adventurous, I can hard-unmask packages. Yes, running a mixed system can require me to unmask dependencies but that has never been a nightmare (for me) and I don't see how the Gentoo devs could possibly prevent it. In fact, it should give you insight into why some packages seem to take longer than usual to make it into stable: all their dependencies need to go stable also.
It could well be that Gentoo is not the right distro for you. If your top priority is ease of use then one of the binary distros would probably be a better choice. But if you want the maximum control over your system, Gentoo is hard to beat.
As I explained at length to Dire Bonobo, the MSM "corrects" the exit poll results so they always add up the the official results. If you
Google(exit poll discrepancy) you will get a bunch of links all about
one paper (warning: pdf) which says:
Part of the reason the issue [exit poll discrepancies] went away for the media and simultaneously raised suspicion on the web is secrecy and confusion about the data and what exactly is being characterized as the exit poll. If you go to the CNN website or any other website on which 2004 exit poll data are available, you'll see numbers very different from those released on election day. This is because the survey results originally collected and presented to subscribers were subsequently "corrected" to conform to official tallies.
Obama was up by 4 or 5% in the exit polls before they were "corrected" to match the official tally.
IMO the MSM is responsible for confusing the heck out of most people because they never mention that the exit poll data gets "corrected" to match the official results. This is especially sad because checks against raw exit poll results are the number one way to detect election fraud. Governments is Serbia, (the former Soviet Republic) Georgia, and the Ukraine were overthrown after large exit poll discrepancies indicated election fraud. You could argue that presenting raw exit poll data to the public and the world is the number one task of a free press in a Democracy.
The exit poll discrepancies in Ohio in 2004 led to a request for a recount. The recount was intentionally rigged to match the official results. This cheating was detected and two election officials recently received prison sentences of 18 months (the maximum) for rigging the recount. Thanks to their crime, we will never know if the election was rigged because the ballots were destroyed shortly after the two criminals showed that their pre-selected sample of ballots matched the official result.
The MSM talks about bringing Democracy to other lands (such as Iraq) but then turns around and does the best job it can to thwart Democracy in America in order to "avoid embarrassment". Judging by the many people on Slashdot that were fooled into thinking the exit poll data is not "corrected", it seems the MSM is doing a heck-of-a job.
If one group is disproportionately likely to avoid talking to pollsters - say, a group that routinely vilifies the "liberal bias in the MSM" - then that group will be systematically underrepresented, regardless of how careful pollsters are in their selection process.
I admit, as does the author of the paper, that it is possible the problem lies in the exit polls and not the official tally but if there is a problem with the polls, it couldn't possibly be a systematic error such as you describe. First of all, if there was a systematic error of this type (and there are potential systematic errors that need correction) then the pollsters would have corrected for it years ago, probably decades ago.
The problem is that the vast majority of exit polls agree with the official results. If this was a systematic error then it would have affected most states and most elections. Instead the anomalies occur in a few key states in a few key elections. For an idea like the one above to fit the data, it would have to explain why conservative voters are almost always willing to talk to pollsters but get pollster-shy in a few battleground states in a few key elections.
Your other suggestion. that the pollsters consistently miss the periods of time when most conservatives vote, doesn't work either. You don't explain why they are only missed in the key states and elections. Worse, since the conservatives need to be bunched for this effect to work then we would expect the pollsters to sometimes oversample the conservatives and this does not match the lopsided deviation almost always favoring the conservatives in the official results.
As a side note, did you know that two Ohio election officials (a Democrat and a Republican) were recently sentenced to the maximum of 18 months in jail for rigging the recount of the 2004 Ohio election? I've got to wonder why someone would go to the trouble and risk of rigging the recount if they didn't think/know there was
fraud in the election. Now, thanks to their crime, we can never find out because the ballots were quickly destroyed after they fudged the data that would have determined if a complete recount was required. The judge gave them the maximum sentence because he believed they were lying to him in order to protect higher-ups.
But you and I agree on the main point that we need verifiable paper trails and we need politicians like Dennis Kucinich who have the courage to demand a recount.
Is the question of tampering worth looking into? Of course, and I'm glad Kucinich is getting a recount going, and anything that gets more paper trails into voting is good,...
I fully agree.
... but the eagerness with which people are leaping on the notion of voting fraud - despite very, very thin evidence - is disturbing.
Perhaps you have the wrong thread. In my original post that you responded to I said:
On the other hand, I think it is possible to explain these very strange results without resorting to election fraud. Even so, I do think the current situation warants further scrutiny.
I then listed possible non-fraud explanations. Since that first post, you and I have been arguing about exit poll results, not fraud. I'm sure there were many people who jumped to the conclusion that this had to be fraud but I wasn't one of them.
Slashdot Mirror
... then write the code top-down.
This means you need one person who groks the entire project but you can have many people writing the code. See Feynman's critique of the design process used in for the Space Shuttle.
Back in WW-II, they added tube based analog computers to the anti-aircraft guns on some ships to help direct the shells/bullets to whatever was in the targeting sight.
IIRC, there were two problems with these devices. First, the tubes took too long to warm up because there was often very little advanced warning of an aerial attack. Second, there was a bug that got the sign of the Coriolis force wrong in the Southern Hemisphere.
The sailors ended up using fire axes to disable the automatic controls and return their guns to manual mode.
I heard this story from a fellow engineer almost 30 years ago and he was close to retirement age when he told it. He was also an artist and did the illustrations for the first edition of Ralph Nader's book, "Unsafe at any Speed".
My point was not "rational is good and non-rational is bad". I used the well documented fact that people respond non-rationally when faced with certain threats to suggest that another terror attack in the US would probably help the Republicans in November even though it happened "on their watch".
If anyone was questioning what I said in my original post, several of the responses, including yours, should erase all possible doubt. Thank you for your help by providing a specific example of the behavior I was discussing.
Your assumption and your misunderstanding of seven simple words combined with the visceral attacks in the rest of your response are an excellent example of the point I was trying to make. Thank you for your help.
The book explains that people are not rational or logical especially when it comes to risk assessment. The best recent example (the book was written in 1989) is America's reaction to the 9/11 attacks. More people died of hunger that day than were killed in the attack. The US response to the attacks was totally illogical because people felt threatened and this caused them to stop using the higher levels of their brains. They instead, reverted to their reptilian "flight or fight" instincts.
Another similar (or worse) attack will most likely produce a similar response from the American people. They will stop thinking rationally, which is probably the only way the Republicans can beat Obama on November 4th.
Oh please. Didn't those go out of style along with buggy whips?
... Com-casket. Or Can-comstic.
If you read all the comments following this rant, you will discover that the person who created the offending patch tried to check it with the OpenSSL devs by posting the patch to the openssl-dev mailing list.
Unfortunately that list is not for OpenSSL devs, instead it is for users of OpenSSL. Therefore only other clueless users saw the patch. To reach the OpenSSL devs one needed to use the openssl-team mailing list instead of openssl-dev.
IMO, this problem was due to a communication problem and it is hard to blame just one person for that. If I had to place blame, I would say the fault was with the poorly chosen names for the OpenSSL mailing lists.
... an iCurtain?
This phenomenon is explained in E. T. Jaynes' book Probability Theory: the Logic of Science. In it he develops the concept of an ideal plausible reasoner who can use inference. He shows that given a small set of desired traits all such reasoner with those traits conform to the same mathematical model which turns to be pretty much the same thing as Baysian probability theory.
Jaynes goes on to show that when two such ideal reasoners have differing views (due to their different past experiences) and are confronted with the same piece of new evidence, each one can believe the new evidence further reinforces their own view, thus polarizing the two ideal reasoners even further. This is a fundamental yet little known problem facing all human societies.
I think this is exactly what is going on in the fascinating study you brought up. It is nothing new. Even if humans were ideal reasoners (in the sense defined by Jaynes) we would still encounter these problems. Instead of blaming the audience for what is essentially a mathematical reality, it would be much more useful if we educated people about the underlying mathematics that causes this behavior.
I'm not familiar with the video game you are referring to. It looks like it is a lot of fun. Can you tell me where I can get a bootleg copy so I can try it out?
Wright's most radical comments are laced with anger and hatred so they inflame anger and hatred in many who hear them. This is why Gandhi, the Dalai Lama, and so many others teach that hatred and anger are not part of the path that will lead us to peace. The reason Obama needed to disavow Wright's words of hate is because they represent the wrong path. The problem is not the realities Wright exposed but the hate and anger he used in exposing them.
Obama's great strength comes from what I would describe as the spiritual balance of his campaign. He has eschewed the politics of hate and division. This gives him such power that the only way his opponents can stop him is to falsely attach to him a message of anger and hate. This is not Obama's message and this is why it was essential for Obama to disavow Wright's words of hate. The message Wright was giving in those sermons was exactly the opposite of Obama's message. This is why Wright's words were used to attack Obama and this is why Obama needed to make it clear that Wright's message of anger is not his message.
The TiVo demonstrates that open source DRM is possible. As many, many others have pointed out, if the end-user has control of the code then they can easily circumvent DRM measures. The TiVo closed this loophole by using DRM/encryption technology to control the code that runs on the user's device, making it impossible for users to run modified software that is not approved by a central authority. Yet the source code is open for all to see.
It is much more difficult (impossible) to use Free (as in freedom) software to implement DRM. Free software ensures the end-user has control over the code running on their own machine and then all of the oxymoron arguments made above apply. All free software is open source but not all open source software is free. When discussing things such as implementating DRM, it is a mistake to assume "open source" and "free" are synonymous. They're not.
Choose your words and your licenses wisely.
Let's recap:
- Many states (such as NH and Ohio) still count votes
using machines with secret sauce source code that have been proved to be trivial to crack, making it easy for a single person to alter the outcome of an entire election.
- The media via a private company have conspired to keep the raw exit poll data secret (see first link above) so it can't be used to check the official results.
- A recount was ordered in one of the states that could possibly change the overall winner of the entire election but that recount was rigged and the ballots were destroyed so we have no idea of who actually won.
This is not proof the election was rigged, if the votes had been honestly recounted they may have matched the official results. But why on earth would the two official in charge of the recount go to the trouble and risk of rigging it if they thought the election was honest? Unfortunately we'll never know if it was honest or not. Never knowing if the outcome of an election was actually fair seems to be more than just a minor problem.I don't agree with your point (2): As I may have said before, you usually want the clipping stage to be early in the processing chain. I would agree with you that the best solution is to have the soft clipping as part of the power amp so it can be calibrated to eliminate any hard clipping in the final power stage. Perhaps I'm old fashioned but I prefer to use a linear power amp (with hard clipping) and put a line level soft clipper in front of it even if it is built into the power amp.
You want the extra harmonics introduced by the soft clipper to be amplified by the power amp. Perhaps it is easiest to visualize this in the time domain. The clipper gracefully shaves off the high peaks and low valleys leaving a signal that is still fairly smooth but of limited range. It's the extra harmonics that do the rounding off. You want the power amp to make an exact replica of this signal except with a larger overall amplitude. That means you need to send the extra harmonics to the power amp. That's the whole point of the clipper.
Also, if the soft clipper is calibrated correctly then the power amp will never hard clip.
the chief drawback in having the soft clipper in what is essentially the preamp is that the soft clipping needs to be tuned to the specs of the power amp. For example, I think an extra volume control after the clipper may suffice. If the board with the tube on the end doesn't have a control to adjust the clipping then I would agree that the clipper won't be useful for all power amps. But there are probably some power amps that it works really well with.
My main point is that soft clipping is a very valuable tool to have in your bag of tricks. Also, IMO I think it is possible that the tube you described could actually do some good, especially if there is a way to tune the clipping to match the off-board power stage.
Here is an article that supports your ideas. It appears to be a London Times editorial even though the domain name is typepad.com.
It says that some women who voted for Clinton might have been lying about it to avoid conflicts with their husbands. I hadn't consider this possibility but it makes sense to me and therefore I am less inclined to believe there was a problem with the official tally, although I'm still glad they're going to do a recount.
Good audio engineers follow some of the same basic principles as good software engineers and good web site designers including:
If you want to know how the law actually works, please simply read the EFF brief. They cite cases where the intent to infringe did not amount to a copyright violation. I don't think it can get any clearer than that. Fair use had nothing to do with it.
If you actual read the fine brief by the EFF you will see that according to the law, intent does not matter. If intent was what mattered then the RIAA would certainly prevail since intentionally "making available" implies intent. The EFF cite the letter of the law and then give many case law examples that make clear that an actual transfer to a 3rd party must occur before it is a copyright violation. They cite cases where unauthorized people advertised copyrighted works and the defendants prevailed because there was no proof of an actual transfer.
They do mention one case where some (dead tree) libraries were successfully sued for copyright infringement without any proof of an actual transfer but that case seemed to be exceptional.
There are many unknowns in this case and in copyright law in general but one thing we do know is: intent does NOT matter in the eyes of the law.
Fair enough.
I've been using Gentoo for years and I love it. In the great FOSS tradition, Gentoo is all about choice (read: control). In the early years the stable branch was not very stable and the forums were filled with emerge problems in the stable branch. In the last few years those problems have almost all gone away. My system is mostly from the stable branch and I almost never run into a problem when I do updates. This is a very good thing[TM] IMO. I'm very glad the devs keep packages out of stable until they are almost always problem free.
But with Gentoo I can keyword-unmask specific packages if I want to be more on the bleeding edge and if I am really adventurous, I can hard-unmask packages. Yes, running a mixed system can require me to unmask dependencies but that has never been a nightmare (for me) and I don't see how the Gentoo devs could possibly prevent it. In fact, it should give you insight into why some packages seem to take longer than usual to make it into stable: all their dependencies need to go stable also.
It could well be that Gentoo is not the right distro for you. If your top priority is ease of use then one of the binary distros would probably be a better choice. But if you want the maximum control over your system, Gentoo is hard to beat.
IMO the MSM is responsible for confusing the heck out of most people because they never mention that the exit poll data gets "corrected" to match the official results. This is especially sad because checks against raw exit poll results are the number one way to detect election fraud. Governments is Serbia, (the former Soviet Republic) Georgia, and the Ukraine were overthrown after large exit poll discrepancies indicated election fraud. You could argue that presenting raw exit poll data to the public and the world is the number one task of a free press in a Democracy.
The exit poll discrepancies in Ohio in 2004 led to a request for a recount. The recount was intentionally rigged to match the official results. This cheating was detected and two election officials recently received prison sentences of 18 months (the maximum) for rigging the recount. Thanks to their crime, we will never know if the election was rigged because the ballots were destroyed shortly after the two criminals showed that their pre-selected sample of ballots matched the official result.
The MSM talks about bringing Democracy to other lands (such as Iraq) but then turns around and does the best job it can to thwart Democracy in America in order to "avoid embarrassment". Judging by the many people on Slashdot that were fooled into thinking the exit poll data is not "corrected", it seems the MSM is doing a heck-of-a job.
The problem is that the vast majority of exit polls agree with the official results. If this was a systematic error then it would have affected most states and most elections. Instead the anomalies occur in a few key states in a few key elections. For an idea like the one above to fit the data, it would have to explain why conservative voters are almost always willing to talk to pollsters but get pollster-shy in a few battleground states in a few key elections.
Your other suggestion. that the pollsters consistently miss the periods of time when most conservatives vote, doesn't work either. You don't explain why they are only missed in the key states and elections. Worse, since the conservatives need to be bunched for this effect to work then we would expect the pollsters to sometimes oversample the conservatives and this does not match the lopsided deviation almost always favoring the conservatives in the official results.
As a side note, did you know that two Ohio election officials (a Democrat and a Republican) were recently sentenced to the maximum of 18 months in jail for rigging the recount of the 2004 Ohio election? I've got to wonder why someone would go to the trouble and risk of rigging the recount if they didn't think/know there was fraud in the election. Now, thanks to their crime, we can never find out because the ballots were quickly destroyed after they fudged the data that would have determined if a complete recount was required. The judge gave them the maximum sentence because he believed they were lying to him in order to protect higher-ups.
But you and I agree on the main point that we need verifiable paper trails and we need politicians like Dennis Kucinich who have the courage to demand a recount.