I don't think I've ever seen such a load of crock as I've seen in the first 51 responses to this article. No one, and I mean, no one, seems to have a clue.
Or maybe I've just missed the sarcasm tags in every post.
It has been noted above, but faith is supposed to be something that gets people to think about their problems and see what they can do for themselves first.
To the extent that atheism also helps people try to solve their own problems, atheism has value in God's universe.
For many people, the prayer for help is less for intervention than for teaching.
Part of the purpose is for the person praying to at least think far enough to figure out what kind of help is really necessary, and the usual result is help figuring out what to do.
Atheists can argue that meditation should be sufficient without having to brown-nose some non-existent supreme being, but that kind of misses the point.
Sometimes there are other things that happen, too, usually people (friends, strangers, even enemies) who provide just enough help, once the person praying has started solving the problem on his or her own. Atheists can argue that such people would likely help anyway, but that's also kind of missing the point.
No, not really everywhere. Not nearly as everywhere as 68K (still). No where nearly as everywhere as ARM.
x86 is not really appropriate to embedded and real time. If the other processors which are tend to get tuned, well, x86 in embedded is going to tend to get even more special hardware.
Not the same x86, however, and that's really the problem we're trying to talk about here. (And failing.)
When you need a secret but verifiable ballot, on-line voting cannot provide either in a way that most voters can understand.
Not just hard problems. Not just NP Complete. We're trying to build systems based on internal contradictions here. You're trying to say that, with a little work, we can make a system where true equals false.
(It hardly takes any work at all to make a system that says true equals false. Different problem, however.)
Butterfly ballots are not the only kind of physical ballot.
The hanging chad is only one of the problems.
They were developed when we didn't yet have optical scanning techniques and machines cheap enough to use for elections, and when paper was (comparatively speaking) expensive enough to motivate separating the candidate list from the ballot. (It was still a bad idea, but the motivation can be understood.)
Bubble sheets incorporating the candidate's names, putting the ballot in a sleeve before leaving the voting booth, is about as good as it gets. (And then someone forgets the blanks for write-in.) And those are better than any computer voting system.
You may have the skills to check a few of those boxes on voting day, but are all the non-geeks just supposed to trust you? And what about all the boxes you can't check?
You can't hide what you do and ask people to trust it. That's the problem with this kind of code. Even for the people that can understand the math, how do they know for sure that the system is not just taking all the shortcuts?
The best system is already in use, and that's the physical ballot, where the voter puts the ballot in a sleeve to hide it from the judges before it goes into the ballot box.
There are some specific details, but computers in any role other than counting physical ballots just get in the way.
In addition to my long tirade above, I'll point out what I've pointed it out elsewhere, the question is not where the genetic patterns came from. It's that we are moving from engineering through the interfaces that nature most commonly uses (breeding) to actually getting into the low-level code, and we still don't have a complete set of documents for any of the systems we are playing with.
It's kind of like script kiddies graduating from playing with VB to playing with the kernel and they have only a partial, reverse-engineered manual. And if the kernel panics, it's the system we are trying to live in that does something we probably didn't want it to.
We want to experiment if we want a good manual, but we want to do it carefully, not on a schedule determined arbitrarlily by some greedy board of directors.
Invoking evolutionary time doesn't help. Actually, it's similar to your Churchill anecdote in some respects.
Leaving aside the historicity, there are a lot of questions begged by your story, and I'm going to ask you to walk through some of them with me.
What is the intent of the story?
If it is not apocryphal, what was Churchill's intent?
How was the woman raised? What are her circumstances? I'm personally of the opinion that women should not sell themselves for any price, but I am not particularly anxious to insult a woman just because she sees a difference between a million pounds and five. And there is a difference in many contexts.
If you absolutely insist on getting laid, five pounds will get you laid in some neighborhoods, 500 pounds will get you laid in others, etc., and in some places it takes a marriage contract.
For example, clock a byte register through all it's possible values. That's fast. Now do it randomly. It will probably take a bit longer, true? But if you have a statistically random sequence generator, it will probably not take too much time. Probably, if the generator has true statistical randomness.
How many effective bits are there in a mosquito's genes?
That's that part that's similar to your anecdote. You are assuming, when you invoke random permutation as if there were no time limits and as if mutation were the same as permutation, that five is as good as a million.
With only 59 effective bits, even at a strictly linear count with one permutation a second, you exceed the expected life of the solar system.
Now, I know you're going to claim that this is not the same problem, that we are picking relatively small strings in the genetic sequence, and that the changes are neither sequential counting nor random. But you are not asserting nature mimicking us, you are asserting coincidence between the processes.
Now that you're thinking, remember that nature works in parallel, but remember also that there is a selection involved. Some of the random stuff gets suppressed before it actually goes live, so we are not talking full permutation. And we don't have docs to tell us which combinations will be suppressed. Well, we can calculate some of the suppressed stuff, but we really don't have enough evidence to be sure of the calcuations that we can do.
The assumption that all possible permutations of a gene sequence will occur eventually is not equivalent to the assumption that all possible mutations will occur, okay?
By the way, remember that nature lopped off the dinosaurs.
Are you realy okay with saying, effectively, that it ought to be okay with everyone if our genetic experiments end up lopping humans off the evolutionary tree a little early? Is a million years not too early for you? How about five years?
A lot of the permutation paths are trimmed by natural selection, so, indeed, we could make mutations that would be impossible in nature, even given a steady state universe and suns that never go nova.
If you aren't convinced there are such things as trimmed selection paths, you still have to consider the limits of time. How many clocks does it run a single 128 bit counter through a full count? How fast is the counter going to have to run to complete the count during the amount of time a planet like our earth is in a mode conducive to our type of life?
The problem of evolving us as a species is dramatically helped by concurrency, because of natural selection.
The problem of evolving a specific genetic permutation is only made worse by the implications of concurrency.
The traditional techniques basically used the closest thing we have to a published API for the systems. And not all of the results there were exactly benign.
The new techniques are not even using "unpublished" APIs. They are basically digging into the code inside the modules and cutting and pasting, and the only documentation we have is very limited partial documentation gained by reverse-engineering.
Slashdot Mirror
I don't think I've ever seen such a load of crock as I've seen in the first 51 responses to this article. No one, and I mean, no one, seems to have a clue.
Or maybe I've just missed the sarcasm tags in every post.
Never heard of spoofing, I guess. Maybe that's why you want to trust on-line voting.
Or maybe you want to be the spoofer?
It has been noted above, but faith is supposed to be something that gets people to think about their problems and see what they can do for themselves first.
To the extent that atheism also helps people try to solve their own problems, atheism has value in God's universe.
For many people, the prayer for help is less for intervention than for teaching.
Part of the purpose is for the person praying to at least think far enough to figure out what kind of help is really necessary, and the usual result is help figuring out what to do.
Atheists can argue that meditation should be sufficient without having to brown-nose some non-existent supreme being, but that kind of misses the point.
Sometimes there are other things that happen, too, usually people (friends, strangers, even enemies) who provide just enough help, once the person praying has started solving the problem on his or her own. Atheists can argue that such people would likely help anyway, but that's also kind of missing the point.
"cuase dingbat" it's time to boycott.
Are we willing to take a few years of limitations on high-def and premium content to take a stand here?
If freedom really matters, it should matter more than the high definition or the premium content, etc.
Standard platforms are okay, but eventually we need to start working on tools to deal with the complexity.
We need a whole bunch of co-existent forks.
And some way to manage them all.
That's the whole point here, I think.
No, not really everywhere. Not nearly as everywhere as 68K (still). No where nearly as everywhere as ARM.
x86 is not really appropriate to embedded and real time. If the other processors which are tend to get tuned, well, x86 in embedded is going to tend to get even more special hardware.
Not the same x86, however, and that's really the problem we're trying to talk about here. (And failing.)
Those are blocks.
When you need a secret but verifiable ballot, on-line voting cannot provide either in a way that most voters can understand.
Not just hard problems. Not just NP Complete. We're trying to build systems based on internal contradictions here. You're trying to say that, with a little work, we can make a system where true equals false.
(It hardly takes any work at all to make a system that says true equals false. Different problem, however.)
As someone else has pointed out, with the bank you only lose money.
Votes are more important than money.
Not worth more in monetary terms, more important. Way more important.
I have long thought the the best approach in many cases is to count a non-vote as a vote against all.
It won't work in some cases, however.
A non-vote can be a protest, but changing to a digital system does nothing to help the protest be registered.
And it opens up a huge bunch of holes to cheating.
Use machines in the counting process, after the voting is done. Keep them away from the actual place where people vote.
Butterfly ballots are not the only kind of physical ballot.
The hanging chad is only one of the problems.
They were developed when we didn't yet have optical scanning techniques and machines cheap enough to use for elections, and when paper was (comparatively speaking) expensive enough to motivate separating the candidate list from the ballot. (It was still a bad idea, but the motivation can be understood.)
Bubble sheets incorporating the candidate's names, putting the ballot in a sleeve before leaving the voting booth, is about as good as it gets. (And then someone forgets the blanks for write-in.) And those are better than any computer voting system.
You may have the skills to check a few of those boxes on voting day, but are all the non-geeks just supposed to trust you? And what about all the boxes you can't check?
You can't hide what you do and ask people to trust it. That's the problem with this kind of code. Even for the people that can understand the math, how do they know for sure that the system is not just taking all the shortcuts?
The best system is already in use, and that's the physical ballot, where the voter puts the ballot in a sleeve to hide it from the judges before it goes into the ballot box.
There are some specific details, but computers in any role other than counting physical ballots just get in the way.
You want to embed x86?
For real?
Seriously?
You're not just joking?
It was INTEL that insisted on this stupid approach with their version of UWB.
A $5.00 part does not raise the price by just $5.00.
FWIW.
But this is bad news all around, and will result in havoc.
In addition to my long tirade above, I'll point out what I've pointed it out elsewhere, the question is not where the genetic patterns came from. It's that we are moving from engineering through the interfaces that nature most commonly uses (breeding) to actually getting into the low-level code, and we still don't have a complete set of documents for any of the systems we are playing with.
It's kind of like script kiddies graduating from playing with VB to playing with the kernel and they have only a partial, reverse-engineered manual. And if the kernel panics, it's the system we are trying to live in that does something we probably didn't want it to.
We want to experiment if we want a good manual, but we want to do it carefully, not on a schedule determined arbitrarlily by some greedy board of directors.
Invoking evolutionary time doesn't help. Actually, it's similar to your Churchill anecdote in some respects.
Leaving aside the historicity, there are a lot of questions begged by your story, and I'm going to ask you to walk through some of them with me.
What is the intent of the story?
If it is not apocryphal, what was Churchill's intent?
How was the woman raised? What are her circumstances? I'm personally of the opinion that women should not sell themselves for any price, but I am not particularly anxious to insult a woman just because she sees a difference between a million pounds and five. And there is a difference in many contexts.
If you absolutely insist on getting laid, five pounds will get you laid in some neighborhoods, 500 pounds will get you laid in others, etc., and in some places it takes a marriage contract.
For example, clock a byte register through all it's possible values. That's fast. Now do it randomly. It will probably take a bit longer, true? But if you have a statistically random sequence generator, it will probably not take too much time. Probably, if the generator has true statistical randomness.
How many effective bits are there in a mosquito's genes?
That's that part that's similar to your anecdote. You are assuming, when you invoke random permutation as if there were no time limits and as if mutation were the same as permutation, that five is as good as a million.
With only 59 effective bits, even at a strictly linear count with one permutation a second, you exceed the expected life of the solar system.
Now, I know you're going to claim that this is not the same problem, that we are picking relatively small strings in the genetic sequence, and that the changes are neither sequential counting nor random. But you are not asserting nature mimicking us, you are asserting coincidence between the processes.
Now that you're thinking, remember that nature works in parallel, but remember also that there is a selection involved. Some of the random stuff gets suppressed before it actually goes live, so we are not talking full permutation. And we don't have docs to tell us which combinations will be suppressed. Well, we can calculate some of the suppressed stuff, but we really don't have enough evidence to be sure of the calcuations that we can do.
The assumption that all possible permutations of a gene sequence will occur eventually is not equivalent to the assumption that all possible mutations will occur, okay?
By the way, remember that nature lopped off the dinosaurs.
Are you realy okay with saying, effectively, that it ought to be okay with everyone if our genetic experiments end up lopping humans off the evolutionary tree a little early? Is a million years not too early for you? How about five years?
Not exactly.
A lot of the permutation paths are trimmed by natural selection, so, indeed, we could make mutations that would be impossible in nature, even given a steady state universe and suns that never go nova.
If you aren't convinced there are such things as trimmed selection paths, you still have to consider the limits of time. How many clocks does it run a single 128 bit counter through a full count? How fast is the counter going to have to run to complete the count during the amount of time a planet like our earth is in a mode conducive to our type of life?
The problem of evolving us as a species is dramatically helped by concurrency, because of natural selection.
The problem of evolving a specific genetic permutation is only made worse by the implications of concurrency.
The traditional techniques basically used the closest thing we have to a published API for the systems. And not all of the results there were exactly benign.
The new techniques are not even using "unpublished" APIs. They are basically digging into the code inside the modules and cutting and pasting, and the only documentation we have is very limited partial documentation gained by reverse-engineering.
I guess I should do some research before hazarding a guess, though.
Until there, I was wondering, now I'm wondering something different.