More information, please...Whose "10 greenest cars" list, for example? If I pick the first Google result from that phrase, Civic Hybrid and Prius top the list.
Granted there are some clean diesel engines available in Europe. Why aren't they here in the U.S.? Incidentally there's no reason diesel and hybrid technology can't be combined as well. I would expect to see such cars sometime in the next several years.
It is also worse then what the last generation of Petrol gets as well.
What the heck is "the last generation of Petrol"?
I'm not at all familiar with the three models you cited from Daihatsu and Toyota. Are they midsized (like the Prius), compact (like the Honda Civic), or subcompact (like the Honda Insight)? I just want to make sure it's a fair comparison.
Honda Civic Hybrid has better fuel consumption and better dynamic characteristics
Are you using the 2004/2005 Prius in your comparison, or the earlier ones? I think your information is old; here's an article to bring you up to date. Note that the Civic Hybrid gets worse mileage, and is slower than the Prius in the 0-to-60 acceleration department. Insight is a bit quicker and gets better MPG, but those things are tiny. BTW, when California has just increased fines for exceeding 100 MPH to $750, who cares if the top speed is 115 or 130?
Think of it this way. You're driving at 50 MPH. Maintaining this speed only requires you to expend enough energy to overcome aerodynamic drag and rolling resistance; let's say you're getting 50 MPG (a reasonable number, maybe even a bit pessimistic, for the Prius at this speed with no wind on flat land). Now if you come to a complete stop, your kinetic energy is converted partially to heat (via drag, rolling resistance, and the brakes) and partially to regenerated electricity (the conversion is not 100% efficient so some more energy is lost to heat in this process). Now you have to get back up to the same speed which means your engine needs to produce the kinetic energy of a 3000-pound object moving at 50 MPH all over again. The electrical power system is only going to provide a fraction of that energy; the rest comes from the ICE, meaning you burn a significant amount of gasoline to reaccelerate. Allowing for gasoline containing 115000 BTU per gallon, and 37% efficiency from the engine, in about a quarter of a mile you have consumed about 0.0076 gallon (i.e., you are only getting about 33 MPG over the acceleration stretch).
That calculation leaves out several considerations, such as rolling resistance and drag during the acceleration. Empirically I'd say the Prius gets about 20 MPG during acceleration (as shown on the consumption display).
Er, that's beside the point. All cars are penalized by red lights. Hybrids only less so. But we were talking about EPA tests that don't feature that kind of incessant stop-and-go; just as with any other car, a hybrid does best when it can keep moving continuously. And that is why real-world experience seldom matches the EPA numbers.
Apples and oranges, I say. Gasoline has an energy content of about 115000 BTU per gallon (it will vary with the formulation due to things like oxygenating additives). Diesel has an energy content of around 130000 BTU per gallon (and varies less, from what I've read).
I agree with you about pollution -- grandparent's Jetta will pollute a lot more than the Prius.
The EPA rates the Prius at 60 mpg. The 55 was from a blog.
The EPA estimates have little to do with real-world mileage. While some people *do* achieve 55 MPG and higher (mostly be driving under 55 MPH everywhere they go), if you drive in real city conditions you will probably see a lot less. Of course it depends on the city. Every traffic light costs energy. If you are lucky to live in a city that has traffic light synchronization then you might see 50 to 60 MPG in the Prius. But forget about it in most Southern California cities, where you have a red light every quarter of a mile or so.
On a recent long trip our Prius got very close to the EPA highway estimate overall, and exceeded it on flat land (east Texas and Louisiana, e.g.). And we were going 75 to 80 MPH most of the time.
The point is that given a particular file, you can compute other files that collide with that file
No, that would be a preimage attack (read the FAQ). The md5sum attack is a collision attack: you can create two files that have the same md5sum, but you can't specify either of the two files in advance.
Except that this does not give you dramatically better security.
Actually it can. You are comparing uncompromised hashes only. Consider that as long as the best known approach is brute force, then the difficulty of finding pairs of data sets that have matching hashes using hash algorithm 1 is exponential w.r.t. the length of the hash. Now find a method that compromises the security of the hash (as these researchers did) and the difficulty becomes merely polynomial, perhaps even linear.
But if you add another constraint (such as requiring both hash function 1 and hash function 2 to produce the same output for the two files), then you have the following situation: Your algorithm can create 2 files that match with hash 1. The probability that those 2 files also match with hash 2 -- provided hash 1 and hash 2 don't have some common periodicity or similar characteristic between them -- is 1 in 2^N, where N is the length in bits of hash 2. So you reseed your algorithm and try again. You must do this an average of 2^(N-1) times before you get a hit. Even having an algorithm for hacking hash 2 doesn't help, because both algorithms must be met with the same pair of files. So it does return you to exponential difficulty (a dramatic improvement over the compromised hash), at least until someone finds an algorithm that compromises the combination of hash 1 and hash 2.
Note: The above argument might not hold in the case of preimage attacks (which the md5sum attack is not), but I can't see why it would not hold for collision attacks.
I have one of these (can't remember the model number) and it is excellent -- maybe not in terms of speed (which isn't a critical factor in my use) but in its support for DVD-RAM. Now I can record DVD-RAMs on my video recorder and read them on my PC. DVD-RAM is also a nice backup medium for a small system.
They reviewed the LG, which is the only other drive I know of that has similar support. I'm not sure if LG will accept DVD-RAM media in cartridges, as the Panasonic does. That is a drawback if they can't.
It would be nice if they had included a Panasonic drive so it could be compared side-by-side with the LG.
Wait a second -- If Condorcet is too difficult for voters to rank when the number of candidates is large, why isn't that true for IRV? The ballots would be identical; only the scoring system would be different.
Given the flaws in IRV I don't think it should be used for any single-winner elections (its close relative STV isn't too bad for multi-winner elections, though).
For single-winner elections, the best solution from a voter simplicity standpoint, for small and large numbers of candidates, would be approval, as the ballots are very easy to mark.
Condorcet will allow both restrictions to be lifted, but of course we know it is not perfect, because there cannot be a perfect voting method...
That is a common misconception, but Arrow's theorem does not mean that there cannot be a perfect voting method. It states that no voting method can meet a certain predefined set of criteria. Whether that set of criteria is necessary for a perfect voting system is a value judgement. As a mathematical theorem, Arrow's is value-free.
One criterion that I believe could be dispensed with is the Independence of Irrelevant Alternatives. The argument for doing so is this: If there is an election with N candidates, and then candidate N+1 enters the race, then each ballot now contains positional information for the other N candidates relevant to candidate N+1. That information was missing before. Why shouldn't the result change in light of new information?
As a more concrete example, suppose I said I prefer Kerry to Bush. That tells you something, but you would learn something about the strength of that preference if I told you whether I preferred Badnarik to Kerry to Bush, or Kerry to Badnarik to Bush.
Another change that could be made to Arrow's criteria is to allow non-ranked voting systems, such as approval voting. The criteria as written simply disallow systems that don't allow handle ranking of the candidates. If this arbitrary restriction is removed, approval voting passes the remaining criteria.
Let's say a group of 3 voters, all of whom like A better than B and B better than C, are trying to decide whether it's worth it to go to the polls.
Suppose the exit polling projects the result as follows:
9 A > B > C
10 B > C > A
13 C > B > A
A is eliminated in the first round, making B the projected winner (19 votes to 13).
Now they figure, with the race this close, they can prevent their favorite candidate, A, from being eliminated in the first round if they vote. So they go to the polls with just minutes remaining, and cast their ballots. Next morning the results are published:
12 A > B > C
10 B > C > A
13 C > B > A
B is eliminated in the first round, so C wins, 23 to 12.
In other words, it's as if this group is punished for voting, since the result of their action is, for them, the worst possible outcome!
I don't know if you mean "can/can not" in a legal sense or in a practical sense. If by "can investigate" you mean that Congress has this power legally then the fact is, they can subpoena anybody within their reach, so "can not compell" [sic] taken in the same sense is not correct. But if you mean it practically, I agree that Congress cannot send a federal marshall to bring in a President (or anybody) that refuses to comply with a subpoena, because Congress does not command any law enforcement officers. The remedy available to Congress in that case would be to cite the President for contempt of Congress, which I believe would result in a criminal trial (in the judicial branch, of course -- separation of powers at work again). Of course, that is never going to happen to a Republican President under a Republican-dominated Congress, so this was a simple face-saving agreement that allowed Congress to get the "testimony" it wanted without appearing to back down.
You'll recall that Clinton testified under oath but he did not have the advantage of having Democrats in control of Congress at the time.
In any case your reply is a change from your earlier tune, which was the congress has no business investigating the office of the president. Certainly that is not implied by "separation of powers".
Um... they didn't have to testify at all - due to separation of powers. Since this wasn't an impeachment hearing (at least not yet) or independent council investigation, the congress has no business investigating the office of the president.
I'm as much a believer in separation powers as anyone. But separation of powers does not prevent Congress from investigating the executive branch. Separation of powers means that the Constitution grants certain exclusive powers to each branch of the government; for example, the executive branch cannot declare war or constitute tribunals; the judicial branch cannot carry out impeachments; and the legislative branch cannot pardon crimes. Separation of powers also encompasses the idea of checks and balances. If one takes the view that this investigatory role is a check on the executive branch then separation of powers actually does give Congress this power.
You might have a more persuasive argument if you had said that the Constitution simply does not grant the power to Congress to conduct investigations. And you'd be right, if we only consider powers that the Constitution grants explicitly. However, it is generally understood that investigatory powers are granted to Congress implicitly, and this power is not so limited as to exclude investigations of the executive branch.
No, I think it was the edits that cut out any smooth transitions between the last four or five scenes that made it seem like it was too long and a bit unfocused. If the EE fills in those breaks then it will be an improvement.
Slashdot Mirror
Granted there are some clean diesel engines available in Europe. Why aren't they here in the U.S.? Incidentally there's no reason diesel and hybrid technology can't be combined as well. I would expect to see such cars sometime in the next several years.
It is also worse then what the last generation of Petrol gets as well.
What the heck is "the last generation of Petrol"?
I'm not at all familiar with the three models you cited from Daihatsu and Toyota. Are they midsized (like the Prius), compact (like the Honda Civic), or subcompact (like the Honda Insight)? I just want to make sure it's a fair comparison.
Honda Civic Hybrid has better fuel consumption and better dynamic characteristics
Are you using the 2004/2005 Prius in your comparison, or the earlier ones? I think your information is old; here's an article to bring you up to date. Note that the Civic Hybrid gets worse mileage, and is slower than the Prius in the 0-to-60 acceleration department. Insight is a bit quicker and gets better MPG, but those things are tiny. BTW, when California has just increased fines for exceeding 100 MPH to $750, who cares if the top speed is 115 or 130?
That calculation leaves out several considerations, such as rolling resistance and drag during the acceleration. Empirically I'd say the Prius gets about 20 MPG during acceleration (as shown on the consumption display).
Er, that's beside the point. All cars are penalized by red lights. Hybrids only less so. But we were talking about EPA tests that don't feature that kind of incessant stop-and-go; just as with any other car, a hybrid does best when it can keep moving continuously. And that is why real-world experience seldom matches the EPA numbers.
Apples and oranges, I say. Gasoline has an energy content of about 115000 BTU per gallon (it will vary with the formulation due to things like oxygenating additives). Diesel has an energy content of around 130000 BTU per gallon (and varies less, from what I've read).
The EPA rates the Prius at 60 mpg. The 55 was from a blog.
The EPA estimates have little to do with real-world mileage. While some people *do* achieve 55 MPG and higher (mostly be driving under 55 MPH everywhere they go), if you drive in real city conditions you will probably see a lot less. Of course it depends on the city. Every traffic light costs energy. If you are lucky to live in a city that has traffic light synchronization then you might see 50 to 60 MPG in the Prius. But forget about it in most Southern California cities, where you have a red light every quarter of a mile or so.
On a recent long trip our Prius got very close to the EPA highway estimate overall, and exceeded it on flat land (east Texas and Louisiana, e.g.). And we were going 75 to 80 MPH most of the time.
No, that would be a preimage attack (read the FAQ). The md5sum attack is a collision attack: you can create two files that have the same md5sum, but you can't specify either of the two files in advance.
Actually it can. You are comparing uncompromised hashes only. Consider that as long as the best known approach is brute force, then the difficulty of finding pairs of data sets that have matching hashes using hash algorithm 1 is exponential w.r.t. the length of the hash. Now find a method that compromises the security of the hash (as these researchers did) and the difficulty becomes merely polynomial, perhaps even linear.
But if you add another constraint (such as requiring both hash function 1 and hash function 2 to produce the same output for the two files), then you have the following situation: Your algorithm can create 2 files that match with hash 1. The probability that those 2 files also match with hash 2 -- provided hash 1 and hash 2 don't have some common periodicity or similar characteristic between them -- is 1 in 2^N, where N is the length in bits of hash 2. So you reseed your algorithm and try again. You must do this an average of 2^(N-1) times before you get a hit. Even having an algorithm for hacking hash 2 doesn't help, because both algorithms must be met with the same pair of files. So it does return you to exponential difficulty (a dramatic improvement over the compromised hash), at least until someone finds an algorithm that compromises the combination of hash 1 and hash 2.
Note: The above argument might not hold in the case of preimage attacks (which the md5sum attack is not), but I can't see why it would not hold for collision attacks.
You might not even need to use different algorithms. What if you simply permute the data you are checking?
Call the original data D, and call the aliased data (that has the same md5sum) E.
md5sum(D) = md5sum(E)
Now permute D and E by some measure, such as reversing the bits in each byte, to get D' and
E'. What are the chances that
md5sum(D') = md5sum(E')?
The goal is to make it harder to construct an alias data set that satisfies both conditions.
Of course the technique is extensible by creating more permutations.
Another possibility is to append a 32-bit CRC to the data before running md5. I suspect
md5sum({D,CRC(D)}) = md5sum({E,CRC(E)})
would be very hard to solve.
I'll take that one step further: You you do not have to "harvest babies" to get embryonic stem cells.
An embryo is not a baby. A fetus is not a baby. An embryo is not a fetus.
I have one of these (can't remember the model number) and it is excellent -- maybe not in terms of speed (which isn't a critical factor in my use) but in its support for DVD-RAM. Now I can record DVD-RAMs on my video recorder and read them on my PC. DVD-RAM is also a nice backup medium for a small system.
They reviewed the LG, which is the only other drive I know of that has similar support. I'm not sure if LG will accept DVD-RAM media in cartridges, as the Panasonic does. That is a drawback if they can't.
It would be nice if they had included a Panasonic drive so it could be compared side-by-side with the LG.
Wait a second -- If Condorcet is too difficult for voters to rank when the number of candidates is large, why isn't that true for IRV? The ballots would be identical; only the scoring system would be different. Given the flaws in IRV I don't think it should be used for any single-winner elections (its close relative STV isn't too bad for multi-winner elections, though). For single-winner elections, the best solution from a voter simplicity standpoint, for small and large numbers of candidates, would be approval, as the ballots are very easy to mark.
Condorcet will allow both restrictions to be lifted, but of course we know it is not perfect,
because there cannot be a perfect voting method...
That is a common misconception, but Arrow's theorem does not mean that there cannot be a perfect voting method. It states that no voting method can meet a certain predefined set of criteria. Whether that set of criteria is necessary for a perfect voting system is a value judgement. As a mathematical theorem, Arrow's is value-free.
One criterion that I believe could be dispensed with is the Independence of Irrelevant Alternatives. The argument for doing so is this: If there is an election with N candidates, and then candidate N+1 enters the race, then each ballot now contains positional information for the other N candidates relevant to candidate N+1. That information was missing before. Why shouldn't the result change in light of new information?
As a more concrete example, suppose I said I prefer Kerry to Bush. That tells you something, but you would learn something about the strength of that preference if I told you whether I preferred Badnarik to Kerry to Bush, or Kerry to Badnarik to Bush.
Another change that could be made to Arrow's criteria is to allow non-ranked voting systems, such as approval voting. The criteria as written simply disallow systems that don't allow handle ranking of the candidates. If this arbitrary restriction is removed, approval voting passes the remaining criteria.
A better example of bad IRV characteristics:
Let's say a group of 3 voters, all of whom like A better than B and B better than C, are trying to decide whether it's worth it to go to the polls.
Suppose the exit polling projects the result as follows:
9 A > B > C
10 B > C > A
13 C > B > A
A is eliminated in the first round, making B the projected winner (19 votes to 13).
Now they figure, with the race this close, they can prevent their favorite candidate, A, from being eliminated in the first round if they vote. So they go to the polls with just minutes remaining, and cast their ballots. Next morning the results are published:
12 A > B > C
10 B > C > A
13 C > B > A
B is eliminated in the first round, so C wins, 23 to 12.
In other words, it's as if this group is punished for voting, since the result of their action is, for them, the worst possible outcome!
I don't know if you mean "can/can not" in a legal sense or in a practical sense. If by "can investigate" you mean that Congress has this power legally then the fact is, they can subpoena anybody within their reach, so "can not compell" [sic] taken in the same sense is not correct. But if you mean it practically, I agree that Congress cannot send a federal marshall to bring in a President (or anybody) that refuses to comply with a subpoena, because Congress does not command any law enforcement officers. The remedy available to Congress in that case would be to cite the President for contempt of Congress, which I believe would result in a criminal trial (in the judicial branch, of course -- separation of powers at work again). Of course, that is never going to happen to a Republican President under a Republican-dominated Congress, so this was a simple face-saving agreement that allowed Congress to get the "testimony" it wanted without appearing to back down.
You'll recall that Clinton testified under oath but he did not have the advantage of having Democrats in control of Congress at the time.
In any case your reply is a change from your earlier tune, which was the congress has no business investigating the office of the president. Certainly that is not implied by "separation of powers".
*Sigh* If only somebody had told this guy...
Um... they didn't have to testify at all - due to separation of powers. Since this wasn't an impeachment hearing (at least not yet) or independent council investigation, the congress has no business investigating the office of the president.
I'm as much a believer in separation powers as anyone. But separation of powers does not prevent Congress from investigating the executive branch. Separation of powers means that the Constitution grants certain exclusive powers to each branch of the government; for example, the executive branch cannot declare war or constitute tribunals; the judicial branch cannot carry out impeachments; and the legislative branch cannot pardon crimes. Separation of powers also encompasses the idea of checks and balances. If one takes the view that this investigatory role is a check on the executive branch then separation of powers actually does give Congress this power.
You might have a more persuasive argument if you had said that the Constitution simply does not grant the power to Congress to conduct investigations. And you'd be right, if we only consider powers that the Constitution grants explicitly. However, it is generally understood that investigatory powers are granted to Congress implicitly, and this power is not so limited as to exclude investigations of the executive branch.
No, I think it was the edits that cut out any smooth transitions between the last four or five scenes that made it seem like it was too long and a bit unfocused. If the EE fills in those breaks then it will be an improvement.