Shout if I'm getting off-topic...
That's exactly the problem I'm trying to address. The point is that both sides can commit themselves to a particular set of results in advance, without letting on what those results are. Being committed to a particular set of results means that any attempt at cheating later on can be detected (as the hash result will be different; it's considered infeasible to forge a document to match a given hash).
When both sides are committed, the results can then be revealed; it doesn't matter who goes first, because they can't cheat without being detected.
DiyAudio is perfectly suitable for this protocol, as there is a well-defined sequence (the same for all participants) to all the messages.
Cheers
IH
Christer said:Ian,
Or rather, of
course it can, and I am sure we trust each other, but still
wouldn't it lower the credibility of the experiment if done
this way. The documents will not be published perfectly
simultaneously, so there is a possibility that someone
reads what has been posted so far and changes his
results/coding scheme/whatever before posting.
That's exactly the problem I'm trying to address. The point is that both sides can commit themselves to a particular set of results in advance, without letting on what those results are. Being committed to a particular set of results means that any attempt at cheating later on can be detected (as the hash result will be different; it's considered infeasible to forge a document to match a given hash).
When both sides are committed, the results can then be revealed; it doesn't matter who goes first, because they can't cheat without being detected.
DiyAudio is perfectly suitable for this protocol, as there is a well-defined sequence (the same for all participants) to all the messages.
Cheers
IH
Ian,
I see now I misunderstood you. Being a computer scientist I
should perhaps have realized that by hashing you actually
meant computing a hash function. However, since I didin't
know your profession and most people wouldn't know what
a hash function is, I wrongly assumed you used the term
hashing in a more diffuse sense, like permuting the numbers
of the cables. Sorry.
I thus agree with what you previously said.
Edit: Thinking about it, what you call a hash value, I would
rather call a checksum. I can't right now think of any real
difference from a mathematical point of view, it is just a
a matter of using different terms depending on the use of it.
A hash value is computed for storing data in hash tables,
while a checksum is computed to verify the integrity of data.
Maybe the term hashing is used elsewhere in the same sense
as in computer science but for different purposes.
I see now I misunderstood you. Being a computer scientist I
should perhaps have realized that by hashing you actually
meant computing a hash function. However, since I didin't
know your profession and most people wouldn't know what
a hash function is, I wrongly assumed you used the term
hashing in a more diffuse sense, like permuting the numbers
of the cables. Sorry.
I thus agree with what you previously said.
Edit: Thinking about it, what you call a hash value, I would
rather call a checksum. I can't right now think of any real
difference from a mathematical point of view, it is just a
a matter of using different terms depending on the use of it.
A hash value is computed for storing data in hash tables,
while a checksum is computed to verify the integrity of data.
Maybe the term hashing is used elsewhere in the same sense
as in computer science but for different purposes.
Christer said:Ian,
Maybe the term hashing is used elsewhere in the same sense
as in computer science but for different purposes.
Yes - I'm a cryptographer by trade, and was using 'hash' in the sense, for instance, of the Secure Hash Standard:
http://www.itl.nist.gov/fipspubs/fip180-1.htm
A cryptographic hash function has a number of important properties to distinguish it from other hashing or checksum algorithms. In this case the properties of interest of:
- (one-way property) given a hash value, it's infeasible to work out what the input to the hash function was (i.e. Eric/Frank shouldn't be able to work out how Steve coded the cables)
- (collision-resistance) it's infeasible to produce two distinct inputs which both give the same hash value output (i.e. Steve can't commit to one coding of the cables, and then switch to another if Eric/Frank get it right)
This has been a public service broadcast; thankyou for listening.
Cheers
IH
This has been a
Ian,
thanks for the explanation. That explains the confusion.
I didn't know the term was used also in cryptograhy, but
it is clear from your explanation that the cryptographic
definition is not compatible with the computer science
definition, so we actually do mean different things with it.
thanks for the explanation. That explains the confusion.
I didn't know the term was used also in cryptograhy, but
it is clear from your explanation that the cryptographic
definition is not compatible with the computer science
definition, so we actually do mean different things with it.
Aren't the results of this test going to be highly dependent on the equipment used?
What about the orientation of the wiring in a pair of headphones or loudspeakers? won't the effect of this be just as great as that of the interconnect used?
How can there ever be absolute agreement if everyone is using different equipment?
What about the orientation of the wiring in a pair of headphones or loudspeakers? won't the effect of this be just as great as that of the interconnect used?
How can there ever be absolute agreement if everyone is using different equipment?
wintermute said:OK that's the bit I was thinking about. How does he know what's the unshifted image and what's shifted. If we are talking a sideways shift of the sound stage, how do you determine what is normal with these particular cables unless you have a known correct one to compare to? Otherwise I agree, he could say same/opposite, but he may decide that the opposite cable is normal and the same cable is the abnormal one especially since we are talking subtle differences here (It would still be a valid result because he identified the difference, but not correct) sorry if I'm being pedantic........[/B]
That will be yet another function of the single initial test set. However they each apparently already know what sounds correct and what sounds shifted. And you're right, it would be just as statistically significant getting all of them wrong as getting all of them right.
se
Richard C said:Aren't the results of this test going to be highly dependent on the equipment used?
What about the orientation of the wiring in a pair of headphones or loudspeakers? won't the effect of this be just as great as that of the interconnect used?
How can there ever be absolute agreement if everyone is using different equipment?
Because they both agree that in their systems they're able to perceive when one cable is wired opposite the other. Also, they'll each be sent an a single initial pair with the proper ends marked to make sure they're confident they're perceiving the difference prior to their being sent the test sets.
In any case, as winterpute pointed out, it would be just as statistically significant getting all of them wrong as getting all of them right.
se
It is not important whether they get the direction right wrt. to
what it is according to the manufacturer. As long as the can
distinguish the two directions from each other, we have an
interesting result. OK, I think Steve said they will get pairs of
cables where either both are directed the same way or they
are in opposite directions. However, it does not change anything.
If they can sort the cable pairs into two groups that coincides
with how Steve has ordered the pairs, there is some kind of
audible difference. Whether they actually manage to tell which
group is which is in my opinion less important, at least for
a first experiment like this one.
Edit: Seems others said about the same thing in parallel.
what it is according to the manufacturer. As long as the can
distinguish the two directions from each other, we have an
interesting result. OK, I think Steve said they will get pairs of
cables where either both are directed the same way or they
are in opposite directions. However, it does not change anything.
If they can sort the cable pairs into two groups that coincides
with how Steve has ordered the pairs, there is some kind of
audible difference. Whether they actually manage to tell which
group is which is in my opinion less important, at least for
a first experiment like this one.
Edit: Seems others said about the same thing in parallel.
Hi,
Maybe we should try to have some backup candidates in case we both fail to spot any difference?
Cheers,😉
Also, they'll each be sent an a single initial pair with the proper ends marked to make sure they're confident they're perceiving the difference prior to their being sent the test sets.
Maybe we should try to have some backup candidates in case we both fail to spot any difference?
Cheers,😉
fdegrove said:Maybe we should try to have some backup candidates in case we both fail to spot any difference?
Sure. I assume by backup candidates you mean alternative cables, yes?
se
Christer said:If you two fail, do you really think there is someone else
who will pass? 🙂
Sure. Bob Crump.
Crump!
Crump!
He's our man!
If he can't do it
No one can!
GOOOOOOOOOOOOOOOOOO CRUMP! 🙂
Damn. Where's that cheerleader emoticon?
But I think Frank meant backup cables, not backup individuals. 🙂
se
Hi,
Actually I did mean backup individuals.
Come to think of it, having a backup set of cables isn't a bad idea either.
Neither is really a necessity for the moment though.
Cheers,😉
But I think Frank meant backup cables, not backup individuals.
Actually I did mean backup individuals.
Come to think of it, having a backup set of cables isn't a bad idea either.
Neither is really a necessity for the moment though.
Cheers,😉
Steve Eddy said:
But I think Frank meant backup cables, not backup individuals. 🙂
Yes, I never thought of that. I merely thought he was joking.
Maybe you should get a few different types of cables for the
initial test, and let them choose which type to use for the
real test? We should give them the best chance possible,
I think.
Christer said:Maybe you should get a few different types of cables for the
initial test, and let them choose which type to use for the
real test? We should give them the best chance possible,
I think.
Agreed. But we're at least somewhat limited in that selection because whatever's used for the test can't have ANY means of visual identification of direction. And there are a surprising number of ways that this could be done.
That's not to say that Eric and Frank are likely to try and cheat, only that it would undermine the credibility of the tests. And if we get a positive result, the credibility will need to be impeccable.
se
fdegrove said:Actually I did mean backup individuals.
Ah... Ok. Well, as I was saying...
Crump!
Crump!
He's our man!
If he can't do it
No one can!
GOOOOOOOOOOOOOO CRUMP!
He also has the good sense to live here in the US. 🙂
Come to think of it, having a backup set of cables isn't a bad idea either.
Neither is really a necessity for the moment though.
True. Let's go with the Vampire as planned and see where that takes us.
se
Steve Eddy said:
And if we get a positive result, the credibility will need to be impeccable.
se
Are we agreed on what constitutes a 'positive result'? 11/20 correctly identified almost certainly isn't, but 20/20 almost certainly is.
Just as a thought re visual cues: How about putting a length of heatshrink over the cable itself. It will act as a sort of tamper-evident seal (unless heatshrink has directionality too...)
One final thought. Steve - don't make exactly half the cables one way and exactly half the other - toss a coin for each cable. Otherwise, the listener will have a statistical advantage.
Cheers
IH
IanHarvey said:Are we agreed on what constitutes a 'positive result'? 11/20 correctly identified almost certainly isn't, but 20/20 almost certainly is.
Certainly.
But I ultimately want to do a number of trials (I was thinking 5) with 20 pairs per trial so that we can get a higher level of confidence. Remember, there will be plenty out there who would see a positive result as an extraordinary claim, and as the saying goes, extraordinary claims demand extraordinary proof.
Just as a thought re visual cues: How about putting a length of heatshrink over the cable itself. It will act as a sort of tamper-evident seal (unless heatshrink has directionality too...)
Hehehe. I've considered that. If nothing else it would add another level of security.
One final thought. Steve - don't make exactly half the cables one way and exactly half the other - toss a coin for each cable. Otherwise, the listener will have a statistical advantage.
Yes, that's the plan. I'd mentioned this previously when I noted that there will be a finite probability that all pairs will be the same in response to the issue of a control set.
se
Setting Some Test Conditions.........
Steve,
The quickest and easiest starter is to send us each two pairs, one pair with both channels in the same direction, and one pair with one channel reversed direction, and you have to be honest about this.
This way we can quickly and assuredly for ourselves determine that we do hear any difference between the two pairs, or not, for this example cable construction/wire type on our systems.
If we are happy that we are hearing reliable differences, then send us the 20 random pairs and we will sort them into categories for you.
So far I am talking about determining image sideways shift due to one channel reversal and therefore no directional (source end and load ends marked) designation is desired.
For the 20 pair experiment we need to decide if source and load ends be marked, and then as well as pairs having one channel reversed direction, some pairs may have both channels reversed.
If we are good enough, we should be able to distinguish both channels reversed also, but in fairness we should know in advance if this is a possibility or not a possibility, and then we know exactly what we are dealing with (listening for).
Throughout these tests, all strands MUST be in the same direction so that we are not dealing with additional variables.
Future tests could include variable strand direction also, but not for now.
Eric.
Steve,
The quickest and easiest starter is to send us each two pairs, one pair with both channels in the same direction, and one pair with one channel reversed direction, and you have to be honest about this.
This way we can quickly and assuredly for ourselves determine that we do hear any difference between the two pairs, or not, for this example cable construction/wire type on our systems.
If we are happy that we are hearing reliable differences, then send us the 20 random pairs and we will sort them into categories for you.
So far I am talking about determining image sideways shift due to one channel reversal and therefore no directional (source end and load ends marked) designation is desired.
For the 20 pair experiment we need to decide if source and load ends be marked, and then as well as pairs having one channel reversed direction, some pairs may have both channels reversed.
If we are good enough, we should be able to distinguish both channels reversed also, but in fairness we should know in advance if this is a possibility or not a possibility, and then we know exactly what we are dealing with (listening for).
Throughout these tests, all strands MUST be in the same direction so that we are not dealing with additional variables.
Future tests could include variable strand direction also, but not for now.
Eric.
Keep it simple
Don't add more variables than absolutely necessary. Don't
bring in both opposite direction pairs and pairs with both cables
the wrong way. Choose the one of these alternatives that
Eric and Frank consider easiest to detect.
We should make the test as easy as possible for them, as long
as it does not affect the validity of the test. The point is not to
try making them fail the test, but to try making them pass it.
If they fail, we are back at square one. We still have no
scientifically valid knowledge either way whether direction is
audible. If they pass the test, we have settled the issue (well,
the test should be repeated by others and with other guinea
pigs, but...)
Don't add more variables than absolutely necessary. Don't
bring in both opposite direction pairs and pairs with both cables
the wrong way. Choose the one of these alternatives that
Eric and Frank consider easiest to detect.
We should make the test as easy as possible for them, as long
as it does not affect the validity of the test. The point is not to
try making them fail the test, but to try making them pass it.
If they fail, we are back at square one. We still have no
scientifically valid knowledge either way whether direction is
audible. If they pass the test, we have settled the issue (well,
the test should be repeated by others and with other guinea
pigs, but...)
- Status
- Not open for further replies.
- Home
- General Interest
- Everything Else
- Cable Directionality (Moved Threadjacking)