[SoundAndMotion]

I did the experiment, blind be it without randomized script etc, with a couple of friends and we were able to separate the various configurations by preference, with fiberoptic being the most preferred option.

Tell more: how was it blind? FYI,"blind" is not about vision, but about knowledge. You can do a sighted test wearing a blindfold (I simply announce what you're listening to) and you can do a blind test with your eyes wide open (as long as you can't see the configuration). If you didn't randomize, ... you knew the order? ...! If so, it wasn't "blind".

EDIT:

@SoundAndMotion : that is pretty much the first reasonable and open minded response in this conversation!

Thanks!

 

[marcelnl]

I work in clinical development, blind means single blind but with intact eyesight here, as in; the observers not knowing which connection was used, and me trying to fool them by making changes in a 'haphazard' way (I will not call it random) manner; as in BA AAB ABA AB BB ABAB etc.

Of course I knew the order, someone has to make the connections. In every (double) blind study there is someone unblinded at some point, to keep the blind you ensure the participant and observer are blind to treatment. I dare say the participant (equipment after the network) was blind to any changes, so were the observants. I was the unblinded person so it's almost a double blind if not randomized study. One other thing I did not do was write a study report and manuscript, that we usually do too

 

[castleofargh]

As always, it comes down to what someone considers to be evidence.
If when coming back home I see a hippopotamus on the road(I’m in France), I will instantly believe that event to be factual. But when telling the story to my friends, at best they'll think I am messing around and they will go for the classic: ”pic or it didn't happen”.

It's the same situation here. Believing something doesn't make it a fact for everybody else. Even being right won't magically cast trust into the mind of others.

I cannot accept sighted experience as evidence of audible impact, because of the many possible biases involved in a sighted impression. Maybe some are correct, but do not expect me to take your word for it. And this section at large will do the same. ”pic or it didn't happen!”

At the same time, I agree with SAM that we must be careful about what we can conclude from @amir’s video and measurements. Getting no difference in a clean environment does not mean some can't happen under more troubled conditions, and any testimony must be BS.
Now based on that vid and the little things I read to try being not so totally ignorant on the subject(work in progress), if I happened to find myself in a situation where I am hearing differences from different switches, I would obviously make sure I am not fooling myself. And my next step would be to try hunting the actual source of the problem. I was never a fan of the band-aid approach, not with USB magic boxes, and not with this.

 

[marcelnl]

A couple of posts earlier I indeed mentioned (or attempted that) that trying to find the root cause for the phenomenon is my goal, I am not one to care for band aids unless there is no solution in sight.

 

[sander99]

Although it can never prove that there are no audible differences or imperfections, if enough tests are done you can have a high confidence that there most likely are no audible differences or imperfections. (And even more likely that even if there was an audible difference or imperfection it is so minimal that it doesn't matter.)

I can see the second sentence as probable, but the first is incorrect. Quantity does not lead to quality. Doing something wrong many, many times does not converge on the right answer. Failure to reject the null hypothesis does not prove the null hypothesis. Absence of evidence is not evidence of absence.

Yes I admit I was a bit sloppy on several levels. But I did say "it can never prove that there are no audible differences or imperfections"! I was sloppy in the next part because I just wanted to mention listening tests for completeness, and didn't go into more detail because I figured listening tests are not needed anyway regarding the issue at hand.
On a more overall level I was also sloppy, I got a bit carried away I guess, and ignored the fact that the measurements we have are of course not covering all possible situations and environments with all possible switches. And just like there can always be the odd DAC that isn't audibly transparent, maybe there also could be the odd switch that under certain conditions does something audible.

Having said that, it remains extremely unlikely that there are "various easily identifiable network switch sounds". One odd one sounding different/worse than all the others would be more easy to believe, but still unlikely. Anyway, if there are "various easily identifiable network switch sounds" then it should be easy to measure some differences that are at least close to the tresholds of audibilty. If someone wants to get at the bottom of this the first thing to do would be more measurements, in more different situations/environments. And only if the measurements results are inconclusive then one could start thinking about (controlled) listening tests.
And again: the manufacturer or seller or other person that claims the unlikely is the one who should prove it. So why don't they?

 

[marcelnl]

'first thing to do is more measurements in more different environments'.....eeeeeh what? Repeating something that did not work? as in 'go off and measure...and don't come back until you can prove a difference'?

'The definition of insanity is doing the same thing over and over again and expecting a different result.'

 

[bfreedma]

'first thing to do is more measurements in more different environments'.....eeeeeh what? Repeating something that did not work? as in 'go off and measure...and don't come back until you can prove a difference'?

'The definition of insanity is doing the same thing over and over again and expecting a different result.'

How would measuring in different environments constitute “doing the same thing over and over again”?

 

[marcelnl]

just consider; if the measurement is the constant factor that did not show a difference is it perhaps not going to show any difference in a(n unspecified) 'different environment"
Would putting down that yardstick on the ground in the rain forest have shown the earth to be a globe?

challenge everything, measurement is a means to an end, not an objective

 

[bfreedma]

just consider; if the measurement is the constant factor that did not show a difference is it perhaps not going to show any difference in a(n unspecified) 'different environment"
Would putting down that yardstick on the ground in the rain forest have shown the earth to be a globe?

challenge everything, measurement is a means to an end, not an objective

Trying hard to wade through that flowery prose. Your point is we shouldn’t test different environments using a common criteria because they might not show a difference? How do we begin to determine the root cause of the “issue” if we don’t examine multiple implementation scenarios?

if you have specific suggestions on what the testing should encompass to identify claimed differences, I’d be interested in hearing them and likely have access to the tools necessary to execute.

 

[sander99]

'first thing to do is more measurements in more different environments'.....eeeeeh what? Repeating something that did not work? as in 'go off and measure...and don't come back until you can prove a difference'?

'The definition of insanity is doing the same thing over and over again and expecting a different result.'

just consider; if the measurement is the constant factor that did not show a difference is it perhaps not going to show any difference in a(n unspecified) 'different environment"
Would putting down that yardstick on the ground in the rain forest have shown the earth to be a globe?

challenge everything, measurement is a means to an end, not an objective

Ah, so following your logic we are finished. Amir has measured a few switches in one setup, one situation, one environment. Those measurements show beyond a shadow of a doubt that in that setup/situation/environment there were no audible differences. I admitted that is not a universal proof because it doesn't exclude the possibility that in another setup/situation/environment there could be different results. You say it doesn't matter. Then the case should be closed for you: network switches can not cause audible differences.
(Unless you still don't understand that measurements in some cases are enough. Not always, because indeed not everything is equally suitable be judged based on measurements, and some measurements are not conclusive. But measurements of electrical audio signals contain no mystery. If such measurements indicate no difference even thousands times more precise than any human can hear then it is impossible that there is an audible difference.)

 

[castleofargh]

just consider; if the measurement is the constant factor that did not show a difference is it perhaps not going to show any difference in a(n unspecified) 'different environment"
Would putting down that yardstick on the ground in the rain forest have shown the earth to be a globe?

challenge everything, measurement is a means to an end, not an objective

I don't know many cases where 2 events can produce a different feeling but no difference can be measured.
At least not many where the subject didn't make up that difference.

Not to say that measuring a frequency response is the same as measuring everything. But if the argument is about sound, we can most definitely confirm variations at the DAC’s output.

 

[marcelnl]

Those measurements show beyond a shadow of a doubt that in that setup/situation/environment there were no audible differences.

that is good to hear, that there is a surrogate parameter (or set) for hearing!

 

[marcelnl]

Trying hard to wade through that flowery prose. Your point is we shouldn’t test different environments using a common criteria because they might not show a difference? How do we begin to determine the root cause of the “issue” if we don’t examine multiple implementation scenarios?

That comes across as a bit condescending, still, I's suggest to start with a known setup causing audible differences for that you'll need to listen first or it would be a fool's errand trying to find a root cause.

 

[Northern]

Archimago had some interesting points in his blog.

 

[bfreedma]

That comes across as a bit condescending, still, I's suggest to start with a known setup causing audible differences for that you'll need to listen first or it would be a fool's errand trying to find a root cause.

With all of the subjective reports of audible differences, you would have thought we would have identified the root cause by now.

The reality you conveniently ignore is that the "audible differences" heard by people attributed to Ethernet switch changes have never resulted in measurements of the Ethernet switch showing those audible differences. Not a single manufacturer has published viable data supporting their subjective performance claims.

If you find a system with audible differences and measurements of an Ethernet switch showing audible impact from the device, please post them. Until then, the claims will remain, for me, filed along side Bigfoot and Loch Ness Monster sightings.