Except as Taleb points out, if you are the casino, there is no gambling involved. You know exactly how much you will take in over the long haul. There is no speculation about it, its pure math. It's only the single player who takes a risk, not the house.
That's the brave new world we're living in.
I was perusing the financial statements from my employer, and their trades were profitable 95% of the time during that particular quarter.
How incredible is that? When technology has reached a point that such a thing is possible?
I also think that an argument can be made that part of the housing crisis was caused by flawed derivatives valuation. For instance a credit default swap contract is basically no different than an insurance policy on a bond. And from reading books like 'The Big Short', I get the impression that one of the things that set the dominoes falling in 2008 was that billions or even trillions of dollars in insurance (or CDS) were sold on bonds, for fractions of a penny on the dollar.
Basically, imagine if you sold your neighbor insurance on his mortgage, but you sold it to him for a dollar a month. If your neighbor stops paying his mortgage, you're going to be paying a very large payout for very little reward.
But all the derivative models underestimated the possibility of falling prices, and so the CDS contracts were very cheap.
Long story short - the math wasn't wrong, the model was wrong. And that's where good quants can improve things.
It touches on a lot of things that are very interesting. Big data, the falling cost of storage, the advent of petabyte databases, scientific computing on an epic scale.
Actually, if you read the book that I suggested, you will see (from a guy who created those models) that the models were right, they always predicted a downside and a huge loss should certain events occur. But those who made the decisions kept on going deeper and deeper into risky territory, ignoring the cautions of the models downside forecasts, until it all collapsed.
Note that in Canada, this did not happen, because - by law - they did not allow such high leveraging. In other words they did not allow the incompetent traders to gamble with other peoples money to anywhere near as much risk as the US allowed. It was not the models that were to blame, it was the decision makers, both in Congress and the board rooms.
Note that in Canada, this did not happen, because - by law - they did not allow such high leveraging. In other words they did not allow the incompetent traders to gamble with other peoples money to anywhere near as much risk as the US allowed. It was not the models that were to blame, it was the decision makers, both in Congress and the board rooms.
Collapse is just removal of wealth that wasn't really created in first place. Sacrifice of civilians and soldiers is secondary catastrophe with tangible loss. Bankers still own the casinos and trading houses, and just print up a new deck of cards, invent a few different rules, and start dealing a new game, which looks very similar to all previous games.
@Geddes
What do you see of studies that support the importance of a speaker having a constant response?
And how much is the design of a horn a question of smoothness (low distortion and diffraction) vs constant directivity in your opinion?
The JMLCs for instance are definetly not constant but they seem to be very smooth.
BEM Simulations
What do you see of studies that support the importance of a speaker having a constant response?
And how much is the design of a horn a question of smoothness (low distortion and diffraction) vs constant directivity in your opinion?
The JMLCs for instance are definetly not constant but they seem to be very smooth.
BEM Simulations
There is a lot of evidence that the power response does matter, but clearly not as much as the direct response. The difference would, of course, be room dependent. the more reverberant room would be more dependent on a smooth power response than a dead room. I strongly prefer a more reverberant space because it always sounds bigger - reverb time and size being highly correlated.
It's more than just smoothness that matters. You cannot toe-in a beaming speaker to expand on the sweet spot as you can with one that has better horizontal control. If you are the selfish type and no one else matters then I guess this doesn't mater so much.
All types of horns and waveguides can be made "smooth" independent of their specific directivity aspects. I prefer to have good directivity control AND smooth response to just having one or the other. I don't see a tradeoff being necessary between smooth response and directivity control as your question implies. There is in a diffraction type device, but not in one of my waveguides.
It's more than just smoothness that matters. You cannot toe-in a beaming speaker to expand on the sweet spot as you can with one that has better horizontal control. If you are the selfish type and no one else matters then I guess this doesn't mater so much.
All types of horns and waveguides can be made "smooth" independent of their specific directivity aspects. I prefer to have good directivity control AND smooth response to just having one or the other. I don't see a tradeoff being necessary between smooth response and directivity control as your question implies. There is in a diffraction type device, but not in one of my waveguides.
Going to the Financial Moon
For those that do not want to study the matter, here is a brief read:
Quants: The Rocket Scientists Of Wall Street - Forbes
Left Wall Street before the last bubble burst & have never looked back since.
All holdings & income now come from real estate & I am still buying.
The need for quants is just a symptom of a financial disease.
Swim with the sharks, you just become more food.
Regards,
WHG
For those that do not want to study the matter, here is a brief read:
Quants: The Rocket Scientists Of Wall Street - Forbes
Left Wall Street before the last bubble burst & have never looked back since.
All holdings & income now come from real estate & I am still buying.
The need for quants is just a symptom of a financial disease.
Swim with the sharks, you just become more food.
Regards,
WHG
See the following post.
http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-640.html#post3692205
Refer to the curves at the above posting. See if you can rank order the systems shown by their power response curves. (Hint, they are already ranked as to preference in a controlled listening test.)
While it is true that in-room response curves show a greater dependence on sound power when the room gets more reverberent, no one has shown that the sonic effect follows the measurement.
As to "a lot of evidence...", I'd like to see a study that shows that sound power is anything more than a secondary factor.
Now response smoothness and flatness near the on-axis position is important. Viewed statistically, listener position relative to the speaker should be considered as a probability cloud. You could draw a countour around the ideal listening axis to encompass, say 90% of the eventual listening positions. Achieving the best possible response family within the 90% probability contour would be a very worthwhile goal.
Shooting for a particular response average outside of it is not supported by research.
David
@David
Sean Olive seems to often mention the power respons as one of the most important factor of speakers. I assume he isn't taking that out of anything, but that it correlates with Harman's blindtests. Let's also remember that those tests are done in a room with a lot of treatment. I have heard however that they remove some of the sidewall treatment in their testings, in order to differentiate the off-axis response more clearly.
Sean Olive seems to often mention the power respons as one of the most important factor of speakers. I assume he isn't taking that out of anything, but that it correlates with Harman's blindtests. Let's also remember that those tests are done in a room with a lot of treatment. I have heard however that they remove some of the sidewall treatment in their testings, in order to differentiate the off-axis response more clearly.
Source: Audio Musings by Sean Olive: Part 3 - Relationship between Loudspeaker Measurements and Listener Preferences
Who said it was more than "secondary"? Did you not read what I said? I clearly said that it was a "secondary" aspect, but it is not irrelevant. That has never been shown. Even in the Lipshitz, Vanderkooy study they do not conclude that it is irrelevant.
We went through all this already. I don't see the point in mis-stating what I said to try and recreate the argument. Dave, that is beneath you.
What Dave is saying that the smoothness of the off-axis response, especially near the listening axis, is important. But any particular shape of the power response is not shown to be preferred. The Olive quote and results agree with this. Smoothness is important, but constant directivity does not correlate with preference.
Out of the 4 speakers Olive is showing, the top two both have rising directivities.
I'm not sure what is really beneath me.😀
We have been through it before but we come back to it when you make further statements inflating the importance of power response.
I am sure there are limits in a system's power response beyond which it will be downgraded in perception, even if the direct response is exemplary.
For that matter there are limits in distortion and phase response, impedance curve and other parameters that we justifiably ignore if they are in the "normal" range.
The question is, can we rank order the goodness of loudspeakers by looking at their power response curve? Is it a necessary and sufficient criteria for a high ranking in blind listening tests? Should we spend money in the design phase (and at the expense of something else) in optimizing it?
I believe the research says, no.
The original Toole papers made that very clear. Floyd states it explicitly, that rather than perhaps being a useful indicator of audible resonances, there was no rhyme or reason to the shape of the power response curves. The 20 curves that I attached to the other post illustrate that perfectly. There are some "bad" power response curves on the well ranked systems and some "good" power response curves on poorly ranked systems (and yes some good with good and bad with bad). It is a random indicator.
Olive found the same thing when he did the multidimensional analysis of performance factors, he could well predict listener preference ranking by looking at axial response flatness and axial response smoothness and power response smoothness (freedom from resonances) but there was no particular power response shape required. No correlation between power response shape and goodness. No ideal power response curve, therefore no ideal directivity index curve, therefore no requirement for constant directivity.
Lipshitz and Vanderkooy found that large holes in power response were not audible (not so with peaks).
I'm not saying that power response can always be ignored, but if it is generally falling in contour and has holes but no significant peaks, then that is all you need to worry about and you should spend all of your efforts on achieving best direct response over a likely listening window.
The fact that a more reverberent room gives a room curve tending more towards the power response of a system is a red herring for many people. You should know better. I'm involved in a SMPTE working group looking at the X curve and the traditional approach of measuring the steady state response of a speaker (a curve dominated by radiated power response) just can't be relied on. Its gotten to the point where they draw 6 different room curves based on room size to get around the simple fact that steady state room curves (and the power response that influences them) are a very poor indicator of perceived frequency balance, especially as the room gets larger.
David
Last edited:
David
Can you rank order the goodness of a loudspeaker by looking at just the direct response? No, you can't do that either. Your own data shows that. You have to do more than just get the axial response right or room EQ devices would work wonders.
When one is making a really great speaker they need to do it all. What I am saying is that control over the directivity is not irrelevant and doing it all means getting it right just as one gets the axial response right.
If you are looking for a weighting of the tradeoffs then look to Olive, he does weight the power response as significant - it is not zero.
The rest of your rant is beside the point as far as I am concerned.
Can you rank order the goodness of a loudspeaker by looking at just the direct response? No, you can't do that either. Your own data shows that. You have to do more than just get the axial response right or room EQ devices would work wonders.
When one is making a really great speaker they need to do it all. What I am saying is that control over the directivity is not irrelevant and doing it all means getting it right just as one gets the axial response right.
If you are looking for a weighting of the tradeoffs then look to Olive, he does weight the power response as significant - it is not zero.
The rest of your rant is beside the point as far as I am concerned.
The unit B that he refers to doesn't really illustrate that point. It has an axial response hole at 3kHz that persists off axis, so yes there is a power response hole but only in conjunction with poorer on axis response (and lumpy bass).
In fact the directivity index curve (blue curve below) is no better or worse than the d.i. of the better ranked L above it. Axial response alone would have predicted the rank order of the 4 speakers.
While we are looking at them, look at how good the directivity index curves of the worst ranked speaker M is. The red hemispheric curve is dead flat: constant directivity. why is it 4th ranked?
David
Attachments
Last edited:
Good point. There are many DSP room EQ devices that are saddled with "house curve" or "room curve" requirements because equalizing to flat sounds wrong. That means that the reverberent component (dominated by the system power response) adds a misleading element to the room curve. If we "heard" the room curve we could just equalize it to flat and we would have the best results every time. It doesn't appear work that way.
The same arguement that most audiophile designers make. I seem to recall that you scoff at those that try and push distortion to especially low levels or try and time allign or achieve flat phase response. By that statement alone you elevate power response to the equal of axial response.
My definition of a great speaker would be to excel in all parameters that matter. In many parameters there is a "good enough" threshold and going beyond that is always wasteful of product cost and inevitably subtracts from achieving the best in other parameters that matter more.
David
He weights the smoothness of the off-axis response, not the shape of the power response curve. It is important to distinguish the qualities of the off-axis response that are important (smoothness and uniformity) from those that are not important (shape).
You see that's where all this got off track. Because it is precisely the off-axis response that matters to me, the power response just got to be synonymous with off-axis response (they are highly correlated, but not exactly the same things).
My goal has always been to have a constant narrow polar response, not to have some particular power response curve. The fact that if you have this constant narrow polar response then you will also have a flat smooth power response, but the former is the goal not the later.
Just looking at the power response does not tell us much about the very early reflections that I personally worry so much about. I never even considered the power response until I started plotting it and then I saw that the two things, directivity control and power response, went pretty much hand in hand. Somewhere along the line the whole argument got to be all about the power response, which I really don't care all that much about. I care about early reflections and the negative effect that they have on image.
Because it looks to have relatively nasty and bright on-axis response compared to the others, AND if anything the more constant directivity should call for a more mellow voicing than the others?
Dave
I think it is ridiculous that you would state this to me as I have made this exact same claim over and over again for a decade or more. I don't even look at THD, it doesn't matter. I don't look at waterfalls, they don't tell us anything new. What I look at is what I post from the software that I custom developed specifically to look at "all the parameters that matter". You see polar maps around here all the time now. Who did these first? Guess. (I understand that Genelec was showing something similar about the same time that I first started showing them back in 2003 or so, but I had never seen theirs when I started doing mine. Now they are ubiquitous.)
Earl,
Could you list in order of importance what it is that you actually think are important parameters when you are designing a system? I really don't care if you don't list order or importance, but what you are looking for. I think I can list a few things that you don't think are important, but I am trying to understand what you do regard as your markers when doing analysis.
Could you list in order of importance what it is that you actually think are important parameters when you are designing a system? I really don't care if you don't list order or importance, but what you are looking for. I think I can list a few things that you don't think are important, but I am trying to understand what you do regard as your markers when doing analysis.