Why is there so much crackpot physics?

[BurntSynapse]

In a series of reports from 1994 through 2009, the Standish Group estimated that one quarter of all software projects are cancelled before delivery, or were delivered but never used. Almost half of the software that was good enough to use was late, over budget, or failed to meet its requirements. Barely one quarter of the projects could be counted as fully successful.

I would have expected an expert on IT Project Management to know that.

...and one would not expect ideologues opposed to the notion of PM efficacy to take heed of the findings in these many studies: application of PM significantly improves success.

You seem possessed by the belief I'm the source of most of my positions when the opposite is true.

 

[BurntSynapse]

You think the vector forms of Maxwell's Equations suffer from gimbal lock? I might have missed this if it came up before.

I was asked for an advantage of Q's, and I provided it courtesy of Wikipedia.

Quaternion experts have credibly maintained a case that using Q's offers advantages vectors lack. That is the basis of my guess.

Better guesses are welcome.

 

[BurntSynapse]

Of course now I'm curious. I can't even think of a basis for comparison between "information systems project management" and "Einstein's best work". To my ear it sounds like trying to compare NATO to a yogurt smoothie. It's like saying the infield fly rule is better than James Joyce's Ulysses.

To the disinterested, I'm sure it does sound like that.

 

[Mashuna]

I was asked for an advantage of Q's, and I provided it courtesy of Wikipedia.

Quaternion experts have credibly maintained a case that using Q's offers advantages vectors lack. That is the basis of my guess.

Better guesses are welcome.

Your guess is based on searching Wikipedia for an example?

 

[ben m]

Quaternion experts have credibly maintained a case that using Q's offers advantages vectors lack. That is the basis of my guess.

You are not merely guessing that "quaternions have advantages". You're guessing that "quaternions have advantages which provide an example of the sort of thing my project-management ideas would consider". Unfortunately this latter claim carries a lot of baggage that you're doing very little to address. Unless quaternions have an advantage over what physicists are actually doing now, and physicists have failed to realize this for the past 150 years despite its being Wikipedia-level common knowledge, and there's a chance that physicists have been laboring under the problems with vectors without doing anything about it ... well, without conditions like that, the fact you've cited (quaternions have computational advantages over Euler rotations) says nothing whatsover in favor of project-management-of-physics-discoveries.

 

[ben m]

To the disinterested, I'm sure it does sound like that.

a) Do you mean disinterested or uninterested? </pedant> </pedant who honestly can't tell what you mean>

b) You really don't like explaining things, do you?

 

[W.D.Clinger]

You seem possessed by the belief I'm the source of most of my positions when the opposite is true.

You're quite wrong about that. In fact, the opposite is closer to being true.

From what you've said in your posts, it looks to me as though many of your positions are derived from reading crackpot web sites uncritically, followed by rejecting expert knowledge.

For example:

I was asked for an advantage of Q's, and I provided it courtesy of Wikipedia.

Quaternion experts have credibly maintained a case that using Q's offers advantages vectors lack. That is the basis of my guess.

Better guesses are welcome.

Quaternions do have some computational advantages over Euler angles, and their avoidance of gimbal lock is one of those advantages. That's a fact, not a guess.

You were guessing, however, when you thought this might be relevant to physics. You can't blame Wikipedia for that guess, and you can't blame anyone who has expert knowledge of modern mathematics and physics. Crackpot web sites are the only possible sources I can find for your stubborn belief that quaternion formulations of Maxwell's equations are less risky than vector or tensor formulations.

 

[godless dave]

Quaternions do have some computational advantages over Euler angles, and their avoidance of gimbal lock is one of those advantages. That's a fact, not a guess.

You were guessing, however, when you thought this might be relevant to physics. You can't blame Wikipedia for that guess, and you can't blame anyone who has expert knowledge of modern mathematics and physics. Crackpot web sites are the only possible sources I can find for your stubborn belief that quaternion formulations of Maxwell's equations are less risky than vector or tensor formulations.

Nor can you blame the field of project management for that guess.

 

[lpetrich]

The gimbal-lock problem is only a problem for specifying rotation matrices with Euler angles. It's a coordinate-singularity problem. For analytic work, it's OK, because one can then do exact cancellations and the like. But for numerical work, it's a serious problem, because the coordinate singularities magnify errors for quantities near them.

However, that problem does not affect most other uses of vectors.

In fact, quaternions are typically implemented as 4-vectors with special manipulation rules.

BurntSynapse, let's see how much you know about quaternions.

1. Can you express them with Pauli matrices? If so, then you can use that expression for the next questions.
2. Is quaternion multiplication associative?
3. Can you find a simple non-quaternion quantity from a quaternion that gets multiplied along with its parent quaternions? Ignore quaternion addition here.
4. Can you find a simple expression for a quaternion's reciprocal? It would be another quaternion, of course.

Symbolically,

1. quaternion units 1, i, j, k -- related to Pauli matrices identity, sigma1, sigma2, sigma3?
2. q1*(q2*q3) = (q1*q2)*q3?
3. non-quaternion S(q1) such that S(q1*q2) = S(q1)*S(q2)?
4. quaternion R(q) such that q*R(q) = R(q)*q = 1?

where the q's are arbitrary quaternions.

 

[ben m]

Let me reply to an older post. I asked what sort of problem quaternions were imagined to solve. He didn't quite answer that, but seemed to indicate that problems were not inevitable:

Not necessarily. It would depend on what a modeling effort compliant with modern good practices yield relative to Heaviside. Perhaps as applied, we would find no discrepancy, perhaps not.

So, BurntS, let's apply this to quaternions. Now that you've gotten specific, you have several domain experts telling you that no discrepancy is possible---that the quaternion and vector versions of Maxwell's Equations are literally the same thing in different notation. Asking for a discrepancy between them is like asking for a discrepancy between binary and hexadecimal arithmetic. Are you willing to accept this answer or not? If not, why not?

 

[Dancing David]

Because uncited experts say so.

 

[BurntSynapse]

Hello Ben M and all,

Sorry for the delay, I've been travelling, went to the Starship Congress in Dallas, and just finished a promotional video for the 100 Year Starship Public Symposium in Houston next week. I stopped back here to review the objections to using information systems PM on physics research.

> I asked what sort of problem quaternions were imagined to solve. He didn't quite answer...
Q's were imagined to offer a consistent algebra where the law of commutativity did not hold, something that I'm told was considered an impossibility at the time.

>you have several domain experts telling you that no discrepancy is possible---that the quaternion and vector versions of Maxwell's Equations are literally the same thing in different notation. Asking for a discrepancy between them is like asking for a discrepancy between binary and hexadecimal arithmetic. Are you willing to accept this answer or not? If not, why not?

Actually this binary vs. hex example strikes me as a very good analogy of an important point. For your direct answer: I accept that the arithmetic results will be the same, just in different notation.

On the other hand: Do you also recognize that there are contextual issues that can make a huge difference in what arithmetic problems we undertake? For example, if we had to do binary arithmetic by hand, it seems unlikely we would do many, and they wouldn't be very complex, quick, or reliable. The opposite would be true if computers had to work in pure hex, with no conversions. If we're looking for discrepancies only in results, clearly we wouldn't find any.

Management science holds that the way we do things matters for what we do and how successful we are at delivering on the desired results. There are many studies proving application of good management positively impacts performance on projects, regardless of application area. Do you accept this includes projects within the physics research area? ...and right back at you: If not, can you explain why not?

 

[ben m]

.
On the other hand: Do you also recognize that there are contextual issues that can make a huge difference in what arithmetic problems we undertake? For example, if we had to do binary arithmetic by hand, it seems unlikely we would do many, and they wouldn't be very complex, quick, or reliable. The opposite would be true if computers had to work in pure hex, with no conversions. If we're looking for discrepancies only in results, clearly we wouldn't find any.

Notice the "with no conversions" you have to specify. Scientists and mathematicians are actually really good at switching conventions when needed. Notice that we *don't* try to build decimal-logic computers, or force people to do computer-related arithmetic in binary, or to represent the web color space in octal. Notice, also, that we managed to accomplish this (AFAIK) without advice from project managers---looking for the most-convenient radix is part of the normal toolkit of computer science, and does not require some special out-of-the-box thinking. Finally, note that there are some domains where the choice-of-radix is utterly meaningless---in symbolic math, solving "ax^2 + bx+c=0" is a math problem that makes no reference whatsoever to the radix.

Same thing in physics. Physicists are *very good at* choosing appropriate conventions and representations. We do so frequently, without prompting from a project-manager. The vast majority of fundamental work in particle theory is done symbolically, using Lie algebras, in which the difference between "R(3) as represented by quaternions" and "R(3) as represented by matrices" is utterly meaningless, just like the difference between "ax^2 + bx + c = 0 in decimal" and "ax^2 + bx + c = 0 in hex".

Finally, while you're talking about cases where the representation biases you towards using (or avoid) certain types of calculations: sure. Welcome to why physicists mostly abandoned quaternions. Quaternions are a wacky and confusing representation of R(3) that is difficult to teach, difficult to learn, do algebra with, difficult to represent on a computer, and extremely difficult to extend to anything other than R(3). Vectors and matrices are simple, easy to learn/teach, and trivial to extend. Vector/matrix representations make it *easy* to think outside the box.

There are many studies proving application of good management positively impacts performance on projects, regardless of application area. Do you accept this includes projects within the physics research area? ...and right back at you: If not, can you explain why not?

a) "Discovering FTL travel" is not a project. Nobody has any idea whether it is possible even in principle, and most people think it's not. There is no known action item, no known milestone, no known method of evaluating progress. You have been unable to clarify *what actions* the physicists under your management might undertake---other than these crackpotty suggestions like "try quaternions"

b) Project management can succeed if it gets people to do something they weren't already doing---reorganizing certain tasks, adding redundancy to risky paths, forcing documentation to be shared more effectively. Project management does not succeed by telling the FEA-analysts that FEA-analysis is important. You have not shown any *awareness of* what physicists are already doing, from which to develop this distinction.

c) Physics is *currently* managed. It is not an anthill. We all get our money from large cash-strapped government funding agencies, which use prioritization panels and peer-reviews and whatnot to decide whose research should continue or accelerate and whose should be discontinued. You have made no attempt to contrast "BurntSynapse-management" with "present management", only with "no management whatsoever".

d) You claim that "studies show GOOD MANAGEMENT" improves project outcomes. However, you are not attempting to apply any form of management that's ever been studied in any way. You're attempting to invent a "predictive" version of a principle (the action/object distinction) which is little-known, little-used, and (in my opinion) sketchy-sounding even in its historical/retrospective version. Even if good management can improve the progress of physics theory, what makes you claim that "action/object" talk makes for good management?

 

[Perpetual Student]

I am smiling at the image of BurntSynapse sitting at the side of Einstein, managing a gedankenexperiment.

 

[TjW]

I am smiling at the image of BurntSynapse sitting at the side of Einstein, managing a gedankenexperiment.

That would be sort of a metagedankenexperiment, wouldn't it?

 

[lpetrich]

I think that Alexander Unzicker is worth a mention here. He is the author of the book "Bankrupting Physics: How Today's Top Scientists are Gambling Away Their Credibility", with Canadian science journalist Sheilla Jones.

It is a translation of his 2010 book Vom Urknall zum Durchknall – die absurde Jagd nach der Weltformel (From the Big Bang to madness – the absurd hunt for the Theory of Everything). He has also written Auf dem Holzweg durchs Universum – warum sich die Physik verlaufen hat (On the wrong track through the Universe – why physics has gotten lost) (2012).

From Amazon's page on that book: "He publicly criticizes theories that are not testable like string theory or cosmic inflation, but also the so-called standard models of particle physics and cosmology, which, according to Unzicker, are too complicated to be credible."


He has written viXra.org e-Print archive, viXra:1212.0100, The Discovery of What? Ten Questions About the Higgs to the Particle Physics Community, but most of it is almost hopelessly amateurish.


In Bankrupting Physics | Not Even Wrong, Peter Woit tells us that he spent "a depressing and tedious few hours" reading that book.

After a while though, it became clear that Unzicker is just a garden-variety crank, of a really tedious sort. ...

Unzicker’s obsessive idea, shared with innumerable other cranks, is that any scientific theory beyond one intuitively clear to them must be nonsense. Similarly, any experimental result beyond one where they can easily understand and analyze the data themselves is also nonsense.

But unlike may physics crackpots, AU does not seem to have any theory of his own. What would he want?


He complains that starting with isospin, particle physicists have gone on a quest for unintuitive symmetries, but without those symmetries, the Standard Model would not be anywhere near where it is today. Without quarks and QCD, the hadrons would be a big unexplained zoo, and without electroweak unification, we'd have a theory of weak interactions that breaks down at about 1 TeV of interaction energy.

 

[lpetrich]

Finally, while you're talking about cases where the representation biases you towards using (or avoid) certain types of calculations: sure. Welcome to why physicists mostly abandoned quaternions. Quaternions are a wacky and confusing representation of R(3) that is difficult to teach, difficult to learn, do algebra with, difficult to represent on a computer, and extremely difficult to extend to anything other than R(3). Vectors and matrices are simple, easy to learn/teach, and trivial to extend. Vector/matrix representations make it *easy* to think outside the box.

R(3)? Do you mean SO(3)?

Quaternions are not that difficult to handle on a computer. All one needs to do is represent them as 4-vectors and to write appropriate multiplication and inverse functions for them. It's something like how one handles complex numbers, as 2-vectors with appropriate multiplication and inverse laws.

Quaternions are widely used for specifying rotations in computer graphics, so this is further evidence that it's not difficult to computerize them.

The unit quaternions can be mapped onto Pauli matrices, more evidence that quaternions have not exactly gone away. In fact, unit quaternions are equivalent to elements of the group SU(2).

But ben m is right that they are difficult to generalize. Though quaternions are not very difficult, they are very specialized.

 

[Perpetual Student]

R(3)? Do you mean SO(3)?

Quaternions are not that difficult to handle on a computer. All one needs to do is represent them as 4-vectors and to write appropriate multiplication and inverse functions for them. It's something like how one handles complex numbers, as 2-vectors with appropriate multiplication and inverse laws.

Quaternions are widely used for specifying rotations in computer graphics, so this is further evidence that it's not difficult to computerize them.

The unit quaternions can be mapped onto Pauli matrices, more evidence that quaternions have not exactly gone away. In fact, unit quaternions are equivalent to elements of the group SU(2).

But ben m is right that they are difficult to generalize. Though quaternions are not very difficult, they are very specialized.

Mathematicians have a good handle on quaternions, where they fit within the context of mathematical theory and their application in physics.
They do not hold some untapped power that the mathematically naïve BurntSynapse has been led to believe by certain crackpots. Perhaps this LINK might be helpful to BurntSynapse
Note the conclusion:

9. Conclusion
The paper has related major covariance groups of
physics to the quaternion group and has given
several physical applications, most of them at
undergraduate level. It is hoped that the treatment
of the representations of the quaternion group will
have shown that quaternion group theory is quite
accessible to undergraduate students and might
further their comprehension of physics by exhibiting
the deep unity of physical phenomena.

 

[CapelDodger]

In a series of reports from 1994 through 2009, the Standish Group estimated that one quarter of all software projects are cancelled before delivery, or were delivered but never used. Almost half of the software that was good enough to use was late, over budget, or failed to meet its requirements. Barely one quarter of the projects could be counted as fully successful.

I would have expected an expert on IT Project Management to know that.

Well jeez, you just had to be there to know that. Never get paid by results - that's the road to ruin. Get paid by fulfilling contractual requirements. Carefully specified.

Fractal dimensions are a real topic in mathematics. (Indeed, my first journal article was co-authored by one of Mandelbrot's PhD students co-authors.) Unfortunately, fractal dimensions have become a strange attractor for the woo-inclined.

Now it's happened, you're not at all surprised, are you?

 

[CapelDodger]

Same thing in physics. Physicists are *very good at* choosing appropriate conventions and representations. We do so frequently, without prompting from a project-manager. The vast majority of fundamental work in particle theory is done symbolically, using Lie algebras, in which the difference between "R(3) as represented by quaternions" and "R(3) as represented by matrices" is utterly meaningless, just like the difference between "ax^2 + bx + c = 0 in decimal" and "ax^2 + bx + c = 0 in hex".

Excellently put.