What objective criteria distinguish between valid science, fringe science and pseudoscience in physics? Plenty of research activity in physics have been vigorously opposed by their opponents as pseudoscience or fringe science, while other research are mainstream. It is possible some topic is pseudoscience if the experts claim it is so, but they could potentially be biased. For the nonexperts out there, short of appealing to authority, what objective criteria can be used to distinguish between valid science, fringe science and pseudoscience, i.e. topics which are a waste of time compared to topics worth pursuing.
 A: I would like to take a stab at this, because I think that this is where we are most privileged. This is perhaps the last moment in time when science is the exclusive domain of the specialist. I am optimistic that soon every person will either directly possess, or personally know some people who together possess, enough expertise to access all human knowledge. Public dissemination makes science into what it was always intended to be: an activity of masses, not elites.
In such an environment, bogus ideas don't stand a chance.
Scientific criteria
Pseudoscience and fringe science are pejorative political terms, which are never used by the proponents of the ideas. I don't like the labels--- I prefer simpler labels, like "right" and "wrong". One can ask, what distinguishes a correct idea from an incorrect idea?
For this case it's simple. The only criteria are


*

*Internal consistency

*Theoretical elegance

*Experiment


Of these, the third is by far the most important--- without it, we would get stuck in ruts all the time, we would run out of new ideas, and research quality would be determined by pure politics, and we know by bitter experience that in politics, Aristotle beats Aristarchus. Further experiment suggest new theories, and new ideas, more often than theory suggests new experiments.
But experiment is not the sole arbiter, because experiment requires theory to interpret, and experiments can be flawed--- consider early studies on ESP, or more recent OPERA results So you need to consider 1 and 2 when evaluating theories and experiments, to see if they make sense in the framework of knowledge already accumulated. But you can't go overboard--- if experiment tells you something, and careful analysis shows that it is correct, then that's that.
It is my opinion that these three principles by themselves suffice to distinguish right ideas from wrong ideas, without any need for political labels like pseudoscience, fringe science, and so on. This way you don't take any chances dismissing an idea politically. If the idea is wrong, it should be trivial to refute it by finding an internal contradiction, a theoretical weakness like a huge amount of unnatural parameter tuning, or a flat out contradiction with experiment.
In today's media climate, it takes less time and effort to fully refute a wrong idea than it takes to come up with it in the first place. So wrong ideas today have a negative multiplication factor, and it should be a short while before they are all extinct. This is a pity, in a certain sense, because it means that future generations will never know an Archimedes Plutonium, or an Alexander Abian, they were quite possibly the last and best of their kind.
Evaluating theories
When evaluating theories, it is important to not restrict yourself to "right" and "wrong", although this classification is still important. You also need to consider "fruitful" vs. "dead-endy", "interesting" vs. "boring", "original" vs. "derivative". These criteria are more human, and more subject to error than the rigorous scientific standard of right and wrong, but they are necessary, because even wrong ideas can often be tweaked into correct ones, and you need to know where to tweak.
Most of the ideas that are traditionally labelled pseudoscience are generally easy to dismiss even in a theoretical evaluation without direct experimental evidence, because they are clearly boring, clearly derivative, and clearly dead-ends. One doesn't have to classify them as pseudoscience to see this--- one can label them boring, derivative, and dead-endy without distinguishing them from their peer-reviewed cousins which are just as boring, just as derivative, and just as dead-endy. I don't see much difference between a numerological analysis of the standard model that appears in a peer reviewed journal and a numerologically motivated analysis of the standard model that somebody publishes on a personal web-page.
Vindication
Many ideas which were dismissed as pseudo-science turned out to be correct. So it is best to ignore the labels and consider the ideas on their merits. Here is an incontrovertible list:


*

*Meteorites: yes, rocks do fall from the sky, and this is the reason that the moon and other rocky bodies are pockmarked with craters.

*Non-Mendelian inheritence: It turns out the Mendel's genes are not the main story in heredity, that most of DNA is non-coding and that the non-coding DNA is transcribed to mystery RNA whose function is to regulate proteins. The major evolution of organisms is in non-coding regions, the coding regions mostly undergo selection neutral mutations which act as a clock for species divergence, and the exact mechanism of RNA function is not known (although I have my own ideas)

*Radiation hormesis: this idea might not be right, the jury is still out, but the statistical evidence does not dismiss the idea that small doses of radiation might not promote cancer, but might counterintuitively halt the production of cancers. The linear-no-threshhold model that cancer rate is linear in the radiation dose was held up as scientific consensus, although the experimental evidence is essentially nonexistent, and the model is now more theoretically dubious than ever.

*Continental drift: this idea was dismissed as pseudoscience for far too long, considering the overwhelming statistical evidence in the matching fossil records at corresponding locations along the African and South-American coastlines

*Jumping Genes: This was dismissed for a short while only, but won the Nobel prize.


There are further examples of theories which were dismissed as vague or ill-defined, perhaps not quite pseudoscience:


*

*S-matrix theory: It is incredible to me that this stuff was dismissed as rubbish, because this theory is more mathematically sophisticated than anything in physics before or since. It is fasionable again, now that the internet allows people to learn the details.

*String theory: This is an outgrowth of S-matrix theory, but again, it was dismissed for a long time, perhaps because it was too good to be true. Well it's too good, and its true.


In addition, I personally find the following vague spiritual ideas impossible to refute, and perhaps they might gain statistical evidence in their favor with time, although they are currently not part of science:


*

*Synchronicity: this is the idea that people's minds are correlated to a greater extent than one would expect from random chance. Some classical examples are flawed, because they require supernatural agency, but the basic idea is that if you have an idea at a certain point in time, somebody far away is very likely to have the same idea too, and likely at the same time. This can be a turn of phrase, an idea for a scientific theory, a fasion scheme, anything at all. This link between minds suggest that a superstructure is active in human experience, a collective mind of some sort, whose logical positive manifestation is this nonlocal correlation between the independent thoughts of individuals

*Perceptual Auras: Some people claim that they are able to see an "aura" surrounding other individuals, and their perception is so strong, that they try to make machinery detect the aura. This idea is clearly bogus as physics, there are no energy fields surrounding people, but the perceptual effect is real--- I have seen the auras myself, mostly in settings such as public performance which are designed to maximize the perception of an individual. It remains to be seen if these perceptual auras are inter-objective, whether two people will agree on the perceptual hue and shape of an individual's aura. James Joyce uses the color term "heliotrope" to refer to the color of a perceptual aura, a sort of golden glow which looks like nothing at all, but directs the mind's attention, and reveals internal activity. The halos of ancient and mideval art are also clear representations of an aura around the holy folks, and the "orgone energy" of pseudoscience is a similar thing. Again, it is clearly perceptual, not physical, but it is clearly perceptually real.

*Hypnosis and Sexual Suggestion: This is the popular idea that there are people with the ability to control others through the force of will. Mesmer was a proponent of such ideas, and surrounded it with the trappings of science, by claiming that the mechanism of control was magnetic fields. This is clearly incorrect--- magnetism has no effect on higher cognition. But the phenomenon of natural hypnosis can be seen in human interactions all the time, especially in relations of women to men. I have seen even at an early age, that women are able to suggest actions to men which they feel compelled to perform, even if these actions are silly. A female friend of mine once turned to me at a party and said "watch this", and then commanded the man standing next to her to stand on the table, which he promptly did. She then made him dance around for a while, and then asked him to come back down (I hope he's not reading this). This suggestion mechanism is clearly related to authoritarian power structures in society, and fetishized sexual submission/domination. But the entire phenomenon, including the simple existence of meditative trances, was dismissed as pseudoscience, it maybe still is.

*Musical synesthesia: this is the perception of colors in the source of music. I have, on rare occasions, seen colored streaks from a guitar during a session, and I have seen artists depict colors eminating from the sources of music. I find it fascinating that the brain places the color at the known source of the music, rather than as a general hue, or at random spots, like when you are hit in the eye. This is clearly high-level perception, but can it be replicated in different people? The notion of musical pitch is also interesting--- it has a perceptual height. Is this pitch/height relationship subjective, or is it the same between different individuals?


Why do wrong ideas persist?
I believe that any persistent wrong idea, especially in today's media environment, is only alive because it is serving a social purpose which is not apparent. Once a good substitute correct idea is formulated which can serve the same social purpose, the idea can die. Alternatively, the social purpose might be rendered obsolete, by an evolution of society.


*

*There is a persistent idiotic idea that global warming is a hoax. This idea is preposterous, especially since global warming was proposed already in the Silent Spring era, when it was called "the Greenhouse Effect". A quick back of the envelope calculation shows that human CO2 emissions match the rise in CO2 levels, and a quick back-of-the-envelope estimate of heat trapping dynamics also shows that it is plausible that it would be responsible for the warming we see today.


This, along with the published results of detailed models for the distribution of warming, ice-core data on the correlation between CO2 and temperature, and projected predictions that each successive year will be warmer than the last (predictions which are each exceedingly unlikely considering how hot current global climate is, and which are uniformly correct), make it certain that the world is warming due to the activity of humans. The only strike against this theory, as far as the media is concerned, is that it was predicted by left-wing activists in the early 1970s.
So there is a social force at work here. The obvious reason to deny warming is to prevent a carbon tax, or emissions cap, and this is supported by big money. So you have denials, but there is no sound science behind the denials. But global warming denial, like genocide denial or big-bang denial, is on the wane.


*

*The idea that the Earth is young seems to never die out. The social reason is obvious--- people want to believe the bible. The creation story in the bible is found here (I translated this, so I know it's accurate: http://en.wikisource.org/wiki/Bible_%28Wikisource%29/Genesis ). It basically depicts the Earth as a flat disk with a big water-covered sky-dome which leaks rain.


I think in this case it is important to recognize that the wrong ideas are persisting because of the important life-altering effects of religious experience, which has motivated people to allow reforms, even to the point where they will die for their beliefs.
The notion of religion, that it is possible to gain experience of consciousness different from that of the individual person, is to my mind entirely reasonable. But the authoritarian claim that the Bible is a good source of ethics is entirely bankrupt.


*

*Prayers/Chants/Magical spells: The idea that an incantation will work to produce a physical effect is manifestly absurd.


But people believe it, because these chants and prayers have a perceptual effect on the people who perform the ritual, and those that come in contact with them. This effect is real, but to what extent it is inter-objective remains to be seen. It is a powerful way to direct people's attention, see Allen Ginsberg's magical "Om": http://www.intrepidtrips.com/pranksters/ginsberg/ . These things transcends their emptiness as physics, because of their mental resonance.
Wrong ideas in science
In many fields of science there are flat out absurd ideas which gain traction, mostly because nobody thinks about them enough to realize that they are absurd. It is hard to name many of these, because I am just as blinded as anyone else by my place in time, but I will try:
These are a counterpoint to false ideas by non-scientists. These are false ideas promoted by scientists. I would call them pseudoscience, but unfortunately, they are mainstream positions today.


*

*Germ line mutations are random: This is a fairy tail which should not by taken seriously by anyone. The idea here that mutations are molecular readjustments caused by x-ray particles hitting DNA, or by mistakes in copying. That this is absurd was already pointed out by Wolfgang Pauli--- consider how long it would take to random mutate your way from a mouse to a human, and compare with the actual time taken by evolution. It has recently been established that bacteria can control their mutation rate, and that stress can induce mutations, not through molecular damage, but through cell signalling. The proper source of mutations is not known at present, although I have my own ideas.

*Modern Synthesis: This is the idea that Mendelian inheritence coupled with Darwin's natural selection can account for the evolution of higher organisms. The idea is that each gene equilibrates according to a separate population genetics model, and the independent selection pressures lead to organism drift with time. This idea is completely insensitive to the detailed computational structure of genetic and RNA networks, and does not allow for coherent system-level evolution. The mechanism is clearly more complicated, and more authorly, involving an actual computing system internal to germline cells, not a blind random number generator plus a blind selector. A parable: once, there was only one book, a cookbook, which described how to make macaroni and cheese. The book was transcribed by scribes, and they sometimes made errors, a letter dropped here, a duplicated passage there, and so on. After many years, behold, we have the Library of Congress! The argument is demolished by the absurd time scale required for the errors to accumulate to produce "War and Peace", if there even is such a timescale at all, considering that you would have to pass through a mountain of gibberish to get there. The existence of collective selection on a system level is not taken seriously in population genetics, likely because such effects are difficult to model.

*Group selection (other than kin selection) is forbidden: There is the following argument against group selection: if a mutation favors the group, but disfavors the individual, then this mutation will be selected against by natural selection, just by the mathematical process of elmination of individuals carrying the gene. This argument ignores the existence of sexual selection, the traits for which it is selecting are themselves selected for. Sexual selection traits can evolve in a group to select for traits which favor the group, but which are not beneficial to the individual. In my opinion, this is the purpose of sexual reproduction--- to allow group selection through sexual selection. But the consensus view is the bullet--- no group selection. It is belied by altruistic scouting behavior in prarie moles, which is beyond the extreme limit predicted by kin selection (although no-one seems to have suggested that this is a sexually selected trait). 

*Fever kills germs: This is the nutty idea that the 3 degree change in body temperature is designed to slow down bacterial or viral replication in the body. This is completely preposterous in the case of bacteria, since they are completely insensitive to such changes-- they live in the real world. It is a little more reasonable for viruses, but it is still absurd. The obvious reason is that the temperature change is used by the body to regulate internal temperature sensitive mechanism. This mechanism is not known at present, but I have my own ideas.


In physics, I compiled some of the persistent wrong ideas here: Common false beliefs in Physics
Cold Fusion
One of the most deplorable cases where the label "fringe science" and "pseudoscience" is thrown around a lot is in the field of cold fusion. Many dozens of groups, mostly in unpopular research centers, although a few in well financed respectable laboratories, have reported nuclear effects in deuterated Palladium. The political structure of physics dismisses these claims, but in my opinion, there is no sound theoretical argument against them.
I have not seen a convincing refutation of cold fusion, but I have seen papers with very solid evidence that nuclear effects are happening. The papers continue to trickle out, and cold fusion is now acceptable again to the American Chemical Society, although not yet to the American Physical Society, or the National Science Foundation.
It would be nice to see a debate of cold fusion on its merits, without political labels getting in the way. Only for some specific theoretical models, are there clear refutations.
A: Science is almost never revolutionary, it is mostly evolutionary: new, good science, usually takes care of explaining all the existing observations, and possibly providing 
1) explanation for previously unexplained observations 
2) new observations not previously thought about
Fringe science is usually quickly detected because it tries to debunk existing well established science without a proper account of the observations already consistent in the existing framework.
For instance, by axiom of choice, lets pick a random theory proving that special relativity is wrong. It almost sure does not care about explaining cosmic rays muon decay mean paths, time dilation in atomic clock or any of that boring stuff. It will cut right to trying to prove that the theory is wrong based on their own pseudo-logical arguments.
Also, another early symptom of fine crackpottery is try to cram a lot of concepts in as little text as possible, in an attempt to intimidate people out of debunking their assertions
A: There are no hard criteria... but, I think a very good general metric is the predictive power of the hypothesized model.  Real science makes quantitative, testable, predictions based on a rigorous mathematical model.  The more closely the proposed mathematics match observed data, the better the science.
Pseudoscience tends to make qualitative predictions based on ill-defined hypotheses.  Creationists, for example, often claim that the speed of light has been slowing down over time (thus purporting to explain why we can see galaxies billions of light years away).  What they don't do, however, is provide a concrete model of the nature of the change.  Do photons change speed in flight or do they keep the speed associated with the time of their generation?  They also don't provide a mathematical model with which to make predictions.  How does the decaying light speed theory improve on calculations based on general relativity?
A: Many of the other answers overrate experiments too highly. Let me emphasize thought experiments instead. Many of the most important breakthroughs in physics, by  Galileo, Newton, Einstein, quantum mechanics, etc. were due to thought experiments. Even the positivists who were so heavily leaning toward empiricism had to acknowledge this sneakily by subtly changing their position to logical positivism, to allow for the "rationalistic" element of thought experiments.
A: Metaphysics basically deals with statements which are undecidable given the current models of physics, i.e. statements which either it or its negation, but not both, can be added as an additional postulate and still give a consistent theory. But it’s entirely possible the next paradigm shift and the next version of physics models can firmly decide the truth or falsehood of some of these statements one way or another. This moves that topic from metaphysics to physics. This is why pursuing metaphysical questions is worthwhile because it can prepare the way for the next paradigm shift.
A: Time will tell. Wait a couple of decades, and nearly everyone will know the difference. Hindsight is 20-20.
A: A long time ago, I was taught that science to be accepted must satisfy a set of simple criteria such that:
Physics can only study phenomena which can be reproduced. A sudden flash of ligth that happens only once can not be studied. Observations must be matched by a mathematical model which can explain the observed facts and predict a few other. At last simplicity should be the thread to use to guide us in the darkness. 
However nice is this set of criteria, one can not use it to discriminate true science from folklore: QM for instance, does not satisfy the reproductibility criterium, because of the impossibility to both know position and momemtum. One can indeed imagine a first experiment where I measure the position, and a second experiment made on a different particle produced in the same conditions where I measure momemtum. Thinking a little bit harder it is probably possible to dismiss all the above criteria with a long established theory, it being QM, statistical physics or else...
I now think that we should extend the postulate that reality is not independant from the observer to: the reality is created by the observer. And that a physical theory to be accepted must be thought and accepted by a large enought number of minds.
This position may appear extreme but consider for instance the paradigm about the motion of the Earth around the sun. Before Galileo everybody was convinced the sun was revolving around the earth, from Galileo's death up to now, everybody is convinced Galileo has proved the Earth is revolving around the sun. What are the facts: it is a daily experiment to see from earth, the sun rising and setting, and if one waits a full year to see the sun revolving around our position. In the same time, Galileo who is the father of the galilean relativity has clearly said that there is no absolute referential. We can thus doubt the words he is supposed to have pronounced about the earth revolving around the sun (making us think the sun is an absolute referential so much better than the geocentric referential). 
There was a turning point in history where a large enought number of minds has deformed Galileo's theory and transform it into a new paradigm: the sun is an absolute referential which is much more conveniant to use. And from this time on, humanity is firmly beleiving the truth is that the Earth is revolving around the sun.
At last, what do you think about the experiments which try to identify the interaction between mind and matter ?
http://www.princeton.edu/~pear/
A: Soft answer:  

Science requires proof. Therefore any experiments ie research carried out to discover these facts is not fringe the word fringe should not be used in this context.

A: @Mike Dunlavey gave the answer in the comment above:

Dead giveaways: 1) it requires belief, 2) little or no math backing it
  up, 3) phony, cherry-picked "statistics", 4) argument by vote, or
  prestige, or by insult, or fear, or "conceivability"

It is enough looking to the history. IMHO with some modification, the real order is the next:


*

*Idea & belief - not really belong here. The author of the pseudo-theory already have it. 

*Courage - the scientist need catch others to believe too. Here are many sources of fear why he don't start catching. One of the main "brake" for the real scientist is the fear from that he will be marked as pseudo-scientists.

*Power - by media-brain-washing, or by vote, or prestige, or by insult, or fear, or "conceivability" (backed up with some little or none experimental stats or math).


If the scientist have enough power, can catch more others, and at some limit the theory became as real. Example: dark matter/energy. Army of scientist believe in its existence, but nobody know what is it. The theory has enough power.
Another example: Here are many pseudo(?) scientists who believe than here is faster speed as light. Of course, never get enough attention, because they are without any power. Now, here is a chance than CERN will have enough power to change this with it's neutrino experiment. Only when/if the experiments confirm the faster speed, the army of physics will start rewriting 100 years of "fundamental" laws. Maybe the next 20 years confirms that some (low power) pseudo-scientist two years ago was right. (But the Nobel Prize but will get someone else). 
So, the answer to the question: Here aren't objective criteria. Only two thing differentiate the real and pseudo-scientists: they's COURAGE and POWER - and those are not objective.
Another question is why real-scientists haven't enough "imagination" to start developing new theories before things happens? E.g. when some "low-power" pseudo-scientists come with some theory why other scientists usually immediately flag them as "pseudo"? The lack of math or simple enviousness? As @AdamRedwine told above - here are believers of non-constant speed of light - why "big" scientists don't start making solid math for the theory? The worst thing what can happens than they confirm the current theory. Simply, the scientists haven't enough imagination and courage to start thinking differently = crowd effect.
It is the same as anywhere else. One of my friend years ago telling me about multi-touch(!!) full-screen mobile phones. He simple hasn't enough courage and power to patent it and start real development. Apple (years later) has enough power...
Ps: if someone care, correct please my "broken english".
