quentron.com

[MarkE]

So far it looks a lot like it's just aimless tinkering.  I think that will be pretty apparent when Pomodoro runs out of caring before profitis runs out of tweaks.

Pomodoro has announced that he is about done and his results do not support Profitis' claims.  Profitis has tacitly acknowledged that the experiments fail to support his claims.  He just recently stated that he will have to do a demonstration himself.

Aren't you directing me to take some kind of action or cease some kind of action based on some outcome that "hasn't been written yet"? How is that different than me taking action based on the probability of success and the probability of other peoples behavior and the probability of the behavior of Profitis?  I just looked at the outcomes, and did an informal decision tree.  The most rational response is...not really to care that much about if Pomodoro stops experimenting because of me.

I was asking you to not rock the boat before Pomodoro is done.  Pomodoro has put a good deal of time and money into his experiments.  He has been careful to address Profitis' objections along the way.  That adds new information.   I for one would like to see that information, and ask that since you don't care, ask that you do not disturb matters. 

[sarkeizen]

Profitis has tacitly acknowledged that the experiments fail to support his claims.

Perhaps but he has tacitly acknowledged that my analysis of his response was correct.  To him Pomodoro didn't provide any evidence against his point.  Pomodoro just didn't do it right either because of some technical detail or that Pomodoro is secretly you or myself.

I was asking you to not rock the boat before Pomodoro is done.  Pomodoro has put a good deal of time and money into his experiments.  He has been careful to address Profitis' objections along the way.

Put another way, if Pomodoro was trying to trisect an angle with only a compass and straightedge.  Should I still alter my behavior?

That adds new information.

Actually if the prior is ridiculously insanely and exceptionally low.  The amount of information an experiment - which only confirms the highly probable outcome - adds can not be very significant.  Negative results are important but not all negative results can be equally important.

since you don't care, ask that you do not disturb matters.

What don't I care about?  What I said was that the probability that this experiment would yield anything is so low that while I don't intend to bother pomodoro.   I believe that what is being risked there is worth almost exactly nothing.

[MarkE]

Perhaps but he has tacitly acknowledged that my analysis of his response was correct.  To him Pomodoro didn't provide any evidence against his point.

Profitis is unlikely to just concede.

Pomodoro just didn't do it right either because of some technical detail or that Pomodoro is secretly you or myself.

_{Don't give anything away!}

Put another way, if Pomodoro was trying to trisect an angle with only a compass and straightedge.  Should I still alter my behavior?Actually if the prior is ridiculously insanely and exceptionally low.  The amount of information an experiment - which only confirms the highly probable outcome - adds can not be very significant. 

The request which is becoming more and more moot as Pomodoro wraps up was for the simple favor of not discouraging it from reaching it's natural if highly predictable outcome.  Let the data tell the story, even if it is the same story as before.

Negative results are important but not all negative results can be equally important.What don't I care about?  What I said was that the probability that this experiment would yield anything is so low that while I don't intend to bother pomodoro.   I believe that what is being risked there is worth almost exactly nothing.

Do I think that there is any merit to Profitis claims?  No, I have not seen any evidence that there is, and his go against well-established understanding. As confident as I may be with conventional understandings, I do not wish to get trapped by hubris. While expectations of anything unusual were always very low, I encourage people who are willing to conduct careful experiments as Pomodoro has done.  Several things could have happened.  One is that Pomodoro could have found unusual results that were the result of an experiment defect.  Ideally, the underlying cause would be found adding to knowledge of possible experiment pit falls.  I consider such knowledge valuable.

[sarkeizen]

Let the data tell the story, even if it is the same story as before.

If the odds are exceptionally low of a meaningful unusual result.  Guess what is many, many times more likely?  A meaningless unusual result.  This is exactly why it is actually detrimental to science to run homeopathy trials.  The prior is so incredibly low that all unusual results are likely to be meaningless.

I do not wish to get trapped by hubris.

Have you considered giving up the irrational belief that any idea should be tested experimentally?  That just as much hubris as believing any other idea is unshakable.  Even with infinite resources highly unlikely experiments will produce many times more noise than signal.  So all you do is produce mountains of bad data.

I encourage people who are willing to conduct careful experiments as Pomodoro has done.

I encourage people to realize that any experiment without a probabilisticly bounded expectation either must have an exceptionally large effect side or is effectively useless.

One is that Pomodoro could have found unusual results that were the result of an experiment defect.  Ideally, the underlying cause would be found adding to knowledge of possible experiment pit falls. I consider such knowledge valuable.

That's your most likely case of getting information out of this kind of experiment?  Please.  P(UR) = probability of unusual result from ordinary circumstance - essentially this is inversely proportional to the strength of your methodology.  Since Karpen's Pile and other magic often have little in the way of explained mechanism.  This will often be higher than in other experiments.  P(FU) = probability of finding the reason for an unusual result.  Often co-morbid with P(UR).  P(URU) = Likelihood that unusual result is unique.  If it's not unique then it adds no information to science.  i.e. finding a possible ordinary explanation for an unusual result in a book.

How likely would you say that  P(UR) * P(FU) * P(URU) is?

I'd guess  "so close to zero that you might as well count it as zero".

[MarkE]

If the odds are exceptionally low of a meaningful unusual result.  Guess what is many, many times more likely?  A meaningless unusual result.  This is exactly why it is actually detrimental to science to run homeopathy trials.

Extraordinary observations require extraordinary validation.  It is a folly of hubris to unilaterally declare that low probability dictates that an experiment should not be conducted.  There is a big difference between reasonable allocation of research resources and dictating that an answer is absolutely known.

The prior is so incredibly low that all unusual results are likely to be meaningless.

It is important to distinguish the differences between:  likely, extremely likely, and absolutely true.  There are many foolish ideas that a number of people assign inexplicable probabilities to, including belief that they are true when solid evidence runs counter.  But none of us are omniscient.

Have you considered giving up the irrational belief that any idea should be tested experimentally?

Before I give up an idea I must first have it.

That just as much hubris as believing any other idea is unshakable.

Agreed.  I have not argued that any idea should be tested.  I argue the idea that if someone wishes to expend their resources testing an idea, even if the idea seems silly, that I see no good reason to discourage such an act.

Even with infinite resources highly unlikely experiments will produce many times more noise than signal.  So all you do is produce mountains of bad data.

Badly conducted experiments make noise.  Well conducted experiments produce reliable data.

I encourage people to realize that any experiment without a probabilisticly bounded expectation either must have an exceptionally large effect side or is effectively useless.That's your most likely case of getting information out of this kind of experiment?  Please.  P(UR) = probability of unusual result from ordinary circumstance - essentially this is inversely proportional to the strength of your methodology.  Since Karpen's Pile and other magic often have little in the way of explained mechanism.  This will often be higher than in other experiments.  P(FU) = probability of finding the reason for an unusual result.  Often co-morbid with P(UR).  P(URU) = Likelihood that unusual result is unique.  If it's not unique then it adds no information to science.  i.e. finding a possible ordinary explanation for an unusual result in a book.

I find value in determining the sorts of errors that occur in experiments so as to improve the reliability of experiments.  That means that an experiment that produces a false positive has educational value provided the reason for the false positive is tracked down.  Case in point was the FTL neutrino experiments at CERN.  An unusual result was reported.  It was the result of all things a loose optical connector.  The episode taught many people valuable lessons in conducting experiments.

How likely would you say that  P(UR) * P(FU) * P(URU) is?

I'd guess  "so close to zero that you might as well count it as zero".

A limit approaching zero is distinct from zero.