PDA

View Full Version : Heisenberger's Uncertainty Principle



Cat
07-01-2004, 07:36 PM
Last year BJC accused me of looking down the 'wrong end of a telescope' when discussing Matt's rating predictions v's Glicko. At the time tensions were high and I didn't want to stir up any controversy before I bolted for Blighty. But I think the statement is worth re-examining now that you all seem to be kicking the embers of the old BB.

When it comes to measuring biological systems I think it is useful to apply Heisenberger's principle in that by measuring something you alter the nature of the thing you are measuring. OK call it Dave's Uncertainty Principle!

Measuring something like chess performance through a rating system has the potential to alter, or bias, the performance of those competitors, therefore altering the system you are measuring. There are at least 2 ways this could happen.

1. Manipulation of results - this clearly happens, as evidenced by Bill's suggestion that Starter should quaranteen his adults against the rampaging Box Hill Juniors.

2. The psychological reaction of players to their published ratings. Excepting players at the top end of the rating system who maintain consistent performance through their professionalism, for most players to think that their abilities can be measured by a single value is fatuous. The Glicko system recognises this to some extent in the RD component. Nevertheless, to the player it is 'the rating value' that is important.

Different players react differently to their published rating, but there is no doubt it affects how the player feels about his abilities. This effect might be more significant in individuals more vulnerable to suggestion, such as children. A child could lose significant self-esteem in regards their playing ability due to a poor rating result, even though the change in value may be more to do with RD or volatility than consistent downturn in performance.

The question is how great is the psychological effect? Perhaps the nearest comparison is the placebo effect. Trials have demonstrated where aspirin is effective for musculo-skeletal pain control in 60% of cases, placebo is effective in 40% of cases. Similarly where anti-depressants medication is effective in 70-80% of cases, placebo is effective in 50%.

But the really interesting thing about the placebo effect is that it may not be entirely psychological, that the effect may be partially exerted biologically. This was beautifully demonstrated in an experiment in the 1970's where mice were given sugar in association with an adjuvant (immune stimulant). Interestingly over a period of time, when the adjuvant was withdrawn and sugar administered alone there was no dimunition in the immune response.

There are 2 conclusions I would draw from this in relation to chess and rating systems;

1. That positive reinforcement may have biological effects on neuronal development as well as psychological effects.

2. The only way to properly test 2 different rating systems vis a vis their ability to predict outcomes is through a prospective double-blind trial.

Kevin Bonham
07-01-2004, 10:48 PM
I do agree that rating systems can alter the behaviour they are intended to measure to a slight degree. Actually one of the potential virtues of the Glicko system is that because the players cannot accurately calculate the impacts of results on their ratings, they could be less inclined to play for ratings points. However, I don't think this point has been studied.

One of my main reactions to your post is: so what? So uncertainties involving measurement affecting performance can lead to a degree of error - this is a big deal in quantum physics arguments about whether you can exactly determine the location of a particle, but probably just another bit of noise if you're talking about a model.

Let's carry your argument to a ridiculous extreme - since rating systems might create measurement-related impurities, let's imagine getting rid of them, so people play tournaments with no idea how those events will be rated. In these circumstances, their form will still be discussed and analysed, just not using a rating system. It may be that A is marginally better than B but B is hugely motivated to be considered the best in the world whereas A doesn't care. So B alters his behaviour to maximise his chances of getting results that will make people think he is "the best in the world" (whatever that means). A rating system does not have to be formal to have the issue you describe.

So the only true test of a rating system along your lines would be to have players playing games of which the results are known to no-one, for no money, possibly without even knowing who their opponent is. But then you are measuring something completely unlike what the rating system actually measures - namely performance in chess tournaments. So it seems to me that your argument is self-defeating because a true double-blind test that accurately simulates real tournament conditions is impossible - having your behaviour observed is a part of those conditions and ratings are an attempt to formalise this.

Now, about those mice, what openings do they play? :shock:

Bill Gletsos
07-01-2004, 11:32 PM
When it comes to measuring biological systems I think it is useful to apply Heisenberger's principle in that by measuring something you alter the nature of the thing you are measuring. OK call it Dave's Uncertainty Principle!
Seems a good enough name to me given the total uncertainty of anything you normally would say in regards to ratings. :rolleyes:


1. Manipulation of results - this clearly happens, as evidenced by Bill's suggestion that Starter should quaranteen his adults against the rampaging Box Hill Juniors.
I never suggested manipulation of results. That has the connotation and implication of cheating. I refute your commnets entirely and take exception to you and your wording. :x
I simply suggested that the composition or the types of tournaments that were rated could be taken into consideration by the organiser.

There are no results here being manipulated. No games thrown, no changing of results.
In fact sometimes clubs may decided they wont rate a tournament. e.g many clubs or staes decide they wont bother rating rapid tournaments. Some tournament organiseres refuse to pay the admin fee thus their tournament is not rated.

There is no manipulation here, simply choice.


2. The psychological reaction of players to their published ratings. Excepting players at the top end of the rating system who maintain consistent performance through their professionalism, for most players to think that their abilities can be measured by a single value is fatuous. The Glicko system recognises this to some extent in the RD component. Nevertheless, to the player it is 'the rating value' that is important.

Different players react differently to their published rating, but there is no doubt it affects how the player feels about his abilities. This effect might be more significant in individuals more vulnerable to suggestion, such as children. A child could lose significant self-esteem in regards their playing ability due to a poor rating result, even though the change in value may be more to do with RD or volatility than consistent downturn in performance.
Who cares. :-''

The rating system is attempting to meaure a players strength based on the results of tournament games. It could therefore be argued that all that is important then is the results of those games and not how the player "felt".


1. That positive reinforcement may have biological effects on neuronal development as well as psychological effects.
Again I say so what. =;


2. The only way to properly test 2 different rating systems vis a vis their ability to predict outcomes is through a prospective double-blind trial.
The only one being blind here I think is you. ](*,)

I now await a response from your cohort Matt who always seems to take your side in these matters. :rolleyes:

Kevin Bonham
07-01-2004, 11:50 PM
Bill has adapted to the new smileys so much better than the rest of us. :(

Rincewind
08-01-2004, 12:14 AM
Just like to point out that it's Heisenberg, not Heisenberger and the Uncertainty describes something about the innate nature of matter at the sub-atomic level. Some people think the "cause" of Heisenberg Uncertainty has to do with the measurement interfering with the system being measured but this is not so (although interference is a separate issue). Heisenberg Uncertainty is just about the what is knowable within levels of certainty. Properties of sub-atomic particles form these "conjugal pairs" where the more accurately you know one property the less accurately you can know the other. One such pair, the classical one, is position and momentum.

Most of this stuff needs to be accepted and stored away by people, which usually means it has to be personally resolved. IE it has to "make sense" to the individual. The idea of measurement interference makes sense and many people attribute Heisenberg Uncertainty to this alone. However, to do so is to miss the point entirely.

Like most QM principles. If you think it makes sense, you haven't understood it fully.
______________________________

Now on to the mice. Is that a case of placebo effect? It sounds more like a Pavlovian response to me (although physiology is not my strong suit). A response is link to an unrelated stimulus and when the cause of one of the responses is removed, the stimulus still produces the second response.

The classic experiment was feeding dogs and ringing a bell. This linked saliva excretion with the sounds of bells. This learned response was reinforced to the extent that even after food was removed, ringing the bell solicited the excretion of saliva in the dogs so trained.

______________________________

Now on to double-blind studies of chess performance.

I don't believe this is required as you are not looking at the effect of something on a population so you effectively don't have a control group. Without a control group the concept of a double-blind is meaningless.

When comparing predictive accuracy all results are passed through both system and a regression analysis is performed. If the regression is favourable (in terms of predictiveness) then the new system is deemed to have won that test.

Sure there is a chance that the performance of the old rating may be impacted by the publishing of the results, but this is not a major problem as the same result is processed by both systems and so no control group is being tainted by placebo, Pavlovian or measurement interference. Both groups contain all games and are therefore affected equally.

The other point is that while these feedback loops of published ratings impacting performance might seem impure, who's to say that they are not as valid as any other response? The purpose of the rating system is not to measure knowledge of chess, or even tactical ability, or anything other abstract quality. It is about predicting the future results of players. Therefore, I think, anything which affects results would and should affect their rating.

That's what I think anyway.

Kevin Bonham
08-01-2004, 02:02 AM
I stand corrected. :) Thanks Barry, I appreciate the correction and look forward to inflicting it on hundreds who use the mistaken version to make pretentious pop-philosophical points about the limits of human knowledge.

Rincewind
08-01-2004, 09:48 AM
I stand corrected. :) Thanks Barry, I appreciate the correction and look forward to inflicting it on hundreds who use the mistaken version to make pretentious pop-philosophical points about the limits of human knowledge.

There are philosophical implications however it was not Heisenberg's work which caused philosophers to question what was within the bounds of what was knowable. Some had already done this before QM was established as a field.

However, I'd take with a grain of salt anyone using QM principles to make philosophical points by analogy. Most of QM doesn't make "sense" and this is because we essentially gave up trying to understand (ie fit within a coherent whole view) and just tried to produce a workable model. So the link between interactions at the quantum level and those at the macro level where most of us live is "tenuous", or at least, not well understood.

If you want some pop-philosophy based on QM you should try to get a copy of "The Self-Aware Universe". The author has an Indian name but I can't bring it to mind now. But from the title, you can probably hazard a guess as to the content. It's a mix of the New Age, Indian Mysticism and Quantum Mechanics in roughly equal measure, mixed well. :rolleyes:

Kevin Bonham
08-01-2004, 04:54 PM
This should probably go to the off-topic board if it carries on much longer but the student mag I used to edit frequently received letters from a crackpot who was obsessed with harmonising some variant of New Age and quantum physics. Exploring the subject to see what on earth he was raving on about, it seems that there is quite a lot of this stuff out there. I was always pretty ordinary at physics, but this QM/New Age hybrid stuff all seemed very woolly-headed and faddish to me.

Rincewind
08-01-2004, 05:30 PM
This should probably go to the off-topic board if it carries on much longer but the student mag I used to edit frequently received letters from a crackpot who was obsessed with harmonising some variant of New Age and quantum physics. Exploring the subject to see what on earth he was raving on about, it seems that there is quite a lot of this stuff out there. I was always pretty ordinary at physics, but this QM/New Age hybrid stuff all seemed very woolly-headed and faddish to me.

Just one last post, as I;ve found the book. The author is Amit Goswami, Ph D and profesor of physics in the Institute of Theoretical Sciences at the University of Oregon.

The back cover blurb has this as the first paragraph:

In this stimulating and timely book, physicist Amit Goswami shatters the widely held worldview that matter is the primary "stuff" of creation and proposes instead that consciousness is the true foundation of all we know and perceive. Consciousness, through our self-reference, creates material reality.

From memory, I didn't get all the way through it. In fact a quick skim of chapter 19 just now pretty much confirms it.

Bob1
08-01-2004, 07:08 PM
Barry, Kevin,
And I thought I had too much idle time - I stand corrected.
remind me NEVER to quote (anything or anyone) !

good work guys - keep it up
Bob

Rincewind
08-01-2004, 07:52 PM
Bill has adapted to the new smileys so much better than the rest of us. :(

He has also adopted an Obi-Wan Kenobi avatar, I note. I'm more of a Qui-Gon Jinn fan myself. ;)

Bill Gletsos
08-01-2004, 08:37 PM
Bill has adapted to the new smileys so much better than the rest of us. :(

He has also adopted an Obi-Wan Kenobi avatar, I note. I'm more of a Qui-Gon Jinn fan myself. ;)
Yoda didn't seem appropriate due to being height challenged. :D

Rincewind
08-01-2004, 09:07 PM
Should I mention Mace Windu before Matt does?

Cat
09-01-2004, 08:50 PM
Yes, it though it was Heisenberg also, and then I stumbled across it written as Heisenberger in a Webster encyclopaedia my son was given at Christmas, and I assumed I was incorrect.

I think you are being a little harsh in some of your criticism, BJC. The only way to measure sub-atomic particles at this point in time is to predict their existence, bombard them and then determine if the resulting emissions fit with predictions. It is inevitable that the particle being studied will be altered in the process, although I agree it is not the complete answer to uncertainty. Of course, more recently it has been suggested the particles may be interacting in dimensions or even universe beyond our perceptibilities, but that we will probably never know.

The salivatory responses elicited in Pavlov's experiments were neuronally induced (there is no known neuronal influence on the immune system), and is essentially the foundation for Cognitive Behavioural Therapy, which is widely used as a therapeutic tool in counseling today. Whereas the palcebo effect is mediated via simple chemical reactions, such as a protein and receptor, it is generally held that operand conditioning is mediated via more complex processes. The point is that CBT is considered to be a therapy based on knowledge and understanding.

Your comparison of the placebo effect and the Pavlovian response however, begs the question is CBT nothing more than a placebo? In other words, is any benefit just simply the consequence of intensive human interaction, or do the techniques have scientific merit? This is a question often raised in medical circles, but studies to answer this question are difficult to conduct due to ethical considerations.

Now to KB's and BG's erudite question, So What?

Well in part I raised the point in response to a question I felt I had left unanswered. However, more importantly it was in response to debates on other threads about junior development in Australian chess. I would suggest we take a leaf out of the Instititute of Sports holisitic approach to development that has been so successful for Australian sport over recent years.

Obviously they have loads of money, which we certainly do not, but there are lessons to learn. I am all for tradition, its a wonderful repository for all sorts of sustaining beliefs, but some traditions can outlive their usefulness and there comes a time when assumptions should be questioned.

A child's brain is plastic, and neuronal development can be influenced hugely by environment. If that environment is rewarding and encouraging, then this may well have biological effects that can promote development. Although coaching is immensely important, maybe we should be also considering more holisitic approaches.

Finally, double blind trials are always difficult to conduct, but not impossible. Many assumptions have been widely held for years, if not centuries, eventually to be overturned by proper scientific enquiry. None of us can possibly know the true limit of these influences I have outlined unless we properly measure them. I agree that in this context this is not necessary as long as we keep our minds open to possibility.

Bill Gletsos
09-01-2004, 09:51 PM
As usual you miss the whole point of ratings. ](*,) ](*,)

They are measuring a players strength based on his RESULTS. #-o

You got that its RESULTS. #-o

Would you like fries with that. :rolleyes:

Not his hair colour, brain size or any other feature. :-({|=

Until you understand that your just wasting everyones time. =;

skip to my lou
09-01-2004, 09:55 PM
\:D/ dance

:-s wt?

:-k hmmm

=D> wd

:mrgreen: hahabs

[-X nono

=P~ eheheehhh

:-# uhoh

[-( nfw

Bill Gletsos
09-01-2004, 09:56 PM
I take it you liked my use of the emoticons.

skip to my lou
09-01-2004, 09:58 PM
No, I was just extremely bored.

Bill Gletsos
09-01-2004, 10:07 PM
How come Jeo is editing your posts.

I didnt think anyone other than paulB and kevin were modererating/admining here.

skip to my lou
09-01-2004, 10:14 PM
hahahah fool \:D/ :-s

skip to my lou
09-01-2004, 10:17 PM
Oh look, now you're editing my posts! CRAX0R! :-s

Bill Gletsos
09-01-2004, 10:23 PM
Think you are 31337 dont ya.

skip to my lou
09-01-2004, 10:26 PM
"Think" :D :x :twisted: :evil: :P :? :shock: :D :) 8) :? :) :( :rolleyes: :oops: :twisted: :?: :idea:

Bill Gletsos
09-01-2004, 10:27 PM
:-k =;

Bill Gletsos
09-01-2004, 10:33 PM
Yes all you are doing is changing your .sig just like I have now done. :D

skip to my lou
09-01-2004, 10:38 PM
.......

Bill Gletsos
09-01-2004, 10:39 PM
And your last trick is just a script running aimed directly at me so that I get a message box appearing.

skip to my lou
09-01-2004, 10:41 PM
AHAHAHAHA ok. :D :D :D :D :D

Rincewind
09-01-2004, 10:51 PM
David,

Sorry if my criticism came over harsh. I certainly didn't mean to offend. As I said measurement interference is an issue (it is for any discipline) perhaps even more so for quantum physics. However, Heisenberg's Uncertainty Principle (HUP) isn't about that. What it says is even if we had a totally passive way to determine the position of a particle to a very low tolerence, the act of taking the measurement will necessarily mean we cannot know it's momentum with as much accuracy.

It's not that we change the nature of the particle when we measure it. It's just we cannot ever know both quantities simultaneously to an combined certainty greater than that perscribed by HUP.That's just the weird way things work in the quantum world.

Anyway, you seemed to be using this as an argument regarding feedback loops into the rating system whereas HUP has absolutely no application in this regard which is why I spoke frankly to ensure there was no misunderstanding. If I offended, I apologise and assure you that was not my intent.

Regarding Placebo vs Pavlov, I bow to your obviously greater wisdom in the area of physiology. I think I flagged my comments as those of a rank amateur. Of course no known neural link to immunity is not the same thing as no neural link but I'm sure you are on the good oil there.

I always regarded the Placebo effect as pretty much requiring conscious knowledge to operate. You have a bunch of people with headaches you give half an aspirin and the other half a sugar tablet. However, some of the people who have sugar report feeling well. The mind over matter response. The belief they are receiving treatment leads to a greater feeling of well being.

Not sure how that relates to chess ratings (in terms of comparing two rating systems for most effective predictor of future performance). Perhaps you need to explain the method you propose in detail and specifically what the advantage would be.

Cat
10-01-2004, 08:45 AM
Barry, you never offend me, you're quite a decent fellow. As to your suggestion that I flesh out my ideas, I'll give it some thought, although I am also keen to avoid offence. Cheers

Cat
10-01-2004, 11:10 AM
Barry, I also made a grammatical error in the post which might be the cause of confusion - where I said 'the placebo effect is mediated' should have read 'the placebo effect described is mediated' - I was referring to the experiment rather than the placebo effect pre se.

Cat
10-01-2004, 11:10 AM
Barry, I also made a grammatical error in the post which might be the cause of confusion - where I said 'the placebo effect is mediated' should have read 'the placebo effect described is mediated' - I was referring to the experiment rather than the placebo effect per se.